POC: batch-granular Parquet scans with the push decoder and scan_plan read-ahead by adriangb · Pull Request #24086 · apache/datafusion

adriangb · 2026-08-04T12:11:19Z

Which issue does this PR close?

Part of [Parquet] Pre-fetch the next row group when reading parquet files #18470 (Pre-fetch the next row group when reading parquet files)
Alternative / companion experiment to Prefetch Parquet row groups with a bounded I/O budget #23492 (Prefetch Parquet row groups with a bounded I/O budget)
Depends on two arrow-rs changes, in this order (as agreed in Incremental Parquet Decoding: avoid the need to buffer pages for all used columns in a RowGroup in Async parquet reader arrow-rs#6946 (comment)):
1. Batch-granular decoding in ParquetPushDecoder, with the configured row selection policy: feat(parquet): predicate cache, page release and selection policy in the batch-granular push decoder arrow-rs#11240 (stacked on feat(parquet): let ParquetPushDecoder decode a batch as soon as its pages are pushed arrow-rs#11238, which starts from the test in test(parquet): show push decoder needs a whole row group before decoding a batch arrow-rs#11218)
2. ParquetPushDecoder::scan_plan (feat(parquet): add ParquetPushDecoder::scan_plan for read-ahead planning arrow-rs#10555)

Rationale for this change

Today a Parquet scan buffers a whole row group before it decodes anything: the push decoder's NeedsData does not resolve until every projected byte of the row group has arrived. Under object-store latency this has two effects:

I/O and decode alternate. The CPU waits for a row group, decodes it, then waits again. [Parquet] Pre-fetch the next row group when reading parquet files #18470 asks for these to overlap.
Buffering is bounded by row-group size. Resident bytes and time-to-first-batch depend on how the file was written, not on the query or the operator.

This PR schedules I/O at batch granularity. It uses the normal push-decoder path and adds only a read-ahead layer on top.

What changes are included in this PR?

One new Parquet read option:

Option	Default	Meaning
`datafusion.execution.parquet.read_ahead_bytes`	`None`	If set, decode a batch at a time and read ahead up to this many bytes per partition stream, bounded by the memory pool. If unset, behavior is the same as `main`.

The intended use: leave it unset for local disks (no speculative reads), set it for object storage.

The decoder decides what each batch needs. The streaming path builds the same ParquetPushDecoder as the default path, with FetchGranularity::Batch, and drives it with try_decode. NeedsData asks only for the pages of the next batch, and the decoder returns the batch as soon as they are pushed. The decoder releases pages when their rows are emitted.
ReadAhead fetches ahead from the decoder's plan. It reads scan_plan() in decode order and fetches in the background while decoder-held bytes plus in-flight bytes stay within the window. NeedsData is always fetched. A blocking fetch is filled with read-ahead up to the window, except the first fetch of a file whose plan is larger than the window, so the first batch comes early.
Filtered scans use the same path, and read-ahead is speculative. With pushdown_filters, the decoder evaluates predicates one window of batch_size rows at a time. Read-ahead also fetches conditional ranges (bytes that a predicate can make unnecessary), in plan order. Without this, each window waits for one round trip per predicate plus one: with simulated latency this made TPC-H 0.50x and TPC-DS 0.14x. With it, TPC-H is 3.79x and TPC-DS 1.74x, with no slower queries. The worst case reads about the bytes that main reads without pushdown. This matches what DuckDB, ClickHouse (remote) and Arrow C++ pre_buffer do by default.
Memory pool accounting. Each stream registers a ParquetReadAhead[partition] reservation that equals the bytes the decoder holds plus the bytes in flight. Bytes that NeedsData requests use grow: they can exceed the pool limit, so the scan never blocks or fails on them. Read-ahead uses try_grow: it fetches only what the pool can give, and tries again at the next decode step. With the default unbounded pool, behavior does not change. FileSource gets create_morselizer_with_context (default: calls create_morselizer) so that the Parquet source can get the pool from the TaskContext.
Row-group pruning and the per-row-group filter toggle work as on main. Read-ahead skips row groups that the dynamic pruner has already pruned. Fetched bytes of row groups that are pruned later are dropped.
Files without a page index get one plan entry per column chunk, so the same code gives overlapped row-group read-ahead.
The page index is loaded only when a row group to read is larger than the window.

Changes in this revision

Before	Now	Why
Decode with a separate sync `ParquetRecordBatchReader` over a custom `ChunkReader` (`SharedBuffers`)	Decode with the push decoder, `FetchGranularity::Batch`	The push decoder can now return a batch when its pages are present. DataFusion has no decoding path of its own.
Filtered scans fall back to the default path	Filtered scans use the streaming path	The decoder evaluates predicates per window.
Runtime dynamic row-group pruning bypassed	Pruning and the filter toggle work as on `main`	The driver is the default driver with `try_decode`.
Scheduler tracks and releases pages itself	The decoder releases pages	Release follows what the decoder has emitted.
Page index always loaded	Loaded only for row groups larger than the window	Pages give nothing over column chunks when the row group fits the window.
`DF_FETCH_POLICY`, `DF_FETCH_BUDGET`, `DF_FETCH_CONDITIONAL` environment variables	`read_ahead_bytes` config option (also in proto)	Review feedback
Conditional ranges skipped by default	Always read ahead (speculative)	Benchmarks and other engines, see above
Window not accounted	`MemoryPool` reservation: required bytes `grow`, read-ahead `try_grow`	Bound memory by the pool, not only by `target_partitions × window`

The pinned arrow-rs branch is pydantic/arrow-rs claude/push-decoder-batch-granular-scan-plan-60.0.0 (commit 6bea7c8969): the 60.0.0 release, plus the test in apache/arrow-rs#11218, plus the selection-policy commit 08d956bca6 (first in apache/arrow-rs#11241, now part of apache/arrow-rs#11240), plus apache/arrow-rs#10555 at 0d91bca5e1, plus the upstream fix apache/arrow-rs#11182 (a test in apache/arrow-rs#10555 needs it). Benchmarks against main therefore measure only these changes. This pin is why the PR cannot merge as-is.

This PR is two commits on main: the pin, then the streaming code. The earlier revisions (sync reader, pipelined and batched policies) are in branch claude/datafusion-rowgroup-buffering-history.

What this does not do

No fixed headroom in the pool. Read-ahead can take the last free bytes of a GreedyMemoryPool before another operator asks for them. A headroom rule (for example, stop read-ahead above X% of the pool) is a follow-up that needs measurements.
One read-ahead fetch is in flight per stream (the same as the previous revision). Each fetch fans out through get_byte_ranges.

Are these changes tested?

sqllogictest: 523 of 523 files pass with the default configuration. With read_ahead_bytes forced to 65536 for the whole suite (a local default change, not committed), 522 of 523 pass. The one difference is information_schema.slt, which prints the changed default. The small window forces many read-ahead waves and releases.
pushdown_filters=true (forced on with a local patch, not committed) for 18 filter-heavy files: the default and streaming policies fail on the same statements, all EXPLAIN differences caused by pushdown itself. The --complete output is byte-identical between the two policies.
datafusion-datasource-parquet unit tests: all pass with read-ahead off and with it forced on. The proto round-trip tests for ParquetOptions pass.
New read-ahead tests (opener::test::read_ahead): a scan with a 64 KiB pool is correct and returns all reserved bytes; the reservation stays within the window plus one fetch and returns to 0; a filtered scan reads conditional ranges ahead with a large pool and reads exactly the baseline bytes with a 1-byte pool. A unit test covers the accounting helper.
The arrow-rs branches have tests that compare batch-granular decoding with row-group decoding (including a randomized equivalence test), and the plan with the ranges that NeedsData requests.
There are no new unit tests here: this is a measurement POC, and the default path is the same as main.

Earlier results

Previous revision (16c5ba3, sync reader), adriangbot GKE vs merge-base, 100 MB window. Full tables are in the benchmark summary.

suite	with `SIMULATE_LATENCY`	without latency
tpch_sf10	7.04x	not run
tpch_sf1	1.73x	1.01x
tpcds_sf1	1.01x	0.98x
clickbench_partitioned	1.96x	0.97x

New runs for this revision are requested below.

Are there any user-facing changes?

Yes: one new configuration option, off by default, and one new FileSource method with a default implementation. With the defaults, behavior is the same as main. The [patch.crates-io] pin means this PR cannot merge in its current form.

🤖 Generated with Claude Code

adriangb · 2026-08-04T14:24:56Z

run benchmarks

env:
    DF_FETCH_POLICY: "streaming"
    DF_FETCH_BUDGET: "104857600"
    SIMULATE_LATENCY: "true"

adriangb · 2026-08-04T14:24:57Z

run benchmarks

env:
    DF_FETCH_POLICY: "streaming"
    DF_FETCH_BUDGET: "104857600"

adriangb · 2026-08-04T14:25:03Z

run benchmarks

env:
    DF_FETCH_POLICY: "pipelined"
    DF_FETCH_BUDGET: "104857600"
    SIMULATE_LATENCY: "true"

adriangb · 2026-08-04T14:25:04Z

run benchmarks

env:
    DF_FETCH_POLICY: "streaming"
    DF_FETCH_BUDGET: "104857600"
    SIMULATE_LATENCY: "true"
    DATAFUSION_RUNTIME_MEMORY_LIMIT: "4G"

adriangbot · 2026-08-04T14:26:55Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180419692-1375-ndr8t 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:27:00Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180419692-1377-vfkqd 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpch
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:27:48Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180419475-1379-h7nxx 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:27:50Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180419692-1376-78n2d 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:28:15Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180419475-1378-h4nsn 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:41:16Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpch
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃     HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │ 38.44 ms │                                    38.84 ms │    no change │
│ QQuery 2  │ 19.38 ms │                                    19.33 ms │    no change │
│ QQuery 3  │ 31.32 ms │                                    32.03 ms │    no change │
│ QQuery 4  │ 17.63 ms │                                    17.53 ms │    no change │
│ QQuery 5  │ 37.58 ms │                                    41.14 ms │ 1.09x slower │
│ QQuery 6  │ 16.61 ms │                                    16.63 ms │    no change │
│ QQuery 7  │ 43.51 ms │                                    46.85 ms │ 1.08x slower │
│ QQuery 8  │ 43.47 ms │                                    43.62 ms │    no change │
│ QQuery 9  │ 50.20 ms │                                    49.18 ms │    no change │
│ QQuery 10 │ 42.25 ms │                                    43.20 ms │    no change │
│ QQuery 11 │ 13.75 ms │                                    13.84 ms │    no change │
│ QQuery 12 │ 24.12 ms │                                    24.09 ms │    no change │
│ QQuery 13 │ 32.53 ms │                                    32.76 ms │    no change │
│ QQuery 14 │ 23.76 ms │                                    24.20 ms │    no change │
│ QQuery 15 │ 31.46 ms │                                    31.72 ms │    no change │
│ QQuery 16 │ 13.96 ms │                                    14.30 ms │    no change │
│ QQuery 17 │ 71.83 ms │                                    72.11 ms │    no change │
│ QQuery 18 │ 59.17 ms │                                    60.45 ms │    no change │
│ QQuery 19 │ 33.10 ms │                                    34.16 ms │    no change │
│ QQuery 20 │ 31.72 ms │                                    33.08 ms │    no change │
│ QQuery 21 │ 55.84 ms │                                    55.60 ms │    no change │
│ QQuery 22 │ 14.39 ms │                                    14.65 ms │    no change │
└───────────┴──────────┴─────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 746.02ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 759.32ms │
│ Average Time (HEAD)                                        │  33.91ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  34.51ms │
│ Queries Faster                                             │        0 │
│ Queries Slower                                             │        2 │
│ Queries with No Change                                     │       20 │
│ Queries with Failure                                       │        0 │
└────────────────────────────────────────────────────────────┴──────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │ 38.44 / 39.58 ±1.15 / 41.52 ms │              38.84 / 39.62 ±1.00 / 41.54 ms │     no change │
│ QQuery 2  │ 19.38 / 20.36 ±0.91 / 21.61 ms │              19.33 / 19.86 ±0.30 / 20.16 ms │     no change │
│ QQuery 3  │ 31.32 / 33.03 ±2.06 / 36.48 ms │              32.03 / 32.78 ±0.81 / 34.01 ms │     no change │
│ QQuery 4  │ 17.63 / 17.80 ±0.10 / 17.90 ms │              17.53 / 18.20 ±0.73 / 19.62 ms │     no change │
│ QQuery 5  │ 37.58 / 39.74 ±1.88 / 42.67 ms │              41.14 / 41.80 ±0.67 / 43.00 ms │  1.05x slower │
│ QQuery 6  │ 16.61 / 18.06 ±1.19 / 19.97 ms │              16.63 / 16.93 ±0.32 / 17.49 ms │ +1.07x faster │
│ QQuery 7  │ 43.51 / 46.57 ±2.24 / 49.45 ms │              46.85 / 49.08 ±3.61 / 56.27 ms │  1.05x slower │
│ QQuery 8  │ 43.47 / 44.25 ±1.14 / 46.51 ms │              43.62 / 43.86 ±0.17 / 44.15 ms │     no change │
│ QQuery 9  │ 50.20 / 50.70 ±0.50 / 51.64 ms │              49.18 / 49.99 ±0.45 / 50.49 ms │     no change │
│ QQuery 10 │ 42.25 / 43.42 ±0.82 / 44.77 ms │              43.20 / 43.62 ±0.42 / 44.25 ms │     no change │
│ QQuery 11 │ 13.75 / 14.11 ±0.66 / 15.43 ms │              13.84 / 13.98 ±0.09 / 14.06 ms │     no change │
│ QQuery 12 │ 24.12 / 24.68 ±0.38 / 25.29 ms │              24.09 / 24.85 ±0.53 / 25.59 ms │     no change │
│ QQuery 13 │ 32.53 / 35.43 ±4.00 / 43.35 ms │              32.76 / 34.20 ±1.12 / 35.75 ms │     no change │
│ QQuery 14 │ 23.76 / 23.92 ±0.22 / 24.35 ms │              24.20 / 24.58 ±0.50 / 25.56 ms │     no change │
│ QQuery 15 │ 31.46 / 32.76 ±1.33 / 35.12 ms │              31.72 / 33.06 ±1.37 / 35.59 ms │     no change │
│ QQuery 16 │ 13.96 / 14.12 ±0.11 / 14.27 ms │              14.30 / 14.50 ±0.21 / 14.89 ms │     no change │
│ QQuery 17 │ 71.83 / 73.57 ±1.58 / 76.14 ms │              72.11 / 73.73 ±0.86 / 74.38 ms │     no change │
│ QQuery 18 │ 59.17 / 61.35 ±2.10 / 64.51 ms │              60.45 / 63.61 ±3.02 / 69.12 ms │     no change │
│ QQuery 19 │ 33.10 / 33.40 ±0.21 / 33.73 ms │              34.16 / 34.43 ±0.23 / 34.83 ms │     no change │
│ QQuery 20 │ 31.72 / 32.13 ±0.33 / 32.72 ms │              33.08 / 33.24 ±0.14 / 33.45 ms │     no change │
│ QQuery 21 │ 55.84 / 56.65 ±0.79 / 58.12 ms │              55.60 / 59.29 ±1.96 / 60.84 ms │     no change │
│ QQuery 22 │ 14.39 / 14.69 ±0.29 / 15.22 ms │              14.65 / 15.02 ±0.19 / 15.22 ms │     no change │
└───────────┴────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 770.32ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 780.20ms │
│ Average Time (HEAD)                                        │  35.01ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  35.46ms │
│ Queries Faster                                             │        1 │
│ Queries Slower                                             │        2 │
│ Queries with No Change                                     │       19 │
│ Queries with Failure                                       │        0 │
└────────────────────────────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	1.2 GiB
Avg memory	505.4 MiB
CPU user	22.1s
CPU sys	1.8s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	1.1 GiB
Avg memory	507.7 MiB
CPU user	22.6s
CPU sys	1.8s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:42:35Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180419475-1380-mkpj5 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpch
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:43:03Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.59 ms │                                     5.86 ms │    no change │
│ QQuery 2  │   80.06 ms │                                    82.46 ms │    no change │
│ QQuery 3  │   29.61 ms │                                    30.33 ms │    no change │
│ QQuery 4  │  518.47 ms │                                   536.84 ms │    no change │
│ QQuery 5  │   51.95 ms │                                    53.13 ms │    no change │
│ QQuery 6  │   37.21 ms │                                    38.19 ms │    no change │
│ QQuery 7  │   95.97 ms │                                    97.81 ms │    no change │
│ QQuery 8  │   37.63 ms │                                    37.94 ms │    no change │
│ QQuery 9  │   53.78 ms │                                    52.98 ms │    no change │
│ QQuery 10 │   64.34 ms │                                    64.87 ms │    no change │
│ QQuery 11 │  327.66 ms │                                   333.62 ms │    no change │
│ QQuery 12 │   29.75 ms │                                    29.57 ms │    no change │
│ QQuery 13 │  121.16 ms │                                   122.51 ms │    no change │
│ QQuery 14 │  420.58 ms │                                   423.28 ms │    no change │
│ QQuery 15 │   60.74 ms │                                    63.70 ms │    no change │
│ QQuery 16 │    6.94 ms │                                     6.94 ms │    no change │
│ QQuery 17 │   81.87 ms │                                    83.03 ms │    no change │
│ QQuery 18 │  127.42 ms │                                   128.86 ms │    no change │
│ QQuery 19 │   42.78 ms │                                    43.27 ms │    no change │
│ QQuery 20 │   37.20 ms │                                    37.19 ms │    no change │
│ QQuery 21 │   18.03 ms │                                    18.08 ms │    no change │
│ QQuery 22 │   66.16 ms │                                    67.57 ms │    no change │
│ QQuery 23 │  366.73 ms │                                   365.48 ms │    no change │
│ QQuery 24 │  227.27 ms │                                   233.07 ms │    no change │
│ QQuery 25 │  112.12 ms │                                   114.53 ms │    no change │
│ QQuery 26 │   57.97 ms │                                    59.77 ms │    no change │
│ QQuery 27 │    6.55 ms │                                     6.65 ms │    no change │
│ QQuery 28 │   57.07 ms │                                    63.03 ms │ 1.10x slower │
│ QQuery 29 │   99.39 ms │                                    99.41 ms │    no change │
│ QQuery 30 │   33.78 ms │                                    34.19 ms │    no change │
│ QQuery 31 │  112.54 ms │                                   115.08 ms │    no change │
│ QQuery 32 │   21.06 ms │                                    21.18 ms │    no change │
│ QQuery 33 │   38.13 ms │                                    38.69 ms │    no change │
│ QQuery 34 │   10.34 ms │                                    10.40 ms │    no change │
│ QQuery 35 │   76.11 ms │                                    77.35 ms │    no change │
│ QQuery 36 │    6.08 ms │                                     6.14 ms │    no change │
│ QQuery 37 │    7.13 ms │                                     7.10 ms │    no change │
│ QQuery 38 │   64.39 ms │                                    64.71 ms │    no change │
│ QQuery 39 │   93.05 ms │                                    93.39 ms │    no change │
│ QQuery 40 │   23.93 ms │                                    24.31 ms │    no change │
│ QQuery 41 │   11.95 ms │                                    11.85 ms │    no change │
│ QQuery 42 │   24.28 ms │                                    25.27 ms │    no change │
│ QQuery 43 │    5.07 ms │                                     5.26 ms │    no change │
│ QQuery 44 │    9.59 ms │                                     9.68 ms │    no change │
│ QQuery 45 │   41.41 ms │                                    42.19 ms │    no change │
│ QQuery 46 │   12.11 ms │                                    12.46 ms │    no change │
│ QQuery 47 │  245.59 ms │                                   252.65 ms │    no change │
│ QQuery 48 │   97.67 ms │                                    99.47 ms │    no change │
│ QQuery 49 │   77.24 ms │                                    78.65 ms │    no change │
│ QQuery 50 │   60.29 ms │                                    61.47 ms │    no change │
│ QQuery 51 │   91.65 ms │                                    93.42 ms │    no change │
│ QQuery 52 │   24.88 ms │                                    25.21 ms │    no change │
│ QQuery 53 │   30.08 ms │                                    30.72 ms │    no change │
│ QQuery 54 │   56.07 ms │                                    58.16 ms │    no change │
│ QQuery 55 │   24.09 ms │                                    24.75 ms │    no change │
│ QQuery 56 │   40.07 ms │                                    40.32 ms │    no change │
│ QQuery 57 │  178.38 ms │                                   182.62 ms │    no change │
│ QQuery 58 │  115.54 ms │                                   118.11 ms │    no change │
│ QQuery 59 │  117.94 ms │                                   119.70 ms │    no change │
│ QQuery 60 │   39.97 ms │                                    40.88 ms │    no change │
│ QQuery 61 │   12.39 ms │                                    12.73 ms │    no change │
│ QQuery 62 │   46.66 ms │                                    48.33 ms │    no change │
│ QQuery 63 │   29.80 ms │                                    30.85 ms │    no change │
│ QQuery 64 │  417.47 ms │                                   425.37 ms │    no change │
│ QQuery 65 │  126.27 ms │                                   127.01 ms │    no change │
│ QQuery 66 │   82.26 ms │                                    82.82 ms │    no change │
│ QQuery 67 │  258.66 ms │                                   263.25 ms │    no change │
│ QQuery 68 │   12.55 ms │                                    12.61 ms │    no change │
│ QQuery 69 │   58.50 ms │                                    59.53 ms │    no change │
│ QQuery 70 │  107.00 ms │                                   109.52 ms │    no change │
│ QQuery 71 │   35.81 ms │                                    36.64 ms │    no change │
│ QQuery 72 │ 2160.72 ms │                                  2174.71 ms │    no change │
│ QQuery 73 │   10.14 ms │                                    10.13 ms │    no change │
│ QQuery 74 │  189.65 ms │                                   189.01 ms │    no change │
│ QQuery 75 │  152.77 ms │                                   152.66 ms │    no change │
│ QQuery 76 │   36.10 ms │                                    37.55 ms │    no change │
│ QQuery 77 │   62.26 ms │                                    62.94 ms │    no change │
│ QQuery 78 │  203.25 ms │                                   206.22 ms │    no change │
│ QQuery 79 │   69.69 ms │                                    69.29 ms │    no change │
│ QQuery 80 │  100.39 ms │                                   101.65 ms │    no change │
│ QQuery 81 │   27.04 ms │                                    26.84 ms │    no change │
│ QQuery 82 │   17.21 ms │                                    17.41 ms │    no change │
│ QQuery 83 │   41.89 ms │                                    40.72 ms │    no change │
│ QQuery 84 │   31.05 ms │                                    31.27 ms │    no change │
│ QQuery 85 │  108.45 ms │                                   110.19 ms │    no change │
│ QQuery 86 │   26.19 ms │                                    26.09 ms │    no change │
│ QQuery 87 │   64.98 ms │                                    64.58 ms │    no change │
│ QQuery 88 │   64.06 ms │                                    63.99 ms │    no change │
│ QQuery 89 │   36.21 ms │                                    37.43 ms │    no change │
│ QQuery 90 │   17.64 ms │                                    17.64 ms │    no change │
│ QQuery 91 │   47.49 ms │                                    47.47 ms │    no change │
│ QQuery 92 │   30.55 ms │                                    30.66 ms │    no change │
│ QQuery 93 │   51.54 ms │                                    52.18 ms │    no change │
│ QQuery 94 │   39.05 ms │                                    39.50 ms │    no change │
│ QQuery 95 │   82.78 ms │                                    82.79 ms │    no change │
│ QQuery 96 │   24.74 ms │                                    24.74 ms │    no change │
│ QQuery 97 │   48.13 ms │                                    47.87 ms │    no change │
│ QQuery 98 │   43.77 ms │                                    44.03 ms │    no change │
│ QQuery 99 │   71.47 ms │                                    71.66 ms │    no change │
└───────────┴────────────┴─────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 10078.99ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 10215.27ms │
│ Average Time (HEAD)                                        │   101.81ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   103.18ms │
│ Queries Faster                                             │          0 │
│ Queries Slower                                             │          1 │
│ Queries with No Change                                     │         98 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.59 / 6.11 ±0.87 / 7.85 ms │                 5.86 / 6.38 ±0.92 / 8.22 ms │     no change │
│ QQuery 2  │        80.06 / 80.45 ±0.40 / 81.14 ms │              82.46 / 82.59 ±0.12 / 82.75 ms │     no change │
│ QQuery 3  │        29.61 / 29.87 ±0.20 / 30.20 ms │              30.33 / 30.61 ±0.20 / 30.94 ms │     no change │
│ QQuery 4  │     518.47 / 526.75 ±8.55 / 542.96 ms │           536.84 / 545.00 ±4.77 / 549.52 ms │     no change │
│ QQuery 5  │        51.95 / 52.50 ±0.39 / 53.06 ms │              53.13 / 53.65 ±0.39 / 54.31 ms │     no change │
│ QQuery 6  │        37.21 / 37.41 ±0.15 / 37.65 ms │              38.19 / 38.63 ±0.38 / 39.31 ms │     no change │
│ QQuery 7  │       95.97 / 97.46 ±1.60 / 100.42 ms │            97.81 / 100.87 ±4.67 / 109.93 ms │     no change │
│ QQuery 8  │        37.63 / 38.31 ±0.65 / 39.38 ms │              37.94 / 38.45 ±0.58 / 39.55 ms │     no change │
│ QQuery 9  │        53.78 / 55.05 ±1.33 / 57.34 ms │              52.98 / 54.68 ±1.10 / 56.01 ms │     no change │
│ QQuery 10 │        64.34 / 65.23 ±0.52 / 65.83 ms │              64.87 / 66.86 ±2.83 / 72.39 ms │     no change │
│ QQuery 11 │     327.66 / 333.00 ±3.80 / 336.91 ms │           333.62 / 341.25 ±5.83 / 350.90 ms │     no change │
│ QQuery 12 │        29.75 / 30.14 ±0.23 / 30.34 ms │              29.57 / 30.11 ±0.33 / 30.44 ms │     no change │
│ QQuery 13 │     121.16 / 123.32 ±3.21 / 129.72 ms │           122.51 / 125.10 ±3.12 / 131.10 ms │     no change │
│ QQuery 14 │     420.58 / 425.58 ±4.70 / 434.16 ms │           423.28 / 429.35 ±7.96 / 444.22 ms │     no change │
│ QQuery 15 │        60.74 / 61.22 ±0.33 / 61.69 ms │              63.70 / 66.92 ±5.49 / 77.87 ms │  1.09x slower │
│ QQuery 16 │           6.94 / 7.07 ±0.12 / 7.22 ms │                 6.94 / 7.12 ±0.17 / 7.42 ms │     no change │
│ QQuery 17 │        81.87 / 84.86 ±3.17 / 90.87 ms │              83.03 / 84.47 ±1.67 / 87.74 ms │     no change │
│ QQuery 18 │     127.42 / 128.86 ±1.04 / 130.27 ms │           128.86 / 131.72 ±3.50 / 138.42 ms │     no change │
│ QQuery 19 │        42.78 / 44.40 ±1.49 / 46.17 ms │              43.27 / 43.75 ±0.32 / 44.21 ms │     no change │
│ QQuery 20 │        37.20 / 38.25 ±0.62 / 39.04 ms │              37.19 / 37.64 ±0.32 / 38.11 ms │     no change │
│ QQuery 21 │        18.03 / 18.26 ±0.15 / 18.48 ms │              18.08 / 18.19 ±0.10 / 18.36 ms │     no change │
│ QQuery 22 │        66.16 / 67.01 ±0.82 / 68.29 ms │              67.57 / 68.25 ±0.49 / 69.08 ms │     no change │
│ QQuery 23 │     366.73 / 371.84 ±4.48 / 379.88 ms │           365.48 / 371.59 ±7.32 / 385.81 ms │     no change │
│ QQuery 24 │     227.27 / 229.92 ±2.77 / 234.61 ms │           233.07 / 234.79 ±1.11 / 236.24 ms │     no change │
│ QQuery 25 │     112.12 / 114.46 ±1.48 / 116.60 ms │           114.53 / 119.88 ±7.74 / 135.06 ms │     no change │
│ QQuery 26 │        57.97 / 58.62 ±0.51 / 59.25 ms │              59.77 / 60.02 ±0.16 / 60.25 ms │     no change │
│ QQuery 27 │           6.55 / 6.79 ±0.21 / 7.15 ms │                 6.65 / 6.70 ±0.04 / 6.78 ms │     no change │
│ QQuery 28 │        57.07 / 60.14 ±2.02 / 61.95 ms │              63.03 / 66.01 ±4.81 / 75.55 ms │  1.10x slower │
│ QQuery 29 │      99.39 / 102.80 ±5.19 / 113.09 ms │            99.41 / 101.69 ±1.95 / 105.25 ms │     no change │
│ QQuery 30 │        33.78 / 34.26 ±0.50 / 35.17 ms │              34.19 / 34.81 ±0.51 / 35.56 ms │     no change │
│ QQuery 31 │     112.54 / 113.60 ±0.75 / 114.57 ms │           115.08 / 117.81 ±4.00 / 125.76 ms │     no change │
│ QQuery 32 │        21.06 / 21.28 ±0.13 / 21.44 ms │              21.18 / 21.54 ±0.24 / 21.90 ms │     no change │
│ QQuery 33 │        38.13 / 38.57 ±0.34 / 39.05 ms │              38.69 / 38.99 ±0.20 / 39.32 ms │     no change │
│ QQuery 34 │        10.34 / 10.50 ±0.20 / 10.88 ms │              10.40 / 11.06 ±0.65 / 12.11 ms │  1.05x slower │
│ QQuery 35 │        76.11 / 76.41 ±0.27 / 76.84 ms │              77.35 / 79.92 ±3.88 / 87.62 ms │     no change │
│ QQuery 36 │           6.08 / 6.20 ±0.14 / 6.48 ms │                 6.14 / 6.26 ±0.15 / 6.55 ms │     no change │
│ QQuery 37 │           7.13 / 7.21 ±0.05 / 7.27 ms │                 7.10 / 7.29 ±0.11 / 7.41 ms │     no change │
│ QQuery 38 │        64.39 / 66.20 ±2.12 / 70.27 ms │              64.71 / 65.57 ±0.71 / 66.79 ms │     no change │
│ QQuery 39 │        93.05 / 93.89 ±1.06 / 95.95 ms │             93.39 / 96.66 ±4.77 / 106.14 ms │     no change │
│ QQuery 40 │        23.93 / 24.14 ±0.14 / 24.35 ms │              24.31 / 24.76 ±0.33 / 25.21 ms │     no change │
│ QQuery 41 │        11.95 / 12.07 ±0.08 / 12.19 ms │              11.85 / 12.10 ±0.15 / 12.29 ms │     no change │
│ QQuery 42 │        24.28 / 25.11 ±0.65 / 25.94 ms │              25.27 / 25.74 ±0.26 / 26.03 ms │     no change │
│ QQuery 43 │           5.07 / 5.26 ±0.14 / 5.47 ms │                 5.26 / 5.40 ±0.16 / 5.68 ms │     no change │
│ QQuery 44 │           9.59 / 9.70 ±0.08 / 9.80 ms │                 9.68 / 9.78 ±0.08 / 9.88 ms │     no change │
│ QQuery 45 │        41.41 / 41.93 ±0.33 / 42.42 ms │              42.19 / 43.14 ±0.88 / 44.73 ms │     no change │
│ QQuery 46 │        12.11 / 12.37 ±0.19 / 12.67 ms │              12.46 / 12.81 ±0.24 / 13.17 ms │     no change │
│ QQuery 47 │     245.59 / 248.45 ±2.06 / 251.42 ms │           252.65 / 256.24 ±3.75 / 263.49 ms │     no change │
│ QQuery 48 │       97.67 / 98.96 ±1.26 / 100.66 ms │            99.47 / 100.02 ±0.63 / 101.15 ms │     no change │
│ QQuery 49 │        77.24 / 77.83 ±0.32 / 78.18 ms │              78.65 / 79.21 ±0.48 / 80.05 ms │     no change │
│ QQuery 50 │        60.29 / 60.69 ±0.23 / 60.92 ms │              61.47 / 63.98 ±3.99 / 71.93 ms │  1.05x slower │
│ QQuery 51 │       91.65 / 95.56 ±4.09 / 103.44 ms │              93.42 / 94.32 ±0.75 / 95.40 ms │     no change │
│ QQuery 52 │        24.88 / 25.02 ±0.14 / 25.28 ms │              25.21 / 25.40 ±0.17 / 25.63 ms │     no change │
│ QQuery 53 │        30.08 / 30.34 ±0.29 / 30.77 ms │              30.72 / 34.21 ±6.55 / 47.31 ms │  1.13x slower │
│ QQuery 54 │        56.07 / 56.45 ±0.22 / 56.73 ms │              58.16 / 58.43 ±0.28 / 58.94 ms │     no change │
│ QQuery 55 │        24.09 / 24.58 ±0.36 / 25.17 ms │              24.75 / 25.16 ±0.35 / 25.66 ms │     no change │
│ QQuery 56 │        40.07 / 41.61 ±2.60 / 46.80 ms │              40.32 / 40.71 ±0.36 / 41.30 ms │     no change │
│ QQuery 57 │     178.38 / 180.68 ±2.91 / 186.43 ms │           182.62 / 185.95 ±3.89 / 193.38 ms │     no change │
│ QQuery 58 │     115.54 / 117.39 ±1.70 / 120.25 ms │           118.11 / 120.14 ±2.10 / 124.02 ms │     no change │
│ QQuery 59 │     117.94 / 119.34 ±1.11 / 120.72 ms │           119.70 / 120.17 ±0.27 / 120.49 ms │     no change │
│ QQuery 60 │        39.97 / 40.59 ±0.52 / 41.27 ms │              40.88 / 41.19 ±0.28 / 41.60 ms │     no change │
│ QQuery 61 │        12.39 / 12.49 ±0.12 / 12.71 ms │              12.73 / 15.08 ±4.43 / 23.95 ms │  1.21x slower │
│ QQuery 62 │        46.66 / 47.65 ±0.82 / 48.72 ms │              48.33 / 48.53 ±0.18 / 48.85 ms │     no change │
│ QQuery 63 │        29.80 / 30.27 ±0.37 / 30.92 ms │              30.85 / 31.42 ±0.40 / 31.91 ms │     no change │
│ QQuery 64 │     417.47 / 422.16 ±3.19 / 427.16 ms │           425.37 / 429.45 ±2.95 / 432.15 ms │     no change │
│ QQuery 65 │     126.27 / 127.37 ±0.80 / 128.16 ms │           127.01 / 129.85 ±4.25 / 138.29 ms │     no change │
│ QQuery 66 │        82.26 / 82.79 ±0.36 / 83.40 ms │              82.82 / 83.34 ±0.55 / 84.37 ms │     no change │
│ QQuery 67 │     258.66 / 263.41 ±6.78 / 276.89 ms │           263.25 / 267.95 ±4.25 / 275.54 ms │     no change │
│ QQuery 68 │        12.55 / 12.64 ±0.08 / 12.75 ms │              12.61 / 16.69 ±6.47 / 29.60 ms │  1.32x slower │
│ QQuery 69 │        58.50 / 59.24 ±0.66 / 60.48 ms │              59.53 / 60.12 ±0.45 / 60.82 ms │     no change │
│ QQuery 70 │     107.00 / 110.52 ±3.86 / 117.92 ms │           109.52 / 112.01 ±3.45 / 118.55 ms │     no change │
│ QQuery 71 │        35.81 / 36.33 ±0.35 / 36.80 ms │              36.64 / 38.72 ±2.91 / 44.45 ms │  1.07x slower │
│ QQuery 72 │ 2160.72 / 2232.41 ±76.68 / 2360.93 ms │       2174.71 / 2203.23 ±31.47 / 2262.39 ms │     no change │
│ QQuery 73 │        10.14 / 10.28 ±0.13 / 10.44 ms │              10.13 / 10.47 ±0.23 / 10.78 ms │     no change │
│ QQuery 74 │     189.65 / 196.56 ±9.94 / 216.17 ms │           189.01 / 192.14 ±3.91 / 199.86 ms │     no change │
│ QQuery 75 │     152.77 / 155.49 ±4.21 / 163.88 ms │           152.66 / 153.36 ±0.41 / 153.70 ms │     no change │
│ QQuery 76 │        36.10 / 36.64 ±0.51 / 37.59 ms │              37.55 / 40.72 ±5.31 / 51.28 ms │  1.11x slower │
│ QQuery 77 │        62.26 / 62.86 ±0.58 / 63.89 ms │              62.94 / 63.66 ±0.38 / 64.00 ms │     no change │
│ QQuery 78 │     203.25 / 208.07 ±3.67 / 214.09 ms │           206.22 / 209.22 ±3.73 / 216.38 ms │     no change │
│ QQuery 79 │        69.69 / 71.01 ±1.17 / 73.10 ms │              69.29 / 70.02 ±0.79 / 71.50 ms │     no change │
│ QQuery 80 │     100.39 / 103.05 ±1.90 / 105.47 ms │           101.65 / 107.25 ±8.55 / 124.26 ms │     no change │
│ QQuery 81 │        27.04 / 27.30 ±0.19 / 27.52 ms │              26.84 / 27.21 ±0.39 / 27.84 ms │     no change │
│ QQuery 82 │        17.21 / 18.64 ±1.74 / 21.98 ms │              17.41 / 17.61 ±0.21 / 17.98 ms │ +1.06x faster │
│ QQuery 83 │        41.89 / 44.67 ±5.05 / 54.76 ms │              40.72 / 41.24 ±0.38 / 41.86 ms │ +1.08x faster │
│ QQuery 84 │        31.05 / 31.78 ±0.52 / 32.54 ms │              31.27 / 31.48 ±0.26 / 31.99 ms │     no change │
│ QQuery 85 │     108.45 / 108.79 ±0.19 / 109.03 ms │           110.19 / 113.12 ±3.76 / 120.23 ms │     no change │
│ QQuery 86 │        26.19 / 28.62 ±2.77 / 33.62 ms │              26.09 / 26.54 ±0.28 / 26.95 ms │ +1.08x faster │
│ QQuery 87 │        64.98 / 66.35 ±1.28 / 68.71 ms │              64.58 / 66.00 ±0.92 / 67.09 ms │     no change │
│ QQuery 88 │        64.06 / 64.37 ±0.28 / 64.73 ms │              63.99 / 67.36 ±4.33 / 75.89 ms │     no change │
│ QQuery 89 │        36.21 / 36.39 ±0.11 / 36.55 ms │              37.43 / 37.82 ±0.34 / 38.32 ms │     no change │
│ QQuery 90 │        17.64 / 18.40 ±0.92 / 19.98 ms │              17.64 / 17.81 ±0.10 / 17.91 ms │     no change │
│ QQuery 91 │        47.49 / 49.37 ±1.85 / 52.85 ms │              47.47 / 48.01 ±0.67 / 49.33 ms │     no change │
│ QQuery 92 │        30.55 / 31.06 ±0.48 / 31.96 ms │              30.66 / 31.17 ±0.35 / 31.68 ms │     no change │
│ QQuery 93 │        51.54 / 52.27 ±0.94 / 54.03 ms │              52.18 / 53.03 ±0.89 / 54.53 ms │     no change │
│ QQuery 94 │        39.05 / 39.36 ±0.23 / 39.74 ms │              39.50 / 40.30 ±0.54 / 41.00 ms │     no change │
│ QQuery 95 │        82.78 / 84.29 ±2.30 / 88.82 ms │              82.79 / 83.67 ±1.16 / 85.91 ms │     no change │
│ QQuery 96 │        24.74 / 24.95 ±0.16 / 25.21 ms │              24.74 / 25.02 ±0.18 / 25.30 ms │     no change │
│ QQuery 97 │        48.13 / 48.56 ±0.35 / 49.18 ms │              47.87 / 48.97 ±0.64 / 49.84 ms │     no change │
│ QQuery 98 │        43.77 / 44.28 ±0.57 / 45.21 ms │              44.03 / 45.02 ±0.65 / 45.99 ms │     no change │
│ QQuery 99 │        71.47 / 73.89 ±2.65 / 78.33 ms │              71.66 / 71.88 ±0.21 / 72.26 ms │     no change │
└───────────┴───────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 10286.42ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 10397.48ms │
│ Average Time (HEAD)                                        │   103.90ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   105.03ms │
│ Queries Faster                                             │          3 │
│ Queries Slower                                             │          9 │
│ Queries with No Change                                     │         87 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	55.0s
Peak memory	2.1 GiB
Avg memory	1.4 GiB
CPU user	232.5s
CPU sys	5.8s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	1.4 GiB
Avg memory	907.0 MiB
CPU user	235.6s
CPU sys	6.8s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:43:56Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180420768-1384-q9724 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "pipelined"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:44:20Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180420975-1381-9sgf5 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "4G"
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:44:26Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180420975-1382-4ntbg 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "4G"
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:44:29Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180420975-1383-m44v4 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "4G"
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:46:28Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.55 ms │                                     1.66 ms │  1.07x slower │
│ QQuery 1  │   13.18 ms │                                    13.14 ms │     no change │
│ QQuery 2  │   37.62 ms │                                    36.75 ms │     no change │
│ QQuery 3  │   33.57 ms │                                    31.72 ms │ +1.06x faster │
│ QQuery 4  │  241.29 ms │                                   220.10 ms │ +1.10x faster │
│ QQuery 5  │  279.49 ms │                                   271.00 ms │     no change │
│ QQuery 6  │    1.26 ms │                                     1.27 ms │     no change │
│ QQuery 7  │   13.93 ms │                                    13.88 ms │     no change │
│ QQuery 8  │  331.27 ms │                                   330.42 ms │     no change │
│ QQuery 9  │  465.35 ms │                                   465.02 ms │     no change │
│ QQuery 10 │   71.61 ms │                                    70.37 ms │     no change │
│ QQuery 11 │   81.72 ms │                                    82.06 ms │     no change │
│ QQuery 12 │  269.53 ms │                                   276.61 ms │     no change │
│ QQuery 13 │  371.14 ms │                                   378.50 ms │     no change │
│ QQuery 14 │  289.73 ms │                                   289.01 ms │     no change │
│ QQuery 15 │  276.12 ms │                                   274.43 ms │     no change │
│ QQuery 16 │  631.72 ms │                                   619.33 ms │     no change │
│ QQuery 17 │  632.04 ms │                                   632.12 ms │     no change │
│ QQuery 18 │ 1448.49 ms │                                  1255.00 ms │ +1.15x faster │
│ QQuery 19 │   30.42 ms │                                    28.31 ms │ +1.07x faster │
│ QQuery 20 │  525.89 ms │                                   522.85 ms │     no change │
│ QQuery 21 │  542.06 ms │                                   516.94 ms │     no change │
│ QQuery 22 │ 1011.60 ms │                                   996.98 ms │     no change │
│ QQuery 23 │ 3139.04 ms │                                  3125.61 ms │     no change │
│ QQuery 24 │   42.01 ms │                                    41.06 ms │     no change │
│ QQuery 25 │  111.70 ms │                                   112.25 ms │     no change │
│ QQuery 26 │   42.21 ms │                                    41.82 ms │     no change │
│ QQuery 27 │  516.57 ms │                                   512.43 ms │     no change │
│ QQuery 28 │ 2926.27 ms │                                  2908.15 ms │     no change │
│ QQuery 29 │   41.22 ms │                                    41.65 ms │     no change │
│ QQuery 30 │  300.47 ms │                                   304.74 ms │     no change │
│ QQuery 31 │  290.59 ms │                                   289.09 ms │     no change │
│ QQuery 32 │  954.20 ms │                                   932.27 ms │     no change │
│ QQuery 33 │ 1503.27 ms │                                  1474.80 ms │     no change │
│ QQuery 34 │ 1481.14 ms │                                  1460.67 ms │     no change │
│ QQuery 35 │  275.22 ms │                                   345.39 ms │  1.25x slower │
│ QQuery 36 │   66.28 ms │                                    72.71 ms │  1.10x slower │
│ QQuery 37 │   35.50 ms │                                    38.67 ms │  1.09x slower │
│ QQuery 38 │   40.06 ms │                                    43.29 ms │  1.08x slower │
│ QQuery 39 │  145.84 ms │                                   168.00 ms │  1.15x slower │
│ QQuery 40 │   14.41 ms │                                    16.04 ms │  1.11x slower │
│ QQuery 41 │   13.94 ms │                                    15.35 ms │  1.10x slower │
│ QQuery 42 │   13.11 ms │                                    14.74 ms │  1.12x slower │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 19553.62ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 19286.23ms │
│ Average Time (HEAD)                                        │   454.74ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   448.52ms │
│ Queries Faster                                             │          4 │
│ Queries Slower                                             │          9 │
│ Queries with No Change                                     │         30 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.55 / 4.54 ±5.82 / 16.18 ms │                1.66 / 4.65 ±5.84 / 16.34 ms │     no change │
│ QQuery 1  │         13.18 / 13.94 ±0.46 / 14.48 ms │              13.14 / 13.57 ±0.36 / 14.11 ms │     no change │
│ QQuery 2  │         37.62 / 37.92 ±0.29 / 38.42 ms │              36.75 / 37.36 ±0.60 / 38.29 ms │     no change │
│ QQuery 3  │         33.57 / 34.32 ±0.61 / 35.37 ms │              31.72 / 31.96 ±0.29 / 32.44 ms │ +1.07x faster │
│ QQuery 4  │     241.29 / 273.11 ±16.55 / 289.63 ms │           220.10 / 228.27 ±8.55 / 244.84 ms │ +1.20x faster │
│ QQuery 5  │      279.49 / 283.86 ±5.52 / 294.43 ms │           271.00 / 275.14 ±2.49 / 277.81 ms │     no change │
│ QQuery 6  │            1.26 / 1.41 ±0.22 / 1.85 ms │                 1.27 / 1.43 ±0.22 / 1.85 ms │     no change │
│ QQuery 7  │         13.93 / 14.27 ±0.22 / 14.47 ms │              13.88 / 14.17 ±0.17 / 14.37 ms │     no change │
│ QQuery 8  │      331.27 / 336.59 ±3.86 / 341.56 ms │           330.42 / 333.40 ±3.20 / 339.04 ms │     no change │
│ QQuery 9  │      465.35 / 474.24 ±8.36 / 485.10 ms │          465.02 / 475.40 ±11.56 / 497.26 ms │     no change │
│ QQuery 10 │         71.61 / 74.38 ±2.63 / 78.83 ms │              70.37 / 71.95 ±1.41 / 74.07 ms │     no change │
│ QQuery 11 │         81.72 / 83.20 ±0.97 / 84.43 ms │             82.06 / 86.53 ±7.47 / 101.45 ms │     no change │
│ QQuery 12 │      269.53 / 275.56 ±3.05 / 277.61 ms │           276.61 / 278.58 ±1.45 / 280.11 ms │     no change │
│ QQuery 13 │      371.14 / 379.25 ±5.45 / 386.08 ms │           378.50 / 394.57 ±9.39 / 402.37 ms │     no change │
│ QQuery 14 │      289.73 / 293.31 ±1.83 / 294.85 ms │           289.01 / 296.69 ±5.90 / 303.58 ms │     no change │
│ QQuery 15 │      276.12 / 285.84 ±7.63 / 297.71 ms │           274.43 / 285.73 ±8.26 / 297.15 ms │     no change │
│ QQuery 16 │      631.72 / 641.41 ±6.88 / 650.93 ms │          619.33 / 680.01 ±46.22 / 732.46 ms │  1.06x slower │
│ QQuery 17 │     632.04 / 674.72 ±42.70 / 735.36 ms │          632.12 / 673.52 ±35.02 / 734.09 ms │     no change │
│ QQuery 18 │  1448.49 / 1486.38 ±23.94 / 1515.67 ms │       1255.00 / 1306.84 ±26.96 / 1331.26 ms │ +1.14x faster │
│ QQuery 19 │         30.42 / 33.99 ±6.36 / 46.69 ms │              28.31 / 35.01 ±9.89 / 53.85 ms │     no change │
│ QQuery 20 │     525.89 / 542.74 ±18.82 / 578.22 ms │           522.85 / 530.83 ±5.52 / 538.68 ms │     no change │
│ QQuery 21 │      542.06 / 551.04 ±6.59 / 560.61 ms │           516.94 / 525.16 ±4.16 / 528.05 ms │     no change │
│ QQuery 22 │   1011.60 / 1025.84 ±8.82 / 1035.15 ms │        996.98 / 1035.70 ±28.38 / 1064.40 ms │     no change │
│ QQuery 23 │  3139.04 / 3195.10 ±60.15 / 3281.58 ms │       3125.61 / 3255.60 ±89.48 / 3356.40 ms │     no change │
│ QQuery 24 │         42.01 / 43.86 ±1.99 / 46.74 ms │              41.06 / 42.22 ±0.90 / 43.51 ms │     no change │
│ QQuery 25 │      111.70 / 115.76 ±5.84 / 127.32 ms │           112.25 / 117.83 ±5.25 / 124.92 ms │     no change │
│ QQuery 26 │         42.21 / 44.61 ±4.01 / 52.60 ms │              41.82 / 44.75 ±2.08 / 47.22 ms │     no change │
│ QQuery 27 │      516.57 / 523.27 ±4.08 / 527.08 ms │           512.43 / 518.86 ±4.83 / 525.40 ms │     no change │
│ QQuery 28 │  2926.27 / 2971.67 ±37.67 / 3017.18 ms │       2908.15 / 2947.38 ±46.73 / 3038.02 ms │     no change │
│ QQuery 29 │         41.22 / 45.62 ±7.68 / 60.95 ms │              41.65 / 44.65 ±3.35 / 48.99 ms │     no change │
│ QQuery 30 │      300.47 / 311.19 ±6.14 / 317.85 ms │           304.74 / 308.25 ±3.15 / 311.94 ms │     no change │
│ QQuery 31 │      290.59 / 293.11 ±2.06 / 296.61 ms │          289.09 / 312.06 ±20.71 / 345.83 ms │  1.06x slower │
│ QQuery 32 │   954.20 / 1098.04 ±78.73 / 1191.97 ms │        932.27 / 1014.07 ±73.49 / 1150.26 ms │ +1.08x faster │
│ QQuery 33 │ 1503.27 / 1671.48 ±121.88 / 1842.16 ms │       1474.80 / 1490.75 ±16.68 / 1513.40 ms │ +1.12x faster │
│ QQuery 34 │ 1481.14 / 1607.74 ±109.63 / 1755.50 ms │       1460.67 / 1537.93 ±89.99 / 1711.61 ms │     no change │
│ QQuery 35 │     275.22 / 331.78 ±83.32 / 495.53 ms │          345.39 / 368.22 ±21.04 / 406.44 ms │  1.11x slower │
│ QQuery 36 │       66.28 / 82.39 ±17.76 / 115.31 ms │              72.71 / 81.59 ±7.92 / 94.25 ms │     no change │
│ QQuery 37 │         35.50 / 39.29 ±4.41 / 47.71 ms │              38.67 / 40.12 ±1.55 / 43.10 ms │     no change │
│ QQuery 38 │         40.06 / 41.00 ±1.01 / 42.38 ms │              43.29 / 47.93 ±4.91 / 57.36 ms │  1.17x slower │
│ QQuery 39 │     145.84 / 160.13 ±10.45 / 175.27 ms │           168.00 / 173.94 ±3.53 / 177.26 ms │  1.09x slower │
│ QQuery 40 │         14.41 / 14.76 ±0.26 / 15.17 ms │              16.04 / 21.13 ±9.46 / 40.04 ms │  1.43x slower │
│ QQuery 41 │         13.94 / 14.20 ±0.25 / 14.60 ms │              15.35 / 15.49 ±0.13 / 15.70 ms │  1.09x slower │
│ QQuery 42 │         13.11 / 13.63 ±0.34 / 14.00 ms │              14.74 / 14.94 ±0.14 / 15.10 ms │  1.10x slower │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 20440.49ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 20014.17ms │
│ Average Time (HEAD)                                        │   475.36ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   465.45ms │
│ Queries Faster                                             │          5 │
│ Queries Slower                                             │          8 │
│ Queries with No Change                                     │         30 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	105.0s
Peak memory	11.5 GiB
Avg memory	4.7 GiB
CPU user	1045.8s
CPU sys	78.2s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	105.0s
Peak memory	11.2 GiB
Avg memory	4.3 GiB
CPU user	1024.9s
CPU sys	76.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:55:45Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │  106.55 ms │                                   106.61 ms │     no change │
│ QQuery 2  │  511.94 ms │                                   764.90 ms │  1.49x slower │
│ QQuery 3  │  382.82 ms │                                   464.40 ms │  1.21x slower │
│ QQuery 4  │ 1194.30 ms │                                  1585.64 ms │  1.33x slower │
│ QQuery 5  │  472.14 ms │                                   606.93 ms │  1.29x slower │
│ QQuery 6  │  610.73 ms │                                   764.11 ms │  1.25x slower │
│ QQuery 7  │  546.23 ms │                                   885.30 ms │  1.62x slower │
│ QQuery 8  │  483.89 ms │                                   759.48 ms │  1.57x slower │
│ QQuery 9  │ 1374.60 ms │                                  1541.04 ms │  1.12x slower │
│ QQuery 10 │  755.03 ms │                                  1004.14 ms │  1.33x slower │
│ QQuery 11 │  872.33 ms │                                  1245.92 ms │  1.43x slower │
│ QQuery 12 │  344.38 ms │                                   379.58 ms │  1.10x slower │
│ QQuery 13 │  657.32 ms │                                   958.21 ms │  1.46x slower │
│ QQuery 14 │ 3929.87 ms │                                  4755.84 ms │  1.21x slower │
│ QQuery 15 │  408.01 ms │                                   583.39 ms │  1.43x slower │
│ QQuery 16 │  108.07 ms │                                   107.92 ms │     no change │
│ QQuery 17 │  631.93 ms │                                   995.17 ms │  1.57x slower │
│ QQuery 18 │  716.07 ms │                                  1013.09 ms │  1.41x slower │
│ QQuery 19 │  533.49 ms │                                   787.93 ms │  1.48x slower │
│ QQuery 20 │  326.84 ms │                                   464.15 ms │  1.42x slower │
│ QQuery 21 │  505.30 ms │                                   482.41 ms │     no change │
│ QQuery 22 │  476.28 ms │                                   525.24 ms │  1.10x slower │
│ QQuery 23 │ 2547.00 ms │                                  3533.07 ms │  1.39x slower │
│ QQuery 24 │ 2187.60 ms │                                  3118.91 ms │  1.43x slower │
│ QQuery 25 │  687.32 ms │                                   913.17 ms │  1.33x slower │
│ QQuery 26 │  599.35 ms │                                   704.95 ms │  1.18x slower │
│ QQuery 27 │  141.55 ms │                                   190.56 ms │  1.35x slower │
│ QQuery 28 │  344.91 ms │                                   425.90 ms │  1.23x slower │
│ QQuery 29 │  720.41 ms │                                  1068.21 ms │  1.48x slower │
│ QQuery 30 │  560.96 ms │                                   857.56 ms │  1.53x slower │
│ QQuery 31 │ 1193.11 ms │                                  2095.41 ms │  1.76x slower │
│ QQuery 32 │  414.92 ms │                                   482.52 ms │  1.16x slower │
│ QQuery 33 │  522.29 ms │                                   736.44 ms │  1.41x slower │
│ QQuery 34 │  194.91 ms │                                   228.07 ms │  1.17x slower │
│ QQuery 35 │  657.11 ms │                                  1012.73 ms │  1.54x slower │
│ QQuery 36 │  110.34 ms │                                   108.19 ms │     no change │
│ QQuery 37 │  220.33 ms │                                   237.11 ms │  1.08x slower │
│ QQuery 38 │  564.16 ms │                                   825.62 ms │  1.46x slower │
│ QQuery 39 │ 1845.58 ms │                                  2251.57 ms │  1.22x slower │
│ QQuery 40 │  466.61 ms │                                   579.31 ms │  1.24x slower │
│ QQuery 41 │  175.86 ms │                                   175.02 ms │     no change │
│ QQuery 42 │  313.33 ms │                                   423.92 ms │  1.35x slower │
│ QQuery 43 │  105.63 ms │                                    76.62 ms │ +1.38x faster │
│ QQuery 44 │  198.81 ms │                                   196.14 ms │     no change │
│ QQuery 45 │  462.94 ms │                                   665.34 ms │  1.44x slower │
│ QQuery 46 │  289.58 ms │                                   525.40 ms │  1.81x slower │
│ QQuery 47 │ 1187.06 ms │                                  1583.24 ms │  1.33x slower │
│ QQuery 48 │  599.59 ms │                                   859.87 ms │  1.43x slower │
│ QQuery 49 │  575.32 ms │                                   655.87 ms │  1.14x slower │
│ QQuery 50 │  528.05 ms │                                   783.09 ms │  1.48x slower │
│ QQuery 51 │  511.88 ms │                                   676.58 ms │  1.32x slower │
│ QQuery 52 │  307.71 ms │                                   442.29 ms │  1.44x slower │
│ QQuery 53 │  408.69 ms │                                   442.07 ms │  1.08x slower │
│ QQuery 54 │  857.62 ms │                                  1122.26 ms │  1.31x slower │
│ QQuery 55 │  377.16 ms │                                   424.88 ms │  1.13x slower │
│ QQuery 56 │  547.20 ms │                                   745.89 ms │  1.36x slower │
│ QQuery 57 │ 1031.63 ms │                                  1460.85 ms │  1.42x slower │
│ QQuery 58 │ 1155.61 ms │                                  1492.95 ms │  1.29x slower │
│ QQuery 59 │  897.47 ms │                                   945.77 ms │  1.05x slower │
│ QQuery 60 │  574.44 ms │                                   698.88 ms │  1.22x slower │
│ QQuery 61 │  394.04 ms │                                   392.37 ms │     no change │
│ QQuery 62 │  581.64 ms │                                   540.77 ms │ +1.08x faster │
│ QQuery 63 │  418.11 ms │                                   448.56 ms │  1.07x slower │
│ QQuery 64 │ 3180.76 ms │                                  4659.33 ms │  1.46x slower │
│ QQuery 65 │  602.43 ms │                                   968.02 ms │  1.61x slower │
│ QQuery 66 │  564.29 ms │                                   683.49 ms │  1.21x slower │
│ QQuery 67 │  612.03 ms │                                   709.45 ms │  1.16x slower │
│ QQuery 68 │  288.29 ms │                                   486.75 ms │  1.69x slower │
│ QQuery 69 │  700.33 ms │                                  1115.17 ms │  1.59x slower │
│ QQuery 70 │  672.84 ms │                                   806.26 ms │  1.20x slower │
│ QQuery 71 │  475.08 ms │                                   675.36 ms │  1.42x slower │
│ QQuery 72 │ 3005.47 ms │                                  3588.68 ms │  1.19x slower │
│ QQuery 73 │  195.44 ms │                                   256.95 ms │  1.31x slower │
│ QQuery 74 │  724.70 ms │                                  1180.16 ms │  1.63x slower │
│ QQuery 75 │  931.22 ms │                                  1242.33 ms │  1.33x slower │
│ QQuery 76 │  521.13 ms │                                   616.34 ms │  1.18x slower │
│ QQuery 77 │  603.99 ms │                                   872.06 ms │  1.44x slower │
│ QQuery 78 │  706.59 ms │                                  1352.28 ms │  1.91x slower │
│ QQuery 79 │  448.82 ms │                                   569.64 ms │  1.27x slower │
│ QQuery 80 │  649.05 ms │                                   898.91 ms │  1.38x slower │
│ QQuery 81 │  477.58 ms │                                   783.97 ms │  1.64x slower │
│ QQuery 82 │  555.56 ms │                                   745.39 ms │  1.34x slower │
│ QQuery 83 │  858.88 ms │                                  1140.15 ms │  1.33x slower │
│ QQuery 84 │  440.73 ms │                                   735.43 ms │  1.67x slower │
│ QQuery 85 │  813.54 ms │                                  1098.30 ms │  1.35x slower │
│ QQuery 86 │  248.85 ms │                                   372.77 ms │  1.50x slower │
│ QQuery 87 │  575.62 ms │                                   911.78 ms │  1.58x slower │
│ QQuery 88 │  737.84 ms │                                   791.46 ms │  1.07x slower │
│ QQuery 89 │  424.96 ms │                                   513.10 ms │  1.21x slower │
│ QQuery 90 │  425.49 ms │                                   457.07 ms │  1.07x slower │
│ QQuery 91 │  526.83 ms │                                  1044.24 ms │  1.98x slower │
│ QQuery 92 │  437.92 ms │                                   508.26 ms │  1.16x slower │
│ QQuery 93 │  458.45 ms │                                   535.88 ms │  1.17x slower │
│ QQuery 94 │  481.71 ms │                                   728.26 ms │  1.51x slower │
│ QQuery 95 │  688.04 ms │                                  1173.07 ms │  1.70x slower │
│ QQuery 96 │  403.26 ms │                                   502.76 ms │  1.25x slower │
│ QQuery 97 │  440.38 ms │                                   626.83 ms │  1.42x slower │
│ QQuery 98 │  401.47 ms │                                   495.62 ms │  1.23x slower │
│ QQuery 99 │  507.71 ms │                                   576.49 ms │  1.14x slower │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 67213.22ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 90351.33ms │
│ Average Time (HEAD)                                        │   678.92ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   912.64ms │
│ Queries Faster                                             │          2 │
│ Queries Slower                                             │         90 │
│ Queries with No Change                                     │          7 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     106.55 / 173.11 ±37.68 / 207.15 ms │          106.61 / 173.08 ±37.78 / 207.08 ms │     no change │
│ QQuery 2  │     511.94 / 649.73 ±72.98 / 712.88 ms │          764.90 / 883.97 ±80.07 / 995.73 ms │  1.36x slower │
│ QQuery 3  │     382.82 / 429.59 ±31.93 / 476.04 ms │         464.40 / 571.90 ±103.92 / 700.77 ms │  1.33x slower │
│ QQuery 4  │  1194.30 / 1311.58 ±93.73 / 1473.53 ms │      1585.64 / 1786.23 ±129.90 / 1978.05 ms │  1.36x slower │
│ QQuery 5  │     472.14 / 569.34 ±70.84 / 678.52 ms │          606.93 / 680.33 ±62.16 / 787.78 ms │  1.19x slower │
│ QQuery 6  │     610.73 / 716.14 ±85.88 / 814.56 ms │          764.11 / 885.31 ±88.72 / 993.62 ms │  1.24x slower │
│ QQuery 7  │     546.23 / 684.97 ±78.48 / 772.18 ms │          885.30 / 949.36 ±32.41 / 971.20 ms │  1.39x slower │
│ QQuery 8  │     483.89 / 573.01 ±59.22 / 646.69 ms │          759.48 / 791.31 ±36.30 / 836.26 ms │  1.38x slower │
│ QQuery 9  │  1374.60 / 1490.74 ±94.29 / 1597.07 ms │       1541.04 / 1641.53 ±80.09 / 1764.57 ms │  1.10x slower │
│ QQuery 10 │     755.03 / 808.04 ±45.05 / 877.62 ms │       1004.14 / 1118.16 ±67.53 / 1190.81 ms │  1.38x slower │
│ QQuery 11 │   872.33 / 1007.07 ±79.70 / 1114.67 ms │       1245.92 / 1296.74 ±57.57 / 1402.36 ms │  1.29x slower │
│ QQuery 12 │     344.38 / 373.35 ±29.51 / 425.91 ms │          379.58 / 476.04 ±66.82 / 541.81 ms │  1.28x slower │
│ QQuery 13 │     657.32 / 746.00 ±50.85 / 794.58 ms │         958.21 / 991.90 ±37.22 / 1063.63 ms │  1.33x slower │
│ QQuery 14 │  3929.87 / 4034.90 ±84.73 / 4161.41 ms │      4755.84 / 4921.46 ±100.96 / 5063.64 ms │  1.22x slower │
│ QQuery 15 │     408.01 / 513.73 ±60.04 / 583.01 ms │          583.39 / 649.71 ±41.33 / 701.20 ms │  1.26x slower │
│ QQuery 16 │     108.07 / 175.42 ±39.05 / 214.08 ms │          107.92 / 170.38 ±35.03 / 207.81 ms │     no change │
│ QQuery 17 │     631.93 / 711.81 ±44.79 / 759.11 ms │        995.17 / 1023.43 ±37.52 / 1097.40 ms │  1.44x slower │
│ QQuery 18 │     716.07 / 800.95 ±60.58 / 903.51 ms │       1013.09 / 1149.03 ±98.30 / 1263.08 ms │  1.43x slower │
│ QQuery 19 │     533.49 / 596.91 ±55.15 / 680.14 ms │          787.93 / 886.32 ±63.38 / 972.99 ms │  1.48x slower │
│ QQuery 20 │     326.84 / 400.19 ±40.13 / 446.80 ms │          464.15 / 512.17 ±49.91 / 572.91 ms │  1.28x slower │
│ QQuery 21 │     505.30 / 570.14 ±48.72 / 629.53 ms │          482.41 / 624.43 ±98.83 / 735.18 ms │  1.10x slower │
│ QQuery 22 │     476.28 / 529.07 ±31.73 / 570.38 ms │          525.24 / 620.11 ±61.66 / 697.54 ms │  1.17x slower │
│ QQuery 23 │ 2547.00 / 2713.35 ±111.15 / 2864.08 ms │      3533.07 / 3669.30 ±114.09 / 3840.70 ms │  1.35x slower │
│ QQuery 24 │ 2187.60 / 2354.21 ±109.67 / 2490.08 ms │       3118.91 / 3139.86 ±18.16 / 3167.27 ms │  1.33x slower │
│ QQuery 25 │     687.32 / 728.63 ±34.47 / 774.02 ms │        913.17 / 1017.29 ±80.95 / 1129.17 ms │  1.40x slower │
│ QQuery 26 │     599.35 / 646.96 ±36.05 / 703.77 ms │          704.95 / 739.51 ±36.55 / 807.82 ms │  1.14x slower │
│ QQuery 27 │     141.55 / 254.15 ±84.00 / 359.41 ms │          190.56 / 224.56 ±35.19 / 285.80 ms │ +1.13x faster │
│ QQuery 28 │     344.91 / 399.67 ±30.96 / 433.37 ms │          425.90 / 526.46 ±69.03 / 642.97 ms │  1.32x slower │
│ QQuery 29 │     720.41 / 787.79 ±48.80 / 839.92 ms │       1068.21 / 1161.04 ±57.29 / 1233.74 ms │  1.47x slower │
│ QQuery 30 │     560.96 / 650.38 ±64.90 / 746.67 ms │          857.56 / 895.55 ±40.91 / 961.13 ms │  1.38x slower │
│ QQuery 31 │  1193.11 / 1272.27 ±57.47 / 1320.37 ms │       2095.41 / 2105.73 ±12.80 / 2130.36 ms │  1.66x slower │
│ QQuery 32 │     414.92 / 450.32 ±31.49 / 485.97 ms │          482.52 / 583.21 ±59.51 / 647.23 ms │  1.30x slower │
│ QQuery 33 │     522.29 / 589.11 ±42.48 / 639.37 ms │          736.44 / 777.67 ±29.55 / 808.31 ms │  1.32x slower │
│ QQuery 34 │     194.91 / 306.28 ±59.56 / 359.41 ms │          228.07 / 292.32 ±61.04 / 374.03 ms │     no change │
│ QQuery 35 │     657.11 / 736.32 ±72.47 / 873.30 ms │       1012.73 / 1115.53 ±70.00 / 1217.86 ms │  1.52x slower │
│ QQuery 36 │     110.34 / 173.88 ±36.77 / 207.67 ms │          108.19 / 174.07 ±38.09 / 210.44 ms │     no change │
│ QQuery 37 │     220.33 / 276.22 ±30.86 / 311.29 ms │          237.11 / 321.02 ±50.97 / 376.53 ms │  1.16x slower │
│ QQuery 38 │     564.16 / 678.42 ±72.29 / 791.56 ms │          825.62 / 920.62 ±51.85 / 962.10 ms │  1.36x slower │
│ QQuery 39 │ 1845.58 / 2008.82 ±111.66 / 2141.17 ms │        2251.57 / 2260.78 ±5.73 / 2267.32 ms │  1.13x slower │
│ QQuery 40 │     466.61 / 494.02 ±22.26 / 513.74 ms │          579.31 / 651.01 ±68.21 / 766.54 ms │  1.32x slower │
│ QQuery 41 │     175.86 / 279.03 ±73.59 / 382.71 ms │          175.02 / 278.90 ±73.83 / 382.74 ms │     no change │
│ QQuery 42 │     313.33 / 405.92 ±65.77 / 468.14 ms │          423.92 / 509.05 ±65.25 / 608.73 ms │  1.25x slower │
│ QQuery 43 │     105.63 / 152.51 ±41.13 / 206.84 ms │           76.62 / 122.96 ±44.05 / 206.63 ms │ +1.24x faster │
│ QQuery 44 │     198.81 / 262.90 ±47.00 / 303.00 ms │          196.14 / 240.78 ±48.50 / 302.06 ms │ +1.09x faster │
│ QQuery 45 │     462.94 / 544.43 ±63.67 / 615.12 ms │          665.34 / 704.58 ±54.74 / 806.01 ms │  1.29x slower │
│ QQuery 46 │     289.58 / 345.73 ±41.84 / 399.58 ms │          525.40 / 582.70 ±47.77 / 668.51 ms │  1.69x slower │
│ QQuery 47 │  1187.06 / 1258.54 ±49.95 / 1323.23 ms │       1583.24 / 1733.12 ±86.93 / 1812.95 ms │  1.38x slower │
│ QQuery 48 │     599.59 / 703.15 ±59.81 / 765.90 ms │          859.87 / 934.92 ±37.75 / 961.49 ms │  1.33x slower │
│ QQuery 49 │     575.32 / 626.63 ±50.48 / 701.96 ms │          655.87 / 724.72 ±57.14 / 819.77 ms │  1.16x slower │
│ QQuery 50 │     528.05 / 635.67 ±57.09 / 696.74 ms │          783.09 / 856.69 ±61.18 / 947.96 ms │  1.35x slower │
│ QQuery 51 │     511.88 / 544.43 ±25.75 / 569.44 ms │          676.58 / 759.29 ±72.60 / 879.73 ms │  1.39x slower │
│ QQuery 52 │     307.71 / 406.47 ±68.78 / 469.10 ms │          442.29 / 498.69 ±43.16 / 558.68 ms │  1.23x slower │
│ QQuery 53 │     408.69 / 453.07 ±49.48 / 515.97 ms │          442.07 / 571.34 ±65.75 / 618.73 ms │  1.26x slower │
│ QQuery 54 │    857.62 / 965.61 ±70.10 / 1038.52 ms │       1122.26 / 1196.36 ±65.00 / 1302.39 ms │  1.24x slower │
│ QQuery 55 │     377.16 / 423.84 ±32.07 / 470.90 ms │          424.88 / 509.20 ±64.66 / 610.51 ms │  1.20x slower │
│ QQuery 56 │     547.20 / 597.36 ±30.78 / 637.22 ms │          745.89 / 826.42 ±64.51 / 938.19 ms │  1.38x slower │
│ QQuery 57 │ 1031.63 / 1169.95 ±108.42 / 1309.42 ms │       1460.85 / 1564.00 ±53.01 / 1605.88 ms │  1.34x slower │
│ QQuery 58 │  1155.61 / 1329.66 ±97.21 / 1424.27 ms │       1492.95 / 1561.45 ±43.71 / 1600.80 ms │  1.17x slower │
│ QQuery 59 │    897.47 / 928.81 ±44.06 / 1016.33 ms │        945.77 / 1044.52 ±64.21 / 1130.09 ms │  1.12x slower │
│ QQuery 60 │     574.44 / 624.69 ±58.11 / 732.08 ms │          698.88 / 763.26 ±60.27 / 872.88 ms │  1.22x slower │
│ QQuery 61 │      394.04 / 395.97 ±1.84 / 399.33 ms │          392.37 / 412.10 ±15.57 / 426.76 ms │     no change │
│ QQuery 62 │     581.64 / 602.74 ±24.98 / 634.62 ms │          540.77 / 650.29 ±82.16 / 787.52 ms │  1.08x slower │
│ QQuery 63 │     418.11 / 440.91 ±27.23 / 475.15 ms │          448.56 / 516.67 ±49.66 / 591.32 ms │  1.17x slower │
│ QQuery 64 │  3180.76 / 3294.58 ±82.59 / 3378.33 ms │       4659.33 / 4701.59 ±39.04 / 4770.47 ms │  1.43x slower │
│ QQuery 65 │     602.43 / 716.35 ±70.94 / 808.47 ms │        968.02 / 1035.74 ±58.60 / 1115.17 ms │  1.45x slower │
│ QQuery 66 │     564.29 / 595.90 ±41.49 / 675.77 ms │          683.49 / 754.68 ±47.72 / 831.03 ms │  1.27x slower │
│ QQuery 67 │     612.03 / 691.70 ±50.02 / 752.46 ms │          709.45 / 754.72 ±34.96 / 797.49 ms │  1.09x slower │
│ QQuery 68 │     288.29 / 307.91 ±15.55 / 327.67 ms │          486.75 / 524.82 ±36.03 / 591.31 ms │  1.70x slower │
│ QQuery 69 │     700.33 / 821.76 ±89.06 / 955.63 ms │       1115.17 / 1262.97 ±75.68 / 1318.50 ms │  1.54x slower │
│ QQuery 70 │     672.84 / 767.70 ±67.45 / 862.41 ms │          806.26 / 885.51 ±55.93 / 968.40 ms │  1.15x slower │
│ QQuery 71 │     475.08 / 561.93 ±65.09 / 667.33 ms │          675.36 / 758.60 ±50.86 / 817.38 ms │  1.35x slower │
│ QQuery 72 │ 3005.47 / 3170.38 ±131.63 / 3394.66 ms │       3588.68 / 3703.35 ±77.47 / 3811.19 ms │  1.17x slower │
│ QQuery 73 │     195.44 / 306.84 ±59.91 / 363.75 ms │          256.95 / 346.75 ±49.80 / 393.23 ms │  1.13x slower │
│ QQuery 74 │     724.70 / 809.69 ±62.40 / 917.67 ms │       1180.16 / 1249.32 ±52.85 / 1315.74 ms │  1.54x slower │
│ QQuery 75 │    931.22 / 989.15 ±40.30 / 1053.70 ms │      1242.33 / 1410.51 ±103.16 / 1539.73 ms │  1.43x slower │
│ QQuery 76 │     521.13 / 599.06 ±55.19 / 653.65 ms │          616.34 / 774.06 ±79.40 / 831.12 ms │  1.29x slower │
│ QQuery 77 │     603.99 / 705.83 ±67.45 / 801.72 ms │         872.06 / 932.93 ±58.55 / 1033.35 ms │  1.32x slower │
│ QQuery 78 │     706.59 / 867.89 ±88.62 / 975.49 ms │       1352.28 / 1379.42 ±17.33 / 1394.82 ms │  1.59x slower │
│ QQuery 79 │     448.82 / 508.35 ±52.10 / 599.51 ms │          569.64 / 689.70 ±71.35 / 767.74 ms │  1.36x slower │
│ QQuery 80 │     649.05 / 734.07 ±62.21 / 829.53 ms │         898.91 / 957.38 ±42.45 / 1031.47 ms │  1.30x slower │
│ QQuery 81 │     477.58 / 609.38 ±84.69 / 722.76 ms │          783.97 / 872.81 ±58.22 / 953.83 ms │  1.43x slower │
│ QQuery 82 │     555.56 / 645.97 ±48.96 / 687.83 ms │          745.39 / 892.41 ±91.88 / 968.78 ms │  1.38x slower │
│ QQuery 83 │    858.88 / 966.42 ±94.01 / 1097.73 ms │       1140.15 / 1229.98 ±72.87 / 1310.35 ms │  1.27x slower │
│ QQuery 84 │     440.73 / 503.17 ±57.69 / 593.02 ms │          735.43 / 786.71 ±44.30 / 859.92 ms │  1.56x slower │
│ QQuery 85 │     813.54 / 887.73 ±53.16 / 954.94 ms │      1098.30 / 1233.90 ±112.02 / 1353.33 ms │  1.39x slower │
│ QQuery 86 │     248.85 / 316.18 ±53.70 / 392.77 ms │          372.77 / 429.13 ±44.55 / 489.25 ms │  1.36x slower │
│ QQuery 87 │     575.62 / 648.43 ±72.57 / 781.59 ms │        911.78 / 1037.51 ±74.48 / 1134.77 ms │  1.60x slower │
│ QQuery 88 │      737.84 / 738.50 ±0.50 / 739.38 ms │          791.46 / 819.99 ±21.28 / 844.83 ms │  1.11x slower │
│ QQuery 89 │     424.96 / 459.58 ±27.81 / 482.49 ms │          513.10 / 548.20 ±34.30 / 609.79 ms │  1.19x slower │
│ QQuery 90 │     425.49 / 465.94 ±26.32 / 502.89 ms │          457.07 / 494.94 ±31.07 / 534.77 ms │  1.06x slower │
│ QQuery 91 │    526.83 / 665.12 ±112.13 / 806.62 ms │        1044.24 / 1048.02 ±2.16 / 1050.94 ms │  1.58x slower │
│ QQuery 92 │     437.92 / 452.51 ±19.81 / 490.58 ms │          508.26 / 587.48 ±42.55 / 636.33 ms │  1.30x slower │
│ QQuery 93 │     458.45 / 481.18 ±17.74 / 510.36 ms │          535.88 / 618.74 ±54.51 / 678.34 ms │  1.29x slower │
│ QQuery 94 │     481.71 / 530.80 ±43.37 / 585.32 ms │          728.26 / 829.06 ±59.60 / 906.99 ms │  1.56x slower │
│ QQuery 95 │     688.04 / 776.20 ±47.70 / 822.89 ms │       1173.07 / 1206.84 ±30.66 / 1246.77 ms │  1.55x slower │
│ QQuery 96 │     403.26 / 445.73 ±48.10 / 506.59 ms │          502.76 / 535.17 ±34.62 / 596.66 ms │  1.20x slower │
│ QQuery 97 │     440.38 / 523.64 ±70.46 / 609.33 ms │          626.83 / 652.20 ±23.27 / 686.72 ms │  1.25x slower │
│ QQuery 98 │     401.47 / 445.79 ±29.34 / 487.02 ms │          495.62 / 556.10 ±54.45 / 641.71 ms │  1.25x slower │
│ QQuery 99 │     507.71 / 618.51 ±64.15 / 693.85 ms │          576.49 / 727.51 ±75.90 / 774.40 ms │  1.18x slower │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 75114.46ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 98169.16ms │
│ Average Time (HEAD)                                        │   758.73ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   991.61ms │
│ Queries Faster                                             │          3 │
│ Queries Slower                                             │         90 │
│ Queries with No Change                                     │          6 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	380.1s
Peak memory	1015.2 MiB
Avg memory	554.3 MiB
CPU user	239.0s
CPU sys	12.0s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	495.1s
Peak memory	810.3 MiB
Avg memory	344.0 MiB
CPU user	242.6s
CPU sys	13.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:55:53Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpch
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │  520.10 ms │                                   298.12 ms │ +1.74x faster │
│ QQuery 2  │  871.00 ms │                                   908.61 ms │     no change │
│ QQuery 3  │  868.80 ms │                                   852.22 ms │     no change │
│ QQuery 4  │  780.04 ms │                                   579.10 ms │ +1.35x faster │
│ QQuery 5  │ 1085.49 ms │                                   956.36 ms │ +1.14x faster │
│ QQuery 6  │  570.05 ms │                                   247.94 ms │ +2.30x faster │
│ QQuery 7  │  993.07 ms │                                  1026.53 ms │     no change │
│ QQuery 8  │ 1170.74 ms │                                  1103.11 ms │ +1.06x faster │
│ QQuery 9  │ 1157.08 ms │                                   945.16 ms │ +1.22x faster │
│ QQuery 10 │  891.03 ms │                                   762.01 ms │ +1.17x faster │
│ QQuery 11 │  822.84 ms │                                   814.38 ms │     no change │
│ QQuery 12 │  738.00 ms │                                   556.76 ms │ +1.33x faster │
│ QQuery 13 │  404.16 ms │                                   505.05 ms │  1.25x slower │
│ QQuery 14 │  593.75 ms │                                   456.20 ms │ +1.30x faster │
│ QQuery 15 │ 1104.76 ms │                                   650.04 ms │ +1.70x faster │
│ QQuery 16 │  325.77 ms │                                   437.17 ms │  1.34x slower │
│ QQuery 17 │ 1077.48 ms │                                   744.61 ms │ +1.45x faster │
│ QQuery 18 │ 1381.10 ms │                                  1099.81 ms │ +1.26x faster │
│ QQuery 19 │  620.10 ms │                                   399.97 ms │ +1.55x faster │
│ QQuery 20 │  883.39 ms │                                   747.06 ms │ +1.18x faster │
│ QQuery 21 │ 2064.56 ms │                                  1292.13 ms │ +1.60x faster │
│ QQuery 22 │  497.00 ms │                                   756.06 ms │  1.52x slower │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 19420.31ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 16138.41ms │
│ Average Time (HEAD)                                        │   882.74ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   733.56ms │
│ Queries Faster                                             │         15 │
│ Queries Slower                                             │          3 │
│ Queries with No Change                                     │          4 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     520.10 / 574.28 ±43.81 / 644.74 ms │          298.12 / 352.53 ±37.59 / 392.36 ms │ +1.63x faster │
│ QQuery 2  │     871.00 / 892.49 ±21.53 / 924.41 ms │        908.61 / 1056.66 ±88.01 / 1151.67 ms │  1.18x slower │
│ QQuery 3  │   868.80 / 1018.72 ±96.26 / 1134.18 ms │          852.22 / 909.66 ±47.23 / 982.65 ms │ +1.12x faster │
│ QQuery 4  │     780.04 / 844.55 ±64.12 / 945.63 ms │          579.10 / 636.42 ±58.28 / 721.32 ms │ +1.33x faster │
│ QQuery 5  │ 1085.49 / 1221.83 ±143.98 / 1458.52 ms │        956.36 / 1055.57 ±67.99 / 1141.81 ms │ +1.16x faster │
│ QQuery 6  │     570.05 / 625.57 ±55.65 / 703.05 ms │          247.94 / 295.01 ±39.66 / 348.84 ms │ +2.12x faster │
│ QQuery 7  │  993.07 / 1165.86 ±119.00 / 1288.10 ms │       1026.53 / 1098.95 ±54.41 / 1167.87 ms │ +1.06x faster │
│ QQuery 8  │ 1170.74 / 1306.16 ±101.24 / 1435.60 ms │       1103.11 / 1165.53 ±55.97 / 1241.57 ms │ +1.12x faster │
│ QQuery 9  │ 1157.08 / 1295.33 ±109.58 / 1427.43 ms │        945.16 / 1053.35 ±82.28 / 1153.90 ms │ +1.23x faster │
│ QQuery 10 │    891.03 / 953.25 ±48.07 / 1006.44 ms │         762.01 / 847.27 ±98.77 / 1020.26 ms │ +1.13x faster │
│ QQuery 11 │     822.84 / 852.63 ±27.97 / 897.75 ms │          814.38 / 871.88 ±50.90 / 936.83 ms │     no change │
│ QQuery 12 │     738.00 / 837.94 ±70.97 / 941.19 ms │          556.76 / 591.67 ±31.28 / 644.05 ms │ +1.42x faster │
│ QQuery 13 │     404.16 / 438.16 ±28.73 / 484.98 ms │          505.05 / 530.34 ±26.43 / 580.95 ms │  1.21x slower │
│ QQuery 14 │     593.75 / 677.83 ±66.25 / 779.80 ms │          456.20 / 505.61 ±41.08 / 565.19 ms │ +1.34x faster │
│ QQuery 15 │ 1104.76 / 1245.41 ±100.60 / 1380.78 ms │          650.04 / 719.14 ±71.32 / 845.41 ms │ +1.73x faster │
│ QQuery 16 │     325.77 / 431.50 ±54.97 / 478.35 ms │          437.17 / 515.55 ±64.17 / 620.22 ms │  1.19x slower │
│ QQuery 17 │ 1077.48 / 1283.44 ±109.00 / 1372.90 ms │          744.61 / 800.66 ±58.35 / 893.11 ms │ +1.60x faster │
│ QQuery 18 │  1381.10 / 1515.65 ±91.28 / 1652.57 ms │        1099.81 / 1102.81 ±2.81 / 1107.76 ms │ +1.37x faster │
│ QQuery 19 │     620.10 / 693.80 ±73.76 / 809.61 ms │          399.97 / 488.27 ±53.75 / 567.49 ms │ +1.42x faster │
│ QQuery 20 │   883.39 / 1042.47 ±81.25 / 1106.50 ms │          747.06 / 817.01 ±54.19 / 901.05 ms │ +1.28x faster │
│ QQuery 21 │ 2064.56 / 2226.49 ±113.78 / 2344.51 ms │       1292.13 / 1358.04 ±46.47 / 1414.66 ms │ +1.64x faster │
│ QQuery 22 │     497.00 / 532.46 ±28.81 / 578.60 ms │           756.06 / 757.97 ±1.18 / 759.42 ms │  1.42x slower │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 21675.83ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 17529.88ms │
│ Average Time (HEAD)                                        │   985.27ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   796.81ms │
│ Queries Faster                                             │         17 │
│ Queries Slower                                             │          4 │
│ Queries with No Change                                     │          1 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	115.0s
Peak memory	834.6 MiB
Avg memory	396.1 MiB
CPU user	30.4s
CPU sys	4.2s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	90.0s
Peak memory	781.5 MiB
Avg memory	398.5 MiB
CPU user	30.9s
CPU sys	3.8s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:57:31Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "4G"
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │  534.12 ms │                                   298.61 ms │ +1.79x faster │
│ QQuery 2  │  813.48 ms │                                   951.53 ms │  1.17x slower │
│ QQuery 3  │  985.52 ms │                                   855.26 ms │ +1.15x faster │
│ QQuery 4  │  707.90 ms │                                   579.89 ms │ +1.22x faster │
│ QQuery 5  │ 1148.40 ms │                                   957.40 ms │ +1.20x faster │
│ QQuery 6  │  533.40 ms │                                   246.16 ms │ +2.17x faster │
│ QQuery 7  │ 1000.29 ms │                                  1024.75 ms │     no change │
│ QQuery 8  │ 1117.15 ms │                                  1099.31 ms │     no change │
│ QQuery 9  │ 1200.50 ms │                                   934.69 ms │ +1.28x faster │
│ QQuery 10 │  932.86 ms │                                   757.91 ms │ +1.23x faster │
│ QQuery 11 │  742.60 ms │                                   813.57 ms │  1.10x slower │
│ QQuery 12 │  727.03 ms │                                   554.46 ms │ +1.31x faster │
│ QQuery 13 │  401.31 ms │                                   502.65 ms │  1.25x slower │
│ QQuery 14 │  624.70 ms │                                   453.51 ms │ +1.38x faster │
│ QQuery 15 │ 1188.19 ms │                                   650.69 ms │ +1.83x faster │
│ QQuery 16 │  325.14 ms │                                   440.47 ms │  1.35x slower │
│ QQuery 17 │ 1131.37 ms │                                   740.73 ms │ +1.53x faster │
│ QQuery 18 │ 1372.30 ms │                                  1098.58 ms │ +1.25x faster │
│ QQuery 19 │  543.64 ms │                                   402.56 ms │ +1.35x faster │
│ QQuery 20 │ 1050.12 ms │                                   744.68 ms │ +1.41x faster │
│ QQuery 21 │ 2061.98 ms │                                  1285.63 ms │ +1.60x faster │
│ QQuery 22 │  497.37 ms │                                   755.17 ms │  1.52x slower │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 19639.36ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 16148.21ms │
│ Average Time (HEAD)                                        │   892.70ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   734.01ms │
│ Queries Faster                                             │         15 │
│ Queries Slower                                             │          5 │
│ Queries with No Change                                     │          2 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     534.12 / 567.20 ±28.72 / 605.57 ms │          298.61 / 354.17 ±39.12 / 399.83 ms │ +1.60x faster │
│ QQuery 2  │     813.48 / 900.44 ±51.57 / 954.71 ms │        951.53 / 1059.50 ±86.57 / 1151.62 ms │  1.18x slower │
│ QQuery 3  │   985.52 / 1022.65 ±28.37 / 1062.90 ms │          855.26 / 911.01 ±46.22 / 980.68 ms │ +1.12x faster │
│ QQuery 4  │     707.90 / 788.82 ±80.46 / 889.80 ms │          579.89 / 636.49 ±57.96 / 721.55 ms │ +1.24x faster │
│ QQuery 5  │  1148.40 / 1265.64 ±84.47 / 1388.91 ms │        957.40 / 1054.37 ±67.04 / 1138.12 ms │ +1.20x faster │
│ QQuery 6  │     533.40 / 622.59 ±74.58 / 759.77 ms │          246.16 / 293.63 ±38.70 / 345.52 ms │ +2.12x faster │
│ QQuery 7  │ 1000.29 / 1154.81 ±119.39 / 1297.81 ms │       1024.75 / 1096.82 ±52.68 / 1168.88 ms │ +1.05x faster │
│ QQuery 8  │ 1117.15 / 1296.68 ±124.31 / 1403.26 ms │       1099.31 / 1160.93 ±57.18 / 1239.55 ms │ +1.12x faster │
│ QQuery 9  │  1200.50 / 1311.45 ±90.12 / 1461.08 ms │       934.69 / 1065.95 ±100.26 / 1179.74 ms │ +1.23x faster │
│ QQuery 10 │   932.86 / 1020.91 ±65.83 / 1111.16 ms │        757.91 / 844.55 ±102.19 / 1025.35 ms │ +1.21x faster │
│ QQuery 11 │     742.60 / 825.85 ±84.75 / 974.20 ms │          813.57 / 870.91 ±52.07 / 940.90 ms │  1.05x slower │
│ QQuery 12 │     727.03 / 807.82 ±67.01 / 887.04 ms │          554.46 / 588.54 ±32.23 / 642.38 ms │ +1.37x faster │
│ QQuery 13 │     401.31 / 440.43 ±23.39 / 465.00 ms │          502.65 / 526.66 ±27.97 / 577.12 ms │  1.20x slower │
│ QQuery 14 │     624.70 / 705.52 ±51.83 / 762.65 ms │          453.51 / 503.03 ±40.46 / 564.25 ms │ +1.40x faster │
│ QQuery 15 │  1188.19 / 1272.73 ±51.08 / 1334.89 ms │          650.69 / 717.86 ±71.39 / 846.01 ms │ +1.77x faster │
│ QQuery 16 │     325.14 / 419.68 ±72.82 / 550.04 ms │          440.47 / 517.08 ±64.04 / 621.81 ms │  1.23x slower │
│ QQuery 17 │ 1131.37 / 1307.73 ±102.98 / 1439.47 ms │          740.73 / 795.30 ±54.73 / 887.48 ms │ +1.64x faster │
│ QQuery 18 │  1372.30 / 1496.32 ±98.09 / 1630.95 ms │        1098.58 / 1106.98 ±7.02 / 1119.59 ms │ +1.35x faster │
│ QQuery 19 │     543.64 / 663.44 ±65.52 / 738.94 ms │          402.56 / 488.91 ±55.34 / 574.50 ms │ +1.36x faster │
│ QQuery 20 │  1050.12 / 1081.28 ±34.44 / 1146.65 ms │          744.68 / 814.44 ±53.86 / 891.15 ms │ +1.33x faster │
│ QQuery 21 │  2061.98 / 2127.53 ±69.70 / 2259.39 ms │       1285.63 / 1354.45 ±49.02 / 1417.05 ms │ +1.57x faster │
│ QQuery 22 │     497.37 / 546.61 ±31.00 / 579.40 ms │           755.17 / 755.93 ±0.76 / 757.14 ms │  1.38x slower │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 21646.11ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 17517.50ms │
│ Average Time (HEAD)                                        │   983.91ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   796.25ms │
│ Queries Faster                                             │         17 │
│ Queries Slower                                             │          5 │
│ Queries with No Change                                     │          0 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only when the benchmark runs with DATAFUSION_RUNTIME_MEMORY_LIMIT set.

Base: 39d5064 (merge-base) | Changed: claude/datafusion-rowgroup-buffering-ec9abc

tpch — tpch_sf1

Query	Base	Changed	Change
Query 1	40.0 MiB	40.0 MiB	+0.0%
Query 2	29.5 MiB	29.7 MiB	+0.6%
Query 3	10.6 MiB	10.6 MiB	+0.0%
Query 4	33.3 MiB	33.3 MiB	-0.0%
Query 5	31.1 MiB	40.6 MiB	+30.6%
Query 6	0 B	0 B	0.0%
Query 7	87.2 MiB	90.0 MiB	+3.3%
Query 8	33.0 MiB	33.4 MiB	+1.1%
Query 9	130.7 MiB	131.3 MiB	+0.5%
Query 10	43.2 MiB	41.5 MiB	-4.0%
Query 11	31.5 MiB	22.1 MiB	-29.9%
Query 12	20.0 MiB	20.0 MiB	+0.0%
Query 13	70.1 MiB	70.1 MiB	-0.0%
Query 14	7.5 MiB	7.5 MiB	+0.0%
Query 15	13.0 MiB	12.9 MiB	-0.6%
Query 16	55.4 MiB	54.8 MiB	-0.9%
Query 17	147.6 MiB	69.3 MiB	-53.1%
Query 18	306.8 MiB	306.8 MiB	+0.0%
Query 19	721.1 KiB	721.1 KiB	+0.0%
Query 20	86.6 MiB	82.5 MiB	-4.7%
Query 21	46.8 MiB	46.8 MiB	+0.0%
Query 22	42.8 MiB	32.2 MiB	-24.8%

Pool accounting vs. process RSS

Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpch	base (`39d5064 (merge-base)`)	306.8 MiB	871.2 MiB	564.4 MiB	2.8×
tpch	changed (`claude/datafusion-rowgroup-buffering-ec9abc`)	306.8 MiB	782.7 MiB	475.9 MiB	2.6×

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	115.0s
Peak memory	871.2 MiB
Avg memory	407.9 MiB
CPU user	30.1s
CPU sys	3.9s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	90.0s
Peak memory	782.7 MiB
Avg memory	395.7 MiB
CPU user	29.8s
CPU sys	3.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:57:42Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.22 ms │                                     1.22 ms │     no change │
│ QQuery 1  │ 1866.15 ms │                                  1840.76 ms │     no change │
│ QQuery 2  │ 2151.50 ms │                                  2188.28 ms │     no change │
│ QQuery 3  │ 2176.97 ms │                                  2141.93 ms │     no change │
│ QQuery 4  │ 2317.57 ms │                                  2328.27 ms │     no change │
│ QQuery 5  │ 2352.26 ms │                                  2360.58 ms │     no change │
│ QQuery 6  │    1.26 ms │                                     1.27 ms │     no change │
│ QQuery 7  │ 1863.22 ms │                                  1970.96 ms │  1.06x slower │
│ QQuery 8  │ 2412.15 ms │                                  2403.23 ms │     no change │
│ QQuery 9  │ 2644.21 ms │                                  2610.42 ms │     no change │
│ QQuery 10 │ 2217.42 ms │                                  2201.51 ms │     no change │
│ QQuery 11 │ 2256.02 ms │                                  2236.91 ms │     no change │
│ QQuery 12 │ 2373.87 ms │                                  2372.98 ms │     no change │
│ QQuery 13 │ 2436.03 ms │                                  2450.08 ms │     no change │
│ QQuery 14 │ 2378.96 ms │                                  2359.17 ms │     no change │
│ QQuery 15 │ 2420.58 ms │                                  2331.05 ms │     no change │
│ QQuery 16 │ 2640.09 ms │                                  2638.22 ms │     no change │
│ QQuery 17 │ 2637.29 ms │                                  2621.47 ms │     no change │
│ QQuery 18 │ 3106.26 ms │                                  3154.47 ms │     no change │
│ QQuery 19 │ 1749.74 ms │                                  1713.74 ms │     no change │
│ QQuery 20 │ 2616.88 ms │                                  2630.15 ms │     no change │
│ QQuery 21 │ 2645.35 ms │                                  2678.52 ms │     no change │
│ QQuery 22 │ 3099.68 ms │                                  3076.14 ms │     no change │
│ QQuery 23 │ 4814.03 ms │                                  4912.75 ms │     no change │
│ QQuery 24 │  457.80 ms │                                   462.93 ms │     no change │
│ QQuery 25 │ 2273.81 ms │                                  2299.50 ms │     no change │
│ QQuery 26 │  455.77 ms │                                   477.00 ms │     no change │
│ QQuery 27 │ 2618.14 ms │                                  2623.13 ms │     no change │
│ QQuery 28 │ 4960.00 ms │                                  4969.36 ms │     no change │
│ QQuery 29 │ 2205.92 ms │                                  2225.96 ms │     no change │
│ QQuery 30 │ 2420.84 ms │                                  2426.61 ms │     no change │
│ QQuery 31 │ 2343.24 ms │                                  2401.83 ms │     no change │
│ QQuery 32 │ 2515.68 ms │                                  2583.14 ms │     no change │
│ QQuery 33 │ 3401.46 ms │                                  3434.92 ms │     no change │
│ QQuery 34 │ 3420.54 ms │                                  3453.66 ms │     no change │
│ QQuery 35 │ 2403.64 ms │                                  2421.84 ms │     no change │
│ QQuery 36 │  187.56 ms │                                   135.19 ms │ +1.39x faster │
│ QQuery 37 │  140.43 ms │                                   137.16 ms │     no change │
│ QQuery 38 │  141.57 ms │                                   129.13 ms │ +1.10x faster │
│ QQuery 39 │  225.42 ms │                                   192.01 ms │ +1.17x faster │
│ QQuery 40 │  120.99 ms │                                   115.97 ms │     no change │
│ QQuery 41 │  121.89 ms │                                   114.62 ms │ +1.06x faster │
│ QQuery 42 │  115.40 ms │                                   114.46 ms │     no change │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 85708.82ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 85942.53ms │
│ Average Time (HEAD)                                        │  1993.23ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  1998.66ms │
│ Queries Faster                                             │          4 │
│ Queries Slower                                             │          1 │
│ Queries with No Change                                     │         38 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.22 / 4.02 ±5.50 / 15.02 ms │                1.22 / 3.95 ±5.37 / 14.68 ms │    no change │
│ QQuery 1  │  1866.15 / 1915.02 ±32.00 / 1958.88 ms │       1840.76 / 1910.60 ±55.91 / 2005.91 ms │    no change │
│ QQuery 2  │  2151.50 / 2224.45 ±64.55 / 2340.18 ms │       2188.28 / 2256.81 ±52.77 / 2327.98 ms │    no change │
│ QQuery 3  │  2176.97 / 2238.37 ±38.83 / 2290.38 ms │       2141.93 / 2208.19 ±43.29 / 2262.28 ms │    no change │
│ QQuery 4  │  2317.57 / 2372.82 ±46.32 / 2426.66 ms │       2328.27 / 2380.17 ±34.50 / 2417.94 ms │    no change │
│ QQuery 5  │  2352.26 / 2414.67 ±48.26 / 2463.76 ms │       2360.58 / 2410.05 ±34.33 / 2460.33 ms │    no change │
│ QQuery 6  │            1.26 / 1.40 ±0.22 / 1.84 ms │                 1.27 / 1.41 ±0.22 / 1.84 ms │    no change │
│ QQuery 7  │  1863.22 / 1985.05 ±80.26 / 2105.91 ms │       1970.96 / 2048.75 ±52.59 / 2132.13 ms │    no change │
│ QQuery 8  │  2412.15 / 2478.43 ±47.99 / 2547.60 ms │       2403.23 / 2468.05 ±38.05 / 2511.52 ms │    no change │
│ QQuery 9  │  2644.21 / 2729.10 ±71.56 / 2834.71 ms │       2610.42 / 2685.29 ±59.33 / 2769.44 ms │    no change │
│ QQuery 10 │  2217.42 / 2255.72 ±35.72 / 2316.73 ms │       2201.51 / 2254.28 ±57.77 / 2366.34 ms │    no change │
│ QQuery 11 │  2256.02 / 2296.37 ±32.52 / 2343.87 ms │       2236.91 / 2293.16 ±44.79 / 2346.49 ms │    no change │
│ QQuery 12 │  2373.87 / 2416.94 ±28.03 / 2452.80 ms │       2372.98 / 2432.70 ±50.41 / 2506.63 ms │    no change │
│ QQuery 13 │  2436.03 / 2507.09 ±59.13 / 2613.85 ms │       2450.08 / 2488.02 ±31.65 / 2539.82 ms │    no change │
│ QQuery 14 │  2378.96 / 2409.24 ±38.08 / 2481.66 ms │       2359.17 / 2464.61 ±59.18 / 2532.60 ms │    no change │
│ QQuery 15 │  2420.58 / 2452.58 ±25.77 / 2483.40 ms │       2331.05 / 2426.11 ±71.60 / 2550.63 ms │    no change │
│ QQuery 16 │  2640.09 / 2689.97 ±61.51 / 2811.21 ms │       2638.22 / 2701.71 ±47.86 / 2776.92 ms │    no change │
│ QQuery 17 │  2637.29 / 2674.44 ±21.76 / 2704.11 ms │       2621.47 / 2706.56 ±59.69 / 2795.57 ms │    no change │
│ QQuery 18 │  3106.26 / 3188.48 ±73.85 / 3292.42 ms │       3154.47 / 3186.95 ±19.51 / 3214.96 ms │    no change │
│ QQuery 19 │  1749.74 / 1824.87 ±78.75 / 1962.30 ms │       1713.74 / 1806.84 ±70.80 / 1932.59 ms │    no change │
│ QQuery 20 │  2616.88 / 2660.98 ±43.80 / 2742.30 ms │       2630.15 / 2664.27 ±44.66 / 2751.38 ms │    no change │
│ QQuery 21 │  2645.35 / 2720.80 ±57.09 / 2795.74 ms │       2678.52 / 2709.29 ±23.44 / 2750.31 ms │    no change │
│ QQuery 22 │  3099.68 / 3160.10 ±60.93 / 3260.10 ms │       3076.14 / 3131.96 ±72.56 / 3267.29 ms │    no change │
│ QQuery 23 │ 4814.03 / 4970.30 ±120.29 / 5090.81 ms │       4912.75 / 4995.21 ±74.08 / 5133.32 ms │    no change │
│ QQuery 24 │     457.80 / 529.41 ±68.94 / 642.13 ms │          462.93 / 542.82 ±57.00 / 612.19 ms │    no change │
│ QQuery 25 │  2273.81 / 2351.86 ±64.22 / 2450.79 ms │       2299.50 / 2350.22 ±46.35 / 2421.47 ms │    no change │
│ QQuery 26 │     455.77 / 477.07 ±15.04 / 499.75 ms │          477.00 / 500.89 ±18.90 / 531.43 ms │    no change │
│ QQuery 27 │  2618.14 / 2689.24 ±51.59 / 2757.99 ms │       2623.13 / 2698.44 ±47.36 / 2768.35 ms │    no change │
│ QQuery 28 │  4960.00 / 5056.86 ±77.29 / 5154.48 ms │       4969.36 / 5039.65 ±43.87 / 5101.17 ms │    no change │
│ QQuery 29 │  2205.92 / 2262.13 ±49.66 / 2331.34 ms │       2225.96 / 2293.16 ±61.44 / 2405.39 ms │    no change │
│ QQuery 30 │  2420.84 / 2491.52 ±66.63 / 2617.01 ms │       2426.61 / 2490.17 ±56.75 / 2584.16 ms │    no change │
│ QQuery 31 │  2343.24 / 2419.58 ±54.35 / 2491.98 ms │       2401.83 / 2454.63 ±39.50 / 2520.82 ms │    no change │
│ QQuery 32 │  2515.68 / 2552.25 ±32.78 / 2604.65 ms │       2583.14 / 2647.08 ±62.02 / 2736.51 ms │    no change │
│ QQuery 33 │  3401.46 / 3457.22 ±52.42 / 3523.36 ms │       3434.92 / 3502.62 ±65.10 / 3620.53 ms │    no change │
│ QQuery 34 │  3420.54 / 3461.20 ±32.72 / 3500.65 ms │       3453.66 / 3500.72 ±47.15 / 3589.64 ms │    no change │
│ QQuery 35 │  2403.64 / 2434.12 ±23.44 / 2474.63 ms │       2421.84 / 2488.41 ±36.26 / 2530.67 ms │    no change │
│ QQuery 36 │     187.56 / 219.63 ±41.96 / 300.69 ms │          135.19 / 220.16 ±45.63 / 260.01 ms │    no change │
│ QQuery 37 │     140.43 / 188.72 ±41.05 / 249.06 ms │          137.16 / 194.38 ±37.92 / 246.71 ms │    no change │
│ QQuery 38 │     141.57 / 193.00 ±34.11 / 244.89 ms │          129.13 / 189.05 ±36.60 / 227.93 ms │    no change │
│ QQuery 39 │     225.42 / 271.04 ±53.27 / 370.74 ms │          192.01 / 286.11 ±51.81 / 350.28 ms │ 1.06x slower │
│ QQuery 40 │     120.99 / 173.78 ±38.48 / 219.30 ms │          115.97 / 175.67 ±45.71 / 250.15 ms │    no change │
│ QQuery 41 │     121.89 / 168.89 ±33.97 / 215.55 ms │          114.62 / 167.98 ±35.54 / 214.15 ms │    no change │
│ QQuery 42 │     115.40 / 168.22 ±36.57 / 216.37 ms │          114.46 / 168.23 ±36.70 / 218.52 ms │    no change │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 88156.96ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 88545.31ms │
│ Average Time (HEAD)                                        │  2050.16ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  2059.19ms │
│ Queries Faster                                             │          0 │
│ Queries Slower                                             │          1 │
│ Queries with No Change                                     │         42 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	445.1s
Peak memory	9.0 GiB
Avg memory	1.8 GiB
CPU user	981.5s
CPU sys	112.2s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	445.1s
Peak memory	9.6 GiB
Avg memory	1.8 GiB
CPU user	983.5s
CPU sys	111.0s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:58:51Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180420768-1386-hfk5m 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpch
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "pipelined"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T14:58:52Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180420768-1385-hr5v7 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "pipelined"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangb · 2026-08-04T15:05:31Z

run benchmarks

env:
    DF_FETCH_POLICY: "streaming"
    DF_FETCH_BUDGET: "104857600"
    SIMULATE_LATENCY: "true"

adriangbot · 2026-08-04T15:08:15Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "pipelined"
  SIMULATE_LATENCY: "true"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.26 ms │                                     1.26 ms │     no change │
│ QQuery 1  │ 1873.75 ms │                                  1342.26 ms │ +1.40x faster │
│ QQuery 2  │ 2135.07 ms │                                  1463.92 ms │ +1.46x faster │
│ QQuery 3  │ 2137.02 ms │                                  1444.51 ms │ +1.48x faster │
│ QQuery 4  │ 2301.92 ms │                                  1570.71 ms │ +1.47x faster │
│ QQuery 5  │ 2312.88 ms │                                  1663.98 ms │ +1.39x faster │
│ QQuery 6  │    1.28 ms │                                     1.28 ms │     no change │
│ QQuery 7  │ 1929.51 ms │                                  1354.42 ms │ +1.42x faster │
│ QQuery 8  │ 2396.28 ms │                                  1686.52 ms │ +1.42x faster │
│ QQuery 9  │ 2596.60 ms │                                  1902.39 ms │ +1.36x faster │
│ QQuery 10 │ 2225.19 ms │                                  1467.72 ms │ +1.52x faster │
│ QQuery 11 │ 2203.65 ms │                                  1477.61 ms │ +1.49x faster │
│ QQuery 12 │ 2338.89 ms │                                  1627.40 ms │ +1.44x faster │
│ QQuery 13 │ 2453.92 ms │                                  1707.80 ms │ +1.44x faster │
│ QQuery 14 │ 2402.82 ms │                                  1686.36 ms │ +1.42x faster │
│ QQuery 15 │ 2386.98 ms │                                  1638.43 ms │ +1.46x faster │
│ QQuery 16 │ 2624.04 ms │                                  1924.58 ms │ +1.36x faster │
│ QQuery 17 │ 2636.77 ms │                                  1927.57 ms │ +1.37x faster │
│ QQuery 18 │ 3102.90 ms │                                  2505.57 ms │ +1.24x faster │
│ QQuery 19 │ 1720.83 ms │                                  1330.13 ms │ +1.29x faster │
│ QQuery 20 │ 2597.30 ms │                                  1917.86 ms │ +1.35x faster │
│ QQuery 21 │ 2632.25 ms │                                  1943.14 ms │ +1.35x faster │
│ QQuery 22 │ 3062.65 ms │                                  2373.14 ms │ +1.29x faster │
│ QQuery 23 │ 4826.15 ms │                                  4603.16 ms │     no change │
│ QQuery 24 │  485.82 ms │                                   397.52 ms │ +1.22x faster │
│ QQuery 25 │ 2283.33 ms │                                  1508.10 ms │ +1.51x faster │
│ QQuery 26 │  496.24 ms │                                   451.51 ms │ +1.10x faster │
│ QQuery 27 │ 2655.00 ms │                                  1954.40 ms │ +1.36x faster │
│ QQuery 28 │ 4896.70 ms │                                  4137.15 ms │ +1.18x faster │
│ QQuery 29 │ 2212.89 ms │                                  1501.21 ms │ +1.47x faster │
│ QQuery 30 │ 2459.04 ms │                                  1693.19 ms │ +1.45x faster │
│ QQuery 31 │ 2388.63 ms │                                  1654.28 ms │ +1.44x faster │
│ QQuery 32 │ 2505.63 ms │                                  1743.25 ms │ +1.44x faster │
│ QQuery 33 │ 3407.68 ms │                                  2729.64 ms │ +1.25x faster │
│ QQuery 34 │ 3457.93 ms │                                  2808.28 ms │ +1.23x faster │
│ QQuery 35 │ 2336.40 ms │                                  1663.16 ms │ +1.40x faster │
│ QQuery 36 │  137.82 ms │                                   139.33 ms │     no change │
│ QQuery 37 │  143.48 ms │                                   134.71 ms │ +1.07x faster │
│ QQuery 38 │  127.39 ms │                                   129.43 ms │     no change │
│ QQuery 39 │  192.85 ms │                                   191.99 ms │     no change │
│ QQuery 40 │  115.12 ms │                                   118.43 ms │     no change │
│ QQuery 41 │  115.09 ms │                                   118.08 ms │     no change │
│ QQuery 42 │  114.30 ms │                                   115.47 ms │     no change │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 85431.25ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 63750.81ms │
│ Average Time (HEAD)                                        │  1986.77ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  1482.58ms │
│ Queries Faster                                             │         34 │
│ Queries Slower                                             │          0 │
│ Queries with No Change                                     │          9 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.26 / 4.03 ±5.48 / 14.99 ms │                1.26 / 4.05 ±5.47 / 15.00 ms │     no change │
│ QQuery 1  │  1873.75 / 1901.21 ±23.96 / 1937.08 ms │       1342.26 / 1393.21 ±53.61 / 1485.64 ms │ +1.36x faster │
│ QQuery 2  │  2135.07 / 2202.91 ±59.05 / 2304.87 ms │       1463.92 / 1521.25 ±48.60 / 1599.03 ms │ +1.45x faster │
│ QQuery 3  │  2137.02 / 2214.71 ±43.39 / 2270.95 ms │       1444.51 / 1499.02 ±58.58 / 1596.01 ms │ +1.48x faster │
│ QQuery 4  │  2301.92 / 2336.68 ±24.89 / 2377.82 ms │       1570.71 / 1634.38 ±63.30 / 1755.43 ms │ +1.43x faster │
│ QQuery 5  │  2312.88 / 2369.77 ±30.87 / 2401.11 ms │       1663.98 / 1703.23 ±32.19 / 1741.47 ms │ +1.39x faster │
│ QQuery 6  │            1.28 / 1.42 ±0.22 / 1.86 ms │                 1.28 / 1.43 ±0.22 / 1.86 ms │     no change │
│ QQuery 7  │  1929.51 / 1985.63 ±49.99 / 2054.87 ms │       1354.42 / 1385.66 ±31.83 / 1444.43 ms │ +1.43x faster │
│ QQuery 8  │  2396.28 / 2451.95 ±29.60 / 2475.70 ms │       1686.52 / 1736.52 ±57.92 / 1841.88 ms │ +1.41x faster │
│ QQuery 9  │  2596.60 / 2650.84 ±33.03 / 2681.42 ms │       1902.39 / 2002.98 ±70.95 / 2085.48 ms │ +1.32x faster │
│ QQuery 10 │  2225.19 / 2246.99 ±13.67 / 2264.00 ms │       1467.72 / 1542.41 ±57.35 / 1638.14 ms │ +1.46x faster │
│ QQuery 11 │  2203.65 / 2281.91 ±65.80 / 2385.13 ms │       1477.61 / 1550.36 ±59.62 / 1638.22 ms │ +1.47x faster │
│ QQuery 12 │  2338.89 / 2388.41 ±40.92 / 2455.58 ms │       1627.40 / 1711.62 ±66.13 / 1818.48 ms │ +1.40x faster │
│ QQuery 13 │  2453.92 / 2511.30 ±59.73 / 2596.87 ms │       1707.80 / 1752.60 ±41.63 / 1819.39 ms │ +1.43x faster │
│ QQuery 14 │  2402.82 / 2457.07 ±46.73 / 2527.91 ms │       1686.36 / 1739.33 ±57.55 / 1816.85 ms │ +1.41x faster │
│ QQuery 15 │  2386.98 / 2434.68 ±47.83 / 2523.35 ms │       1638.43 / 1677.71 ±34.40 / 1737.63 ms │ +1.45x faster │
│ QQuery 16 │  2624.04 / 2671.94 ±33.23 / 2706.84 ms │       1924.58 / 1969.65 ±35.86 / 2033.68 ms │ +1.36x faster │
│ QQuery 17 │  2636.77 / 2693.68 ±43.30 / 2762.32 ms │       1927.57 / 1973.27 ±31.24 / 2016.66 ms │ +1.37x faster │
│ QQuery 18 │  3102.90 / 3161.42 ±38.70 / 3207.75 ms │       2505.57 / 2524.99 ±16.58 / 2555.08 ms │ +1.25x faster │
│ QQuery 19 │  1720.83 / 1827.11 ±72.06 / 1916.17 ms │       1330.13 / 1374.93 ±40.82 / 1446.11 ms │ +1.33x faster │
│ QQuery 20 │  2597.30 / 2682.75 ±54.35 / 2763.13 ms │       1917.86 / 1982.14 ±57.45 / 2089.60 ms │ +1.35x faster │
│ QQuery 21 │  2632.25 / 2700.32 ±34.85 / 2726.35 ms │       1943.14 / 2018.77 ±55.66 / 2084.37 ms │ +1.34x faster │
│ QQuery 22 │  3062.65 / 3162.16 ±63.24 / 3259.96 ms │       2373.14 / 2389.86 ±12.01 / 2403.79 ms │ +1.32x faster │
│ QQuery 23 │  4826.15 / 4959.76 ±75.70 / 5039.15 ms │       4603.16 / 4711.08 ±75.86 / 4829.91 ms │ +1.05x faster │
│ QQuery 24 │     485.82 / 514.03 ±23.10 / 551.44 ms │          397.52 / 495.88 ±69.66 / 598.26 ms │     no change │
│ QQuery 25 │  2283.33 / 2312.63 ±18.80 / 2333.35 ms │       1508.10 / 1557.51 ±39.91 / 1625.07 ms │ +1.48x faster │
│ QQuery 26 │     496.24 / 547.48 ±54.64 / 653.73 ms │          451.51 / 483.41 ±25.96 / 529.22 ms │ +1.13x faster │
│ QQuery 27 │  2655.00 / 2728.76 ±66.07 / 2808.20 ms │       1954.40 / 1989.37 ±42.13 / 2062.79 ms │ +1.37x faster │
│ QQuery 28 │ 4896.70 / 5070.86 ±129.88 / 5277.81 ms │       4137.15 / 4298.66 ±88.39 / 4390.35 ms │ +1.18x faster │
│ QQuery 29 │  2212.89 / 2263.60 ±55.27 / 2348.40 ms │       1501.21 / 1548.92 ±44.77 / 1606.10 ms │ +1.46x faster │
│ QQuery 30 │  2459.04 / 2535.63 ±50.25 / 2599.26 ms │       1693.19 / 1760.35 ±56.66 / 1859.43 ms │ +1.44x faster │
│ QQuery 31 │  2388.63 / 2452.96 ±61.04 / 2539.92 ms │       1654.28 / 1681.11 ±27.30 / 1730.43 ms │ +1.46x faster │
│ QQuery 32 │ 2505.63 / 2636.24 ±103.59 / 2751.60 ms │       1743.25 / 1785.50 ±48.97 / 1877.24 ms │ +1.48x faster │
│ QQuery 33 │  3407.68 / 3497.33 ±65.16 / 3552.30 ms │       2729.64 / 2752.35 ±20.96 / 2790.94 ms │ +1.27x faster │
│ QQuery 34 │  3457.93 / 3557.46 ±63.72 / 3629.68 ms │       2808.28 / 2833.56 ±25.26 / 2882.15 ms │ +1.26x faster │
│ QQuery 35 │  2336.40 / 2440.42 ±65.81 / 2539.61 ms │       1663.16 / 1732.67 ±64.41 / 1816.92 ms │ +1.41x faster │
│ QQuery 36 │     137.82 / 221.05 ±44.19 / 260.04 ms │          139.33 / 296.18 ±95.16 / 407.66 ms │  1.34x slower │
│ QQuery 37 │     143.48 / 192.27 ±32.57 / 236.10 ms │          134.71 / 253.04 ±78.68 / 360.26 ms │  1.32x slower │
│ QQuery 38 │     127.39 / 189.93 ±36.89 / 229.03 ms │          129.43 / 268.83 ±86.60 / 376.54 ms │  1.42x slower │
│ QQuery 39 │     192.85 / 284.20 ±50.92 / 343.32 ms │          191.99 / 340.63 ±91.34 / 448.39 ms │  1.20x slower │
│ QQuery 40 │     115.12 / 176.03 ±41.40 / 231.73 ms │          118.43 / 253.07 ±84.87 / 349.24 ms │  1.44x slower │
│ QQuery 41 │     115.09 / 168.47 ±35.81 / 214.14 ms │          118.08 / 242.02 ±81.33 / 348.02 ms │  1.44x slower │
│ QQuery 42 │     114.30 / 168.63 ±36.17 / 215.25 ms │          115.47 / 241.65 ±82.13 / 348.88 ms │  1.43x slower │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 88256.64ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 66615.16ms │
│ Average Time (HEAD)                                        │  2052.48ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  1549.19ms │
│ Queries Faster                                             │         33 │
│ Queries Slower                                             │          7 │
│ Queries with No Change                                     │          3 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	445.1s
Peak memory	9.1 GiB
Avg memory	1.8 GiB
CPU user	985.7s
CPU sys	109.6s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	335.1s
Peak memory	9.6 GiB
Avg memory	2.1 GiB
CPU user	989.5s
CPU sys	102.5s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T15:09:04Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (5ff7af6) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "4G"
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │  107.71 ms │                                   106.92 ms │     no change │
│ QQuery 2  │  519.86 ms │                                   773.75 ms │  1.49x slower │
│ QQuery 3  │  382.34 ms │                                   466.37 ms │  1.22x slower │
│ QQuery 4  │ 1262.77 ms │                                  1801.25 ms │  1.43x slower │
│ QQuery 5  │  497.70 ms │                                   605.72 ms │  1.22x slower │
│ QQuery 6  │  544.19 ms │                                   765.68 ms │  1.41x slower │
│ QQuery 7  │  562.37 ms │                                   775.41 ms │  1.38x slower │
│ QQuery 8  │  527.95 ms │                                   755.83 ms │  1.43x slower │
│ QQuery 9  │ 1379.77 ms │                                  1535.65 ms │  1.11x slower │
│ QQuery 10 │  704.14 ms │                                  1052.73 ms │  1.50x slower │
│ QQuery 11 │  943.87 ms │                                  1225.46 ms │  1.30x slower │
│ QQuery 12 │  348.95 ms │                                   380.14 ms │  1.09x slower │
│ QQuery 13 │  676.71 ms │                                   966.01 ms │  1.43x slower │
│ QQuery 14 │ 3809.02 ms │                                  4729.46 ms │  1.24x slower │
│ QQuery 15 │  405.51 ms │                                   620.20 ms │  1.53x slower │
│ QQuery 16 │  108.23 ms │                                   108.33 ms │     no change │
│ QQuery 17 │  636.77 ms │                                   998.48 ms │  1.57x slower │
│ QQuery 18 │  714.25 ms │                                  1012.36 ms │  1.42x slower │
│ QQuery 19 │  512.10 ms │                                   730.97 ms │  1.43x slower │
│ QQuery 20 │  325.57 ms │                                   461.87 ms │  1.42x slower │
│ QQuery 21 │  506.17 ms │                                   490.02 ms │     no change │
│ QQuery 22 │  486.21 ms │                                   523.01 ms │  1.08x slower │
│ QQuery 23 │ 2489.99 ms │                                  3303.62 ms │  1.33x slower │
│ QQuery 24 │ 2189.10 ms │                                  3117.24 ms │  1.42x slower │
│ QQuery 25 │  690.46 ms │                                   911.78 ms │  1.32x slower │
│ QQuery 26 │  585.06 ms │                                   720.73 ms │  1.23x slower │
│ QQuery 27 │  141.55 ms │                                   191.42 ms │  1.35x slower │
│ QQuery 28 │  343.39 ms │                                   424.48 ms │  1.24x slower │
│ QQuery 29 │  729.12 ms │                                  1072.42 ms │  1.47x slower │
│ QQuery 30 │  541.46 ms │                                   773.58 ms │  1.43x slower │
│ QQuery 31 │ 1137.40 ms │                                  2104.86 ms │  1.85x slower │
│ QQuery 32 │  411.45 ms │                                   484.98 ms │  1.18x slower │
│ QQuery 33 │  522.89 ms │                                   735.65 ms │  1.41x slower │
│ QQuery 34 │  196.28 ms │                                   231.27 ms │  1.18x slower │
│ QQuery 35 │  713.57 ms │                                  1019.30 ms │  1.43x slower │
│ QQuery 36 │  107.12 ms │                                   108.61 ms │     no change │
│ QQuery 37 │  222.01 ms │                                   235.41 ms │  1.06x slower │
│ QQuery 38 │  641.58 ms │                                   868.19 ms │  1.35x slower │
│ QQuery 39 │ 1847.35 ms │                                  2253.37 ms │  1.22x slower │
│ QQuery 40 │  464.29 ms │                                   582.24 ms │  1.25x slower │
│ QQuery 41 │  175.93 ms │                                   176.34 ms │     no change │
│ QQuery 42 │  313.04 ms │                                   424.75 ms │  1.36x slower │
│ QQuery 43 │  106.27 ms │                                    76.87 ms │ +1.38x faster │
│ QQuery 44 │  197.68 ms │                                   197.45 ms │     no change │
│ QQuery 45 │  451.01 ms │                                   566.88 ms │  1.26x slower │
│ QQuery 46 │  290.87 ms │                                   525.32 ms │  1.81x slower │
│ QQuery 47 │ 1185.24 ms │                                  1531.58 ms │  1.29x slower │
│ QQuery 48 │  586.89 ms │                                   860.34 ms │  1.47x slower │
│ QQuery 49 │  579.55 ms │                                   713.51 ms │  1.23x slower │
│ QQuery 50 │  542.57 ms │                                   787.60 ms │  1.45x slower │
│ QQuery 51 │  520.67 ms │                                   682.18 ms │  1.31x slower │
│ QQuery 52 │  312.58 ms │                                   445.38 ms │  1.42x slower │
│ QQuery 53 │  408.33 ms │                                   446.71 ms │  1.09x slower │
│ QQuery 54 │  856.79 ms │                                  1146.19 ms │  1.34x slower │
│ QQuery 55 │  376.31 ms │                                   427.07 ms │  1.13x slower │
│ QQuery 56 │  522.52 ms │                                   713.75 ms │  1.37x slower │
│ QQuery 57 │ 1086.59 ms │                                  1383.35 ms │  1.27x slower │
│ QQuery 58 │ 1161.46 ms │                                  1486.29 ms │  1.28x slower │
│ QQuery 59 │  908.67 ms │                                  1006.37 ms │  1.11x slower │
│ QQuery 60 │  582.73 ms │                                   645.35 ms │  1.11x slower │
│ QQuery 61 │  395.13 ms │                                   390.40 ms │     no change │
│ QQuery 62 │  579.22 ms │                                   541.37 ms │ +1.07x faster │
│ QQuery 63 │  412.95 ms │                                   450.44 ms │  1.09x slower │
│ QQuery 64 │ 3103.02 ms │                                  4581.49 ms │  1.48x slower │
│ QQuery 65 │  608.88 ms │                                   966.99 ms │  1.59x slower │
│ QQuery 66 │  559.87 ms │                                   684.01 ms │  1.22x slower │
│ QQuery 67 │  676.39 ms │                                   671.85 ms │     no change │
│ QQuery 68 │  282.13 ms │                                   505.36 ms │  1.79x slower │
│ QQuery 69 │  699.05 ms │                                  1117.58 ms │  1.60x slower │
│ QQuery 70 │  701.82 ms │                                   792.69 ms │  1.13x slower │
│ QQuery 71 │  475.37 ms │                                   649.94 ms │  1.37x slower │
│ QQuery 72 │ 3044.03 ms │                                  3546.89 ms │  1.17x slower │
│ QQuery 73 │  194.32 ms │                                   259.27 ms │  1.33x slower │
│ QQuery 74 │  754.02 ms │                                  1070.66 ms │  1.42x slower │
│ QQuery 75 │  926.92 ms │                                  1259.71 ms │  1.36x slower │
│ QQuery 76 │  544.05 ms │                                   613.68 ms │  1.13x slower │
│ QQuery 77 │  613.58 ms │                                   758.76 ms │  1.24x slower │
│ QQuery 78 │  701.90 ms │                                  1357.90 ms │  1.93x slower │
│ QQuery 79 │  451.00 ms │                                   583.44 ms │  1.29x slower │
│ QQuery 80 │  649.68 ms │                                   919.94 ms │  1.42x slower │
│ QQuery 81 │  477.80 ms │                                   785.57 ms │  1.64x slower │
│ QQuery 82 │  555.59 ms │                                   746.15 ms │  1.34x slower │
│ QQuery 83 │  864.28 ms │                                  1142.97 ms │  1.32x slower │
│ QQuery 84 │  444.24 ms │                                   692.08 ms │  1.56x slower │
│ QQuery 85 │  819.22 ms │                                  1090.85 ms │  1.33x slower │
│ QQuery 86 │  247.55 ms │                                   376.78 ms │  1.52x slower │
│ QQuery 87 │  591.86 ms │                                   941.08 ms │  1.59x slower │
│ QQuery 88 │  736.39 ms │                                   811.79 ms │  1.10x slower │
│ QQuery 89 │  425.01 ms │                                   511.71 ms │  1.20x slower │
│ QQuery 90 │  425.49 ms │                                   470.96 ms │  1.11x slower │
│ QQuery 91 │  524.19 ms │                                  1043.10 ms │  1.99x slower │
│ QQuery 92 │  439.63 ms │                                   504.60 ms │  1.15x slower │
│ QQuery 93 │  471.06 ms │                                   529.59 ms │  1.12x slower │
│ QQuery 94 │  477.38 ms │                                   690.56 ms │  1.45x slower │
│ QQuery 95 │  695.69 ms │                                  1141.76 ms │  1.64x slower │
│ QQuery 96 │  403.54 ms │                                   502.38 ms │  1.24x slower │
│ QQuery 97 │  439.31 ms │                                   622.15 ms │  1.42x slower │
│ QQuery 98 │  395.91 ms │                                   478.60 ms │  1.21x slower │
│ QQuery 99 │  506.66 ms │                                   581.55 ms │  1.15x slower │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 67416.08ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 89654.66ms │
│ Average Time (HEAD)                                        │   680.97ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   905.60ms │
│ Queries Faster                                             │          2 │
│ Queries Slower                                             │         89 │
│ Queries with No Change                                     │          8 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     107.71 / 173.55 ±37.42 / 207.78 ms │          106.92 / 173.36 ±37.67 / 207.90 ms │     no change │
│ QQuery 2  │     519.86 / 648.67 ±69.09 / 702.66 ms │          773.75 / 886.18 ±78.92 / 996.08 ms │  1.37x slower │
│ QQuery 3  │     382.34 / 428.09 ±31.45 / 474.55 ms │          466.37 / 561.00 ±91.00 / 695.94 ms │  1.31x slower │
│ QQuery 4  │  1262.77 / 1329.47 ±57.10 / 1409.15 ms │       1801.25 / 1898.68 ±83.14 / 2048.82 ms │  1.43x slower │
│ QQuery 5  │     497.70 / 577.30 ±69.21 / 689.98 ms │          605.72 / 671.59 ±68.61 / 789.23 ms │  1.16x slower │
│ QQuery 6  │    544.19 / 697.87 ±103.39 / 817.41 ms │          765.68 / 886.51 ±88.59 / 998.68 ms │  1.27x slower │
│ QQuery 7  │     562.37 / 685.24 ±68.45 / 754.39 ms │          775.41 / 890.32 ±79.48 / 956.27 ms │  1.30x slower │
│ QQuery 8  │     527.95 / 590.21 ±44.74 / 659.85 ms │           755.83 / 758.18 ±1.77 / 760.24 ms │  1.28x slower │
│ QQuery 9  │  1379.77 / 1487.73 ±89.54 / 1589.83 ms │       1535.65 / 1640.71 ±78.21 / 1758.75 ms │  1.10x slower │
│ QQuery 10 │     704.14 / 767.59 ±45.43 / 843.34 ms │       1052.73 / 1164.10 ±60.95 / 1237.13 ms │  1.52x slower │
│ QQuery 11 │   943.87 / 1063.11 ±83.75 / 1196.90 ms │       1225.46 / 1295.64 ±57.10 / 1392.13 ms │  1.22x slower │
│ QQuery 12 │     348.95 / 374.92 ±25.96 / 419.37 ms │          380.14 / 476.93 ±69.29 / 542.78 ms │  1.27x slower │
│ QQuery 13 │     676.71 / 745.19 ±47.62 / 799.45 ms │         966.01 / 998.04 ±39.71 / 1075.88 ms │  1.34x slower │
│ QQuery 14 │ 3809.02 / 4091.94 ±171.24 / 4292.78 ms │       4729.46 / 4801.14 ±92.52 / 4983.94 ms │  1.17x slower │
│ QQuery 15 │     405.51 / 480.39 ±44.61 / 533.75 ms │          620.20 / 656.25 ±21.45 / 683.87 ms │  1.37x slower │
│ QQuery 16 │     108.23 / 175.17 ±38.04 / 208.28 ms │          108.33 / 170.38 ±34.94 / 208.87 ms │     no change │
│ QQuery 17 │     636.77 / 712.96 ±42.69 / 759.40 ms │        998.48 / 1024.67 ±32.16 / 1086.81 ms │  1.44x slower │
│ QQuery 18 │     714.25 / 802.92 ±61.23 / 903.90 ms │      1012.36 / 1157.97 ±113.81 / 1264.33 ms │  1.44x slower │
│ QQuery 19 │     512.10 / 573.77 ±60.89 / 680.69 ms │          730.97 / 863.79 ±95.44 / 974.30 ms │  1.51x slower │
│ QQuery 20 │     325.57 / 399.20 ±40.60 / 446.56 ms │          461.87 / 515.65 ±51.89 / 579.05 ms │  1.29x slower │
│ QQuery 21 │     506.17 / 570.62 ±48.67 / 631.19 ms │          490.02 / 626.54 ±96.71 / 735.26 ms │  1.10x slower │
│ QQuery 22 │     486.21 / 534.45 ±29.82 / 579.41 ms │          523.01 / 619.30 ±62.33 / 694.23 ms │  1.16x slower │
│ QQuery 23 │ 2489.99 / 2753.66 ±141.72 / 2904.31 ms │      3303.62 / 3620.08 ±184.87 / 3826.36 ms │  1.31x slower │
│ QQuery 24 │ 2189.10 / 2350.68 ±102.16 / 2465.04 ms │       3117.24 / 3132.89 ±10.64 / 3147.87 ms │  1.33x slower │
│ QQuery 25 │     690.46 / 737.28 ±35.86 / 781.66 ms │        911.78 / 1010.45 ±78.62 / 1131.16 ms │  1.37x slower │
│ QQuery 26 │     585.06 / 630.46 ±27.68 / 672.14 ms │          720.73 / 751.77 ±27.46 / 798.17 ms │  1.19x slower │
│ QQuery 27 │     141.55 / 254.33 ±84.46 / 359.66 ms │          191.42 / 223.57 ±32.82 / 279.89 ms │ +1.14x faster │
│ QQuery 28 │     343.39 / 400.05 ±32.95 / 433.33 ms │          424.48 / 524.45 ±69.62 / 643.12 ms │  1.31x slower │
│ QQuery 29 │     729.12 / 783.16 ±45.45 / 834.83 ms │       1072.42 / 1160.59 ±58.83 / 1223.57 ms │  1.48x slower │
│ QQuery 30 │     541.46 / 630.49 ±76.61 / 740.32 ms │          773.58 / 856.62 ±60.44 / 957.67 ms │  1.36x slower │
│ QQuery 31 │  1137.40 / 1242.13 ±83.53 / 1346.97 ms │       2104.86 / 2205.35 ±77.96 / 2283.89 ms │  1.78x slower │
│ QQuery 32 │     411.45 / 456.31 ±36.23 / 487.32 ms │          484.98 / 583.56 ±58.32 / 648.42 ms │  1.28x slower │
│ QQuery 33 │     522.89 / 576.47 ±28.94 / 610.77 ms │          735.65 / 775.66 ±29.49 / 811.22 ms │  1.35x slower │
│ QQuery 34 │     196.28 / 306.78 ±58.97 / 360.12 ms │          231.27 / 294.75 ±60.79 / 375.90 ms │     no change │
│ QQuery 35 │     713.57 / 771.51 ±56.55 / 874.90 ms │       1019.30 / 1121.13 ±70.53 / 1222.43 ms │  1.45x slower │
│ QQuery 36 │     107.12 / 173.61 ±37.91 / 208.53 ms │          108.61 / 174.39 ±37.78 / 210.11 ms │     no change │
│ QQuery 37 │     222.01 / 277.69 ±30.13 / 310.94 ms │          235.41 / 321.24 ±51.65 / 377.65 ms │  1.16x slower │
│ QQuery 38 │     641.58 / 695.74 ±37.44 / 739.51 ms │          868.19 / 919.84 ±37.65 / 963.58 ms │  1.32x slower │
│ QQuery 39 │ 1847.35 / 2004.60 ±113.16 / 2139.06 ms │        2253.37 / 2256.26 ±1.94 / 2258.46 ms │  1.13x slower │
│ QQuery 40 │     464.29 / 493.73 ±23.16 / 514.15 ms │          582.24 / 651.11 ±67.24 / 764.67 ms │  1.32x slower │
│ QQuery 41 │     175.93 / 278.68 ±73.48 / 381.49 ms │          176.34 / 278.89 ±73.18 / 381.66 ms │     no change │
│ QQuery 42 │     313.04 / 406.58 ±66.54 / 469.44 ms │          424.75 / 508.06 ±64.51 / 608.33 ms │  1.25x slower │
│ QQuery 43 │     106.27 / 153.42 ±40.93 / 206.85 ms │           76.87 / 122.80 ±44.31 / 207.30 ms │ +1.25x faster │
│ QQuery 44 │     197.68 / 262.02 ±46.86 / 301.64 ms │          197.45 / 241.40 ±47.81 / 302.02 ms │ +1.09x faster │
│ QQuery 45 │     451.01 / 532.42 ±78.73 / 646.05 ms │          566.88 / 676.77 ±58.96 / 729.30 ms │  1.27x slower │
│ QQuery 46 │     290.87 / 347.90 ±40.26 / 400.02 ms │          525.32 / 595.42 ±55.74 / 666.91 ms │  1.71x slower │
│ QQuery 47 │  1185.24 / 1277.11 ±56.32 / 1339.46 ms │      1531.58 / 1691.49 ±123.96 / 1828.87 ms │  1.32x slower │
│ QQuery 48 │     586.89 / 704.35 ±64.89 / 767.71 ms │          860.34 / 940.31 ±40.65 / 972.04 ms │  1.33x slower │
│ QQuery 49 │     579.55 / 633.07 ±47.90 / 700.54 ms │          713.51 / 762.19 ±41.24 / 812.47 ms │  1.20x slower │
│ QQuery 50 │     542.57 / 640.05 ±52.65 / 690.72 ms │          787.60 / 861.02 ±65.80 / 959.68 ms │  1.35x slower │
│ QQuery 51 │     520.67 / 537.99 ±14.14 / 550.77 ms │          682.18 / 761.79 ±70.10 / 874.33 ms │  1.42x slower │
│ QQuery 52 │     312.58 / 407.67 ±64.79 / 474.39 ms │          445.38 / 501.36 ±42.92 / 559.57 ms │  1.23x slower │
│ QQuery 53 │     408.33 / 450.05 ±47.26 / 509.37 ms │          446.71 / 572.80 ±64.03 / 619.85 ms │  1.27x slower │
│ QQuery 54 │    856.79 / 951.74 ±64.38 / 1036.33 ms │       1146.19 / 1202.47 ±51.05 / 1298.23 ms │  1.26x slower │
│ QQuery 55 │     376.31 / 424.24 ±32.17 / 472.94 ms │          427.07 / 510.18 ±64.94 / 611.92 ms │  1.20x slower │
│ QQuery 56 │     522.52 / 587.33 ±34.44 / 618.46 ms │          713.75 / 791.18 ±52.54 / 847.87 ms │  1.35x slower │
│ QQuery 57 │  1086.59 / 1209.58 ±84.71 / 1322.89 ms │       1383.35 / 1526.50 ±90.48 / 1641.59 ms │  1.26x slower │
│ QQuery 58 │  1161.46 / 1328.95 ±95.04 / 1424.12 ms │       1486.29 / 1557.80 ±43.91 / 1600.89 ms │  1.17x slower │
│ QQuery 59 │    908.67 / 933.47 ±48.51 / 1030.49 ms │       1006.37 / 1093.11 ±69.39 / 1210.71 ms │  1.17x slower │
│ QQuery 60 │     582.73 / 668.18 ±59.01 / 737.15 ms │          645.35 / 754.09 ±84.78 / 862.24 ms │  1.13x slower │
│ QQuery 61 │      395.13 / 395.72 ±0.48 / 396.30 ms │          390.40 / 406.63 ±17.32 / 427.91 ms │     no change │
│ QQuery 62 │     579.22 / 601.14 ±24.69 / 632.90 ms │          541.37 / 651.02 ±81.35 / 786.06 ms │  1.08x slower │
│ QQuery 63 │     412.95 / 440.16 ±28.07 / 475.79 ms │          450.44 / 516.72 ±48.80 / 589.09 ms │  1.17x slower │
│ QQuery 64 │ 3103.02 / 3328.53 ±120.38 / 3445.36 ms │       4581.49 / 4701.73 ±72.41 / 4767.96 ms │  1.41x slower │
│ QQuery 65 │     608.88 / 721.26 ±70.16 / 817.53 ms │        966.99 / 1037.07 ±61.62 / 1133.67 ms │  1.44x slower │
│ QQuery 66 │     559.87 / 599.59 ±41.66 / 678.05 ms │          684.01 / 759.71 ±53.18 / 818.52 ms │  1.27x slower │
│ QQuery 67 │     676.39 / 756.49 ±65.27 / 865.22 ms │          671.85 / 734.72 ±42.09 / 790.85 ms │     no change │
│ QQuery 68 │     282.13 / 306.93 ±18.14 / 330.96 ms │          505.36 / 550.88 ±51.70 / 645.64 ms │  1.79x slower │
│ QQuery 69 │     699.05 / 775.50 ±84.41 / 930.68 ms │       1117.58 / 1268.38 ±85.55 / 1370.94 ms │  1.64x slower │
│ QQuery 70 │     701.82 / 774.27 ±66.53 / 883.88 ms │          792.69 / 881.32 ±61.41 / 969.41 ms │  1.14x slower │
│ QQuery 71 │     475.37 / 559.70 ±62.68 / 660.60 ms │          649.94 / 730.37 ±53.22 / 806.66 ms │  1.30x slower │
│ QQuery 72 │ 3044.03 / 3165.95 ±115.92 / 3380.91 ms │      3546.89 / 3750.38 ±123.34 / 3892.05 ms │  1.18x slower │
│ QQuery 73 │     194.32 / 310.16 ±59.67 / 362.85 ms │          259.27 / 344.28 ±46.67 / 395.58 ms │  1.11x slower │
│ QQuery 74 │     754.02 / 813.94 ±54.84 / 906.22 ms │       1070.66 / 1154.38 ±63.05 / 1256.96 ms │  1.42x slower │
│ QQuery 75 │    926.92 / 992.83 ±49.53 / 1078.85 ms │      1259.71 / 1446.22 ±130.80 / 1631.07 ms │  1.46x slower │
│ QQuery 76 │     544.05 / 608.33 ±50.96 / 659.16 ms │          613.68 / 731.80 ±95.34 / 812.26 ms │  1.20x slower │
│ QQuery 77 │     613.58 / 726.39 ±69.44 / 804.74 ms │        758.76 / 930.82 ±110.02 / 1077.70 ms │  1.28x slower │
│ QQuery 78 │     701.90 / 857.68 ±82.35 / 946.42 ms │       1357.90 / 1380.76 ±17.85 / 1401.57 ms │  1.61x slower │
│ QQuery 79 │     451.00 / 511.41 ±53.61 / 604.97 ms │          583.44 / 686.79 ±63.22 / 747.65 ms │  1.34x slower │
│ QQuery 80 │     649.68 / 745.90 ±66.55 / 827.47 ms │         919.94 / 977.45 ±49.99 / 1037.83 ms │  1.31x slower │
│ QQuery 81 │     477.80 / 627.44 ±78.18 / 689.01 ms │          785.57 / 895.89 ±58.70 / 952.08 ms │  1.43x slower │
│ QQuery 82 │     555.59 / 642.92 ±47.76 / 681.70 ms │          746.15 / 892.27 ±91.56 / 968.23 ms │  1.39x slower │
│ QQuery 83 │    864.28 / 982.67 ±78.78 / 1098.97 ms │       1142.97 / 1212.42 ±39.60 / 1264.48 ms │  1.23x slower │
│ QQuery 84 │     444.24 / 503.37 ±56.72 / 595.05 ms │          692.08 / 771.19 ±61.79 / 879.61 ms │  1.53x slower │
│ QQuery 85 │     819.22 / 889.00 ±52.15 / 955.65 ms │      1090.85 / 1233.08 ±117.86 / 1352.21 ms │  1.39x slower │
│ QQuery 86 │     247.55 / 316.54 ±56.17 / 395.79 ms │          376.78 / 428.52 ±44.71 / 489.65 ms │  1.35x slower │
│ QQuery 87 │     591.86 / 666.41 ±66.35 / 787.35 ms │        941.08 / 1008.22 ±68.99 / 1131.74 ms │  1.51x slower │
│ QQuery 88 │      736.39 / 737.64 ±0.82 / 738.89 ms │          811.79 / 856.19 ±50.30 / 950.52 ms │  1.16x slower │
│ QQuery 89 │     425.01 / 459.35 ±27.91 / 483.21 ms │          511.71 / 545.87 ±32.75 / 604.73 ms │  1.19x slower │
│ QQuery 90 │     425.49 / 466.52 ±26.00 / 500.42 ms │          470.96 / 508.61 ±24.42 / 533.97 ms │  1.09x slower │
│ QQuery 91 │    524.19 / 663.79 ±112.28 / 805.35 ms │        1043.10 / 1046.85 ±2.41 / 1049.77 ms │  1.58x slower │
│ QQuery 92 │      439.63 / 445.22 ±3.91 / 449.72 ms │          504.60 / 578.65 ±43.93 / 632.00 ms │  1.30x slower │
│ QQuery 93 │     471.06 / 482.26 ±11.68 / 501.88 ms │          529.59 / 617.69 ±55.73 / 677.86 ms │  1.28x slower │
│ QQuery 94 │     477.38 / 528.91 ±44.14 / 588.01 ms │          690.56 / 798.78 ±66.49 / 878.12 ms │  1.51x slower │
│ QQuery 95 │     695.69 / 779.66 ±45.63 / 825.44 ms │       1141.76 / 1200.99 ±40.20 / 1245.61 ms │  1.54x slower │
│ QQuery 96 │     403.54 / 446.45 ±49.78 / 507.91 ms │          502.38 / 535.03 ±34.64 / 595.84 ms │  1.20x slower │
│ QQuery 97 │     439.31 / 557.41 ±89.83 / 660.61 ms │          622.15 / 658.20 ±41.46 / 731.49 ms │  1.18x slower │
│ QQuery 98 │     395.91 / 439.87 ±31.40 / 484.66 ms │          478.60 / 555.32 ±59.56 / 639.39 ms │  1.26x slower │
│ QQuery 99 │     506.66 / 617.98 ±64.64 / 694.81 ms │          581.55 / 737.34 ±80.10 / 811.33 ms │  1.19x slower │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 75429.13ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 97992.87ms │
│ Average Time (HEAD)                                        │   761.91ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   989.83ms │
│ Queries Faster                                             │          3 │
│ Queries Slower                                             │         89 │
│ Queries with No Change                                     │          7 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only when the benchmark runs with DATAFUSION_RUNTIME_MEMORY_LIMIT set.

Base: 39d5064 (merge-base) | Changed: claude/datafusion-rowgroup-buffering-ec9abc

tpcds — tpcds_sf1

Query	Base	Changed	Change
Query 1	219 B	219 B	+0.0%
Query 2	73.4 MiB	62.4 MiB	-15.0%
Query 3	5.1 MiB	4.8 MiB	-6.2%
Query 4	153.1 MiB	154.2 MiB	+0.7%
Query 5	13.2 MiB	13.9 MiB	+4.8%
Query 6	8.0 MiB	7.8 MiB	-2.7%
Query 7	6.9 MiB	6.9 MiB	+0.0%
Query 8	2.4 MiB	2.6 MiB	+7.7%
Query 9	32.1 KiB	32.1 KiB	+0.0%
Query 10	4.4 MiB	4.4 MiB	+0.0%
Query 11	102.5 MiB	102.5 MiB	+0.0%
Query 12	132.7 MiB	132.7 MiB	+0.0%
Query 13	18.7 MiB	18.0 MiB	-3.6%
Query 14	159.6 MiB	159.5 MiB	-0.1%
Query 15	7.0 MiB	7.0 MiB	+0.0%
Query 16	539 B	603 B	+11.9%
Query 17	37.3 MiB	37.3 MiB	+0.0%
Query 18	16.6 MiB	18.3 MiB	+10.1%
Query 19	7.8 MiB	7.4 MiB	-5.7%
Query 20	132.8 MiB	132.8 MiB	+0.0%
Query 21	5.1 MiB	5.3 MiB	+4.2%
Query 22	55.5 MiB	65.4 MiB	+17.9%
Query 23	368.2 MiB	372.6 MiB	+1.2%
Query 24	36.7 MiB	36.7 MiB	+0.0%
Query 25	80.4 MiB	78.9 MiB	-1.9%
Query 26	7.5 MiB	7.5 MiB	-0.1%
Query 27	1.7 MiB	1.7 MiB	-0.0%
Query 28	1.2 MiB	679.9 KiB	-44.4%
Query 29	90.8 MiB	91.0 MiB	+0.1%
Query 30	6.6 MiB	6.6 MiB	-0.8%
Query 31	32.3 MiB	32.0 MiB	-1.0%
Query 32	2.7 MiB	2.7 MiB	+0.0%
Query 33	4.8 MiB	4.9 MiB	+2.2%
Query 34	8.0 MiB	8.0 MiB	+0.0%
Query 35	28.6 MiB	28.6 MiB	+0.0%
Query 36	10.6 MiB	10.6 MiB	+0.0%
Query 37	3.8 MiB	2.8 MiB	-25.6%
Query 38	20.3 MiB	20.6 MiB	+1.4%
Query 39	45.2 MiB	54.6 MiB	+20.6%
Query 40	11.1 MiB	11.1 MiB	+0.0%
Query 41	2.1 MiB	2.1 MiB	+0.0%
Query 42	2.5 MiB	2.5 MiB	+0.0%
Query 43	2.4 KiB	2.7 KiB	+13.9%
Query 44	1008 B	1008 B	+0.0%
Query 45	8.4 MiB	8.8 MiB	+4.5%
Query 46	8.1 MiB	8.0 MiB	-1.2%
Query 47	163.0 MiB	163.0 MiB	+0.0%
Query 48	15.5 MiB	15.5 MiB	+0.0%
Query 49	71.1 MiB	52.3 MiB	-26.4%
Query 50	13.6 MiB	13.6 MiB	+0.0%
Query 51	200.1 MiB	200.1 MiB	+0.0%
Query 52	4.2 MiB	4.2 MiB	+0.0%
Query 53	30.8 MiB	30.8 MiB	+0.0%
Query 54	3.4 MiB	3.3 MiB	-1.8%
Query 55	4.1 MiB	3.8 MiB	-7.0%
Query 56	5.5 MiB	5.3 MiB	-3.7%
Query 57	156.9 MiB	156.9 MiB	+0.0%
Query 58	6.7 MiB	6.7 MiB	+0.0%
Query 59	24.9 MiB	24.9 MiB	+0.0%
Query 60	7.6 MiB	7.6 MiB	+0.2%
Query 61	1.4 MiB	1.4 MiB	+2.3%
Query 62	7.7 MiB	7.7 MiB	+0.0%
Query 63	30.8 MiB	30.8 MiB	-0.0%
Query 64	70.6 MiB	68.3 MiB	-3.3%
Query 65	34.0 MiB	34.0 MiB	+0.1%
Query 66	15.9 MiB	15.8 MiB	-0.6%
Query 67	515.6 MiB	515.5 MiB	-0.0%
Query 68	8.1 MiB	8.0 MiB	-1.5%
Query 69	47.9 MiB	47.9 MiB	+0.0%
Query 70	52.0 MiB	52.0 MiB	-0.0%
Query 71	93.0 MiB	93.0 MiB	+0.0%
Query 72	199.2 MiB	235.3 MiB	+18.1%
Query 73	8.0 MiB	8.0 MiB	+0.0%
Query 74	57.0 MiB	57.6 MiB	+1.0%
Query 75	21.5 MiB	26.4 MiB	+22.9%
Query 76	9.5 MiB	9.5 MiB	-0.0%
Query 77	3.5 MiB	3.3 MiB	-3.6%
Query 78	111.8 MiB	113.8 MiB	+1.8%
Query 79	9.6 MiB	9.6 MiB	+0.0%
Query 80	39.5 MiB	39.9 MiB	+1.0%
Query 81	17.1 MiB	17.1 MiB	+0.1%
Query 82	3.8 MiB	2.9 MiB	-21.9%
Query 83	11.4 MiB	11.6 MiB	+1.7%
Query 84	13.5 MiB	13.5 MiB	+0.0%
Query 85	13.9 MiB	13.9 MiB	+0.0%
Query 86	105.5 MiB	105.5 MiB	-0.0%
Query 87	20.4 MiB	20.1 MiB	-1.7%
Query 88	1.2 MiB	1.2 MiB	+5.1%
Query 89	130.4 MiB	131.2 MiB	+0.6%
Query 90	619.0 KiB	619.0 KiB	+0.0%
Query 91	22.1 MiB	31.4 MiB	+41.8%
Query 92	2.7 MiB	2.9 MiB	+8.0%
Query 93	14.0 MiB	13.8 MiB	-1.8%
Query 94	3.8 MiB	3.8 MiB	+0.0%
Query 95	14.3 MiB	14.3 MiB	+0.0%
Query 96	224.6 KiB	224.6 KiB	+0.0%
Query 97	46.2 MiB	46.7 MiB	+0.9%
Query 98	133.1 MiB	133.1 MiB	+0.0%
Query 99	7.8 MiB	7.8 MiB	-0.1%

Pool accounting vs. process RSS

Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpcds	base (`39d5064 (merge-base)`)	515.6 MiB	1.0 GiB	554.7 MiB	2.1×
tpcds	changed (`claude/datafusion-rowgroup-buffering-ec9abc`)	515.5 MiB	906.7 MiB	391.1 MiB	1.8×

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	385.1s
Peak memory	1.0 GiB
Avg memory	543.1 MiB
CPU user	238.7s
CPU sys	12.2s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	495.1s
Peak memory	906.7 MiB
Avg memory	370.4 MiB
CPU user	246.0s
CPU sys	13.2s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-08-04T15:10:02Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180904706-1387-pckl6 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (bcbbda1) to 39d5064 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-08-04T15:11:11Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5180904706-1388-gfkbh 6.12.85+ #1 SMP Wed Jun 17 20:31:55 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (bcbbda1) to 39d5064 (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DF_FETCH_BUDGET: "104857600"
  DF_FETCH_POLICY: "streaming"
  SIMULATE_LATENCY: "true"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:27:17Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5897086822-2918-6jb8f 6.12.94+ #1 SMP Wed Aug 19 07:47:20 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: "true"
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:38:50Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch
env:
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │  568.36 ms │                                   187.44 ms │ +3.03x faster │
│ QQuery 2  │  636.80 ms │                                   535.52 ms │ +1.19x faster │
│ QQuery 3  │  927.83 ms │                                   443.48 ms │ +2.09x faster │
│ QQuery 4  │  664.25 ms │                                   348.57 ms │ +1.91x faster │
│ QQuery 5  │  729.90 ms │                                   527.54 ms │ +1.38x faster │
│ QQuery 6  │  482.42 ms │                                   168.73 ms │ +2.86x faster │
│ QQuery 7  │  774.86 ms │                                   563.57 ms │ +1.37x faster │
│ QQuery 8  │  933.40 ms │                                   502.07 ms │ +1.86x faster │
│ QQuery 9  │ 1093.09 ms │                                   618.00 ms │ +1.77x faster │
│ QQuery 10 │  839.41 ms │                                   451.04 ms │ +1.86x faster │
│ QQuery 11 │  580.59 ms │                                   431.62 ms │ +1.35x faster │
│ QQuery 12 │  619.65 ms │                                   320.64 ms │ +1.93x faster │
│ QQuery 13 │  336.78 ms │                                   254.18 ms │ +1.32x faster │
│ QQuery 14 │  561.15 ms │                                   210.03 ms │ +2.67x faster │
│ QQuery 15 │  930.11 ms │                                   369.97 ms │ +2.51x faster │
│ QQuery 16 │  228.64 ms │                                   182.90 ms │ +1.25x faster │
│ QQuery 17 │ 1112.16 ms │                                   453.35 ms │ +2.45x faster │
│ QQuery 18 │ 1366.04 ms │                                   726.77 ms │ +1.88x faster │
│ QQuery 19 │  632.51 ms │                                   249.58 ms │ +2.53x faster │
│ QQuery 20 │  801.74 ms │                                   434.10 ms │ +1.85x faster │
│ QQuery 21 │ 1751.17 ms │                                   853.56 ms │ +2.05x faster │
│ QQuery 22 │  445.38 ms │                                   440.36 ms │     no change │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 17016.23ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  9273.01ms │
│ Average Time (HEAD)                                        │   773.47ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   421.50ms │
│ Queries Faster                                             │         21 │
│ Queries Slower                                             │          0 │
│ Queries with No Change                                     │          1 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     568.36 / 640.95 ±56.80 / 732.74 ms │          187.44 / 209.80 ±16.57 / 224.16 ms │ +3.06x faster │
│ QQuery 2  │   636.80 / 895.43 ±182.05 / 1181.61 ms │         535.52 / 741.60 ±114.59 / 840.15 ms │ +1.21x faster │
│ QQuery 3  │    927.83 / 991.35 ±37.12 / 1034.21 ms │          443.48 / 530.85 ±68.29 / 637.91 ms │ +1.87x faster │
│ QQuery 4  │     664.25 / 766.11 ±68.34 / 866.06 ms │          348.57 / 376.46 ±33.29 / 417.83 ms │ +2.04x faster │
│ QQuery 5  │   729.90 / 912.35 ±136.91 / 1034.00 ms │          527.54 / 580.53 ±42.45 / 639.47 ms │ +1.57x faster │
│ QQuery 6  │     482.42 / 535.18 ±29.16 / 559.14 ms │          168.73 / 199.36 ±18.77 / 219.83 ms │ +2.68x faster │
│ QQuery 7  │  774.86 / 1027.01 ±156.25 / 1189.67 ms │          563.57 / 644.73 ±80.73 / 796.06 ms │ +1.59x faster │
│ QQuery 8  │    933.40 / 977.05 ±34.43 / 1039.32 ms │          502.07 / 632.61 ±70.04 / 690.90 ms │ +1.54x faster │
│ QQuery 9  │ 1093.09 / 1283.15 ±102.96 / 1371.85 ms │         618.00 / 735.96 ±128.16 / 947.24 ms │ +1.74x faster │
│ QQuery 10 │  839.41 / 1023.91 ±142.25 / 1217.09 ms │          451.04 / 531.62 ±44.79 / 573.83 ms │ +1.93x faster │
│ QQuery 11 │    580.59 / 746.07 ±133.45 / 978.64 ms │        431.62 / 758.38 ±192.36 / 1019.35 ms │     no change │
│ QQuery 12 │     619.65 / 719.63 ±91.05 / 827.10 ms │          320.64 / 396.01 ±38.72 / 426.93 ms │ +1.82x faster │
│ QQuery 13 │     336.78 / 374.82 ±34.20 / 437.87 ms │          254.18 / 357.93 ±54.53 / 407.07 ms │     no change │
│ QQuery 14 │    561.15 / 684.07 ±152.67 / 944.47 ms │          210.03 / 293.70 ±61.81 / 381.23 ms │ +2.33x faster │
│ QQuery 15 │  930.11 / 1141.42 ±108.55 / 1233.84 ms │          369.97 / 440.15 ±52.71 / 520.47 ms │ +2.59x faster │
│ QQuery 16 │     228.64 / 345.30 ±61.10 / 398.68 ms │          182.90 / 295.92 ±89.31 / 415.09 ms │ +1.17x faster │
│ QQuery 17 │ 1112.16 / 1291.33 ±120.56 / 1453.64 ms │          453.35 / 545.33 ±72.73 / 657.61 ms │ +2.37x faster │
│ QQuery 18 │ 1366.04 / 1587.14 ±169.01 / 1842.47 ms │          726.77 / 749.77 ±25.99 / 796.16 ms │ +2.12x faster │
│ QQuery 19 │    632.51 / 772.35 ±120.25 / 967.40 ms │          249.58 / 340.22 ±54.63 / 395.47 ms │ +2.27x faster │
│ QQuery 20 │    801.74 / 911.13 ±76.77 / 1031.17 ms │          434.10 / 543.30 ±62.59 / 615.49 ms │ +1.68x faster │
│ QQuery 21 │ 1751.17 / 1945.96 ±110.83 / 2040.58 ms │         853.56 / 909.15 ±70.24 / 1043.58 ms │ +2.14x faster │
│ QQuery 22 │     445.38 / 494.19 ±37.04 / 538.88 ms │          440.36 / 466.49 ±27.84 / 519.31 ms │ +1.06x faster │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 20065.91ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 11279.87ms │
│ Average Time (HEAD)                                        │   912.09ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   512.72ms │
│ Queries Faster                                             │         20 │
│ Queries Slower                                             │          0 │
│ Queries with No Change                                     │          2 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	105.0s
Peak memory	890.3 MiB
Avg memory	451.8 MiB
CPU user	30.3s
CPU sys	4.1s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	60.0s
Peak memory	1.0 GiB
Avg memory	505.2 MiB
CPU user	30.5s
CPU sys	3.5s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:40:35Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: "true"
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 1  │  899.50 ms │                                   175.33 ms │  +5.13x faster │
│ QQuery 2  │ 1659.41 ms │                                   648.99 ms │  +2.56x faster │
│ QQuery 3  │ 1926.78 ms │                                   377.83 ms │  +5.10x faster │
│ QQuery 4  │ 1323.04 ms │                                   356.75 ms │  +3.71x faster │
│ QQuery 5  │ 1665.73 ms │                                   448.32 ms │  +3.72x faster │
│ QQuery 6  │ 1860.15 ms │                                   171.39 ms │ +10.85x faster │
│ QQuery 7  │ 2352.35 ms │                                   552.45 ms │  +4.26x faster │
│ QQuery 8  │ 2667.71 ms │                                   596.50 ms │  +4.47x faster │
│ QQuery 9  │ 2014.77 ms │                                   710.97 ms │  +2.83x faster │
│ QQuery 10 │ 1964.01 ms │                                   480.27 ms │  +4.09x faster │
│ QQuery 11 │ 1166.39 ms │                                   469.17 ms │  +2.49x faster │
│ QQuery 12 │ 1927.90 ms │                                   347.13 ms │  +5.55x faster │
│ QQuery 13 │  576.02 ms │                                   313.67 ms │  +1.84x faster │
│ QQuery 14 │ 1054.44 ms │                                   262.89 ms │  +4.01x faster │
│ QQuery 15 │ 2129.43 ms │                                   396.61 ms │  +5.37x faster │
│ QQuery 16 │  595.94 ms │                                   233.77 ms │  +2.55x faster │
│ QQuery 17 │ 2150.99 ms │                                   392.46 ms │  +5.48x faster │
│ QQuery 18 │ 1848.25 ms │                                   703.56 ms │  +2.63x faster │
│ QQuery 19 │ 2229.46 ms │                                   225.77 ms │  +9.87x faster │
│ QQuery 20 │ 1917.40 ms │                                   439.68 ms │  +4.36x faster │
│ QQuery 21 │ 3927.95 ms │                                   832.04 ms │  +4.72x faster │
│ QQuery 22 │  725.85 ms │                                   386.68 ms │  +1.88x faster │
└───────────┴────────────┴─────────────────────────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 38583.47ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  9522.23ms │
│ Average Time (HEAD)                                        │  1753.79ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   432.83ms │
│ Queries Faster                                             │         22 │
│ Queries Slower                                             │          0 │
│ Queries with No Change                                     │          0 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │   899.50 / 1030.66 ±97.89 / 1195.58 ms │          175.33 / 208.09 ±23.22 / 235.88 ms │ +4.95x faster │
│ QQuery 2  │ 1659.41 / 2020.85 ±232.80 / 2325.03 ms │          648.99 / 769.36 ±93.90 / 909.79 ms │ +2.63x faster │
│ QQuery 3  │  1926.78 / 2101.59 ±92.41 / 2171.99 ms │          377.83 / 503.83 ±80.17 / 595.45 ms │ +4.17x faster │
│ QQuery 4  │  1323.04 / 1374.64 ±52.46 / 1467.40 ms │          356.75 / 393.69 ±30.88 / 432.29 ms │ +3.49x faster │
│ QQuery 5  │ 1665.73 / 1876.19 ±159.96 / 2129.53 ms │          448.32 / 506.23 ±38.35 / 561.80 ms │ +3.71x faster │
│ QQuery 6  │  1860.15 / 1903.33 ±39.56 / 1964.33 ms │          171.39 / 193.48 ±23.57 / 223.13 ms │ +9.84x faster │
│ QQuery 7  │  2352.35 / 2484.53 ±85.00 / 2594.10 ms │          552.45 / 630.98 ±61.19 / 739.86 ms │ +3.94x faster │
│ QQuery 8  │ 2667.71 / 2999.84 ±281.97 / 3406.21 ms │          596.50 / 657.60 ±45.23 / 734.38 ms │ +4.56x faster │
│ QQuery 9  │ 2014.77 / 2440.69 ±258.42 / 2782.56 ms │          710.97 / 752.54 ±38.02 / 804.16 ms │ +3.24x faster │
│ QQuery 10 │ 1964.01 / 2095.55 ±135.81 / 2304.64 ms │          480.27 / 559.36 ±69.01 / 649.11 ms │ +3.75x faster │
│ QQuery 11 │ 1166.39 / 1419.20 ±245.49 / 1753.73 ms │          469.17 / 525.17 ±41.78 / 589.18 ms │ +2.70x faster │
│ QQuery 12 │ 1927.90 / 2072.88 ±116.14 / 2277.55 ms │          347.13 / 383.52 ±38.78 / 440.38 ms │ +5.40x faster │
│ QQuery 13 │     576.02 / 650.14 ±39.33 / 687.79 ms │          313.67 / 401.18 ±44.67 / 430.44 ms │ +1.62x faster │
│ QQuery 14 │ 1054.44 / 1203.28 ±120.31 / 1412.72 ms │          262.89 / 339.01 ±54.12 / 429.63 ms │ +3.55x faster │
│ QQuery 15 │ 2129.43 / 2394.81 ±211.84 / 2647.66 ms │          396.61 / 445.18 ±38.33 / 488.87 ms │ +5.38x faster │
│ QQuery 16 │     595.94 / 724.01 ±84.91 / 824.22 ms │          233.77 / 348.22 ±61.44 / 417.36 ms │ +2.08x faster │
│ QQuery 17 │ 2150.99 / 2314.42 ±103.25 / 2431.36 ms │          392.46 / 531.57 ±82.88 / 628.20 ms │ +4.35x faster │
│ QQuery 18 │ 1848.25 / 2158.00 ±174.90 / 2327.38 ms │          703.56 / 765.55 ±62.12 / 847.07 ms │ +2.82x faster │
│ QQuery 19 │ 2229.46 / 2486.43 ±202.13 / 2835.69 ms │          225.77 / 327.59 ±80.21 / 442.88 ms │ +7.59x faster │
│ QQuery 20 │ 1917.40 / 2177.79 ±159.13 / 2390.10 ms │          439.68 / 537.82 ±58.40 / 623.27 ms │ +4.05x faster │
│ QQuery 21 │ 3927.95 / 4213.48 ±261.32 / 4580.09 ms │         832.04 / 890.46 ±71.81 / 1028.51 ms │ +4.73x faster │
│ QQuery 22 │     725.85 / 839.49 ±97.35 / 960.14 ms │          386.68 / 472.71 ±77.84 / 567.16 ms │ +1.78x faster │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 42981.80ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 11143.14ms │
│ Average Time (HEAD)                                        │  1953.72ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   506.51ms │
│ Queries Faster                                             │         22 │
│ Queries Slower                                             │          0 │
│ Queries with No Change                                     │          0 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	220.0s
Peak memory	677.1 MiB
Avg memory	451.1 MiB
CPU user	34.4s
CPU sys	4.9s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	60.0s
Peak memory	712.8 MiB
Avg memory	470.3 MiB
CPU user	40.5s
CPU sys	3.5s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:40:41Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "1G"
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 1  │  816.11 ms │                                   215.52 ms │  +3.79x faster │
│ QQuery 2  │ 1553.91 ms │                                   783.94 ms │  +1.98x faster │
│ QQuery 3  │ 2111.39 ms │                                   454.78 ms │  +4.64x faster │
│ QQuery 4  │ 1046.91 ms │                                   356.64 ms │  +2.94x faster │
│ QQuery 5  │ 1598.57 ms │                                   434.66 ms │  +3.68x faster │
│ QQuery 6  │ 1650.82 ms │                                   131.06 ms │ +12.60x faster │
│ QQuery 7  │ 2583.90 ms │                                   486.88 ms │  +5.31x faster │
│ QQuery 8  │ 2829.05 ms │                                   568.10 ms │  +4.98x faster │
│ QQuery 9  │ 2124.36 ms │                                   616.45 ms │  +3.45x faster │
│ QQuery 10 │ 2032.69 ms │                                   477.40 ms │  +4.26x faster │
│ QQuery 11 │ 1314.36 ms │                                   591.26 ms │  +2.22x faster │
│ QQuery 12 │ 2152.31 ms │                                   385.32 ms │  +5.59x faster │
│ QQuery 13 │  531.66 ms │                                   327.71 ms │  +1.62x faster │
│ QQuery 14 │  988.97 ms │                                   243.42 ms │  +4.06x faster │
│ QQuery 15 │ 2067.25 ms │                                   360.62 ms │  +5.73x faster │
│ QQuery 16 │  646.47 ms │                                   264.12 ms │  +2.45x faster │
│ QQuery 17 │ 1955.89 ms │                                   414.80 ms │  +4.72x faster │
│ QQuery 18 │ 2048.64 ms │                                   663.33 ms │  +3.09x faster │
│ QQuery 19 │ 2229.85 ms │                                   286.28 ms │  +7.79x faster │
│ QQuery 20 │ 2154.52 ms │                                   349.59 ms │  +6.16x faster │
│ QQuery 21 │ 4267.84 ms │                                   796.58 ms │  +5.36x faster │
│ QQuery 22 │  828.50 ms │                                   425.58 ms │  +1.95x faster │
└───────────┴────────────┴─────────────────────────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 39533.97ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  9634.02ms │
│ Average Time (HEAD)                                        │  1797.00ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   437.91ms │
│ Queries Faster                                             │         22 │
│ Queries Slower                                             │          0 │
│ Queries with No Change                                     │          0 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │   816.11 / 918.45 ±117.07 / 1132.88 ms │          215.52 / 230.03 ±11.63 / 243.97 ms │ +3.99x faster │
│ QQuery 2  │ 1553.91 / 1915.66 ±199.98 / 2128.25 ms │        783.94 / 937.23 ±108.57 / 1062.29 ms │ +2.04x faster │
│ QQuery 3  │ 2111.39 / 2262.08 ±106.35 / 2403.36 ms │          454.78 / 507.01 ±49.19 / 575.84 ms │ +4.46x faster │
│ QQuery 4  │ 1046.91 / 1322.61 ±152.18 / 1473.70 ms │          356.64 / 400.41 ±25.42 / 430.85 ms │ +3.30x faster │
│ QQuery 5  │ 1598.57 / 1870.71 ±144.39 / 2013.70 ms │          434.66 / 529.07 ±55.98 / 601.74 ms │ +3.54x faster │
│ QQuery 6  │  1650.82 / 1768.54 ±83.00 / 1879.63 ms │          131.06 / 198.22 ±35.03 / 226.53 ms │ +8.92x faster │
│ QQuery 7  │ 2583.90 / 2830.34 ±169.34 / 3111.84 ms │         486.88 / 617.01 ±134.97 / 822.36 ms │ +4.59x faster │
│ QQuery 8  │ 2829.05 / 3083.01 ±211.39 / 3444.16 ms │          568.10 / 620.35 ±45.20 / 700.29 ms │ +4.97x faster │
│ QQuery 9  │ 2124.36 / 2399.30 ±178.36 / 2641.69 ms │          616.45 / 739.26 ±79.77 / 863.59 ms │ +3.25x faster │
│ QQuery 10 │ 2032.69 / 2193.09 ±159.36 / 2496.00 ms │          477.40 / 541.65 ±43.44 / 602.03 ms │ +4.05x faster │
│ QQuery 11 │  1314.36 / 1405.62 ±75.12 / 1501.22 ms │          591.26 / 682.00 ±65.61 / 758.21 ms │ +2.06x faster │
│ QQuery 12 │ 2152.31 / 2246.12 ±134.11 / 2502.44 ms │          385.32 / 410.35 ±17.64 / 434.58 ms │ +5.47x faster │
│ QQuery 13 │     531.66 / 669.73 ±89.22 / 808.60 ms │          327.71 / 362.61 ±42.97 / 429.97 ms │ +1.85x faster │
│ QQuery 14 │   988.97 / 1108.31 ±95.56 / 1243.71 ms │          243.42 / 310.09 ±49.64 / 377.41 ms │ +3.57x faster │
│ QQuery 15 │ 2067.25 / 2400.23 ±230.50 / 2695.86 ms │          360.62 / 492.02 ±75.73 / 574.52 ms │ +4.88x faster │
│ QQuery 16 │   646.47 / 840.42 ±147.43 / 1052.46 ms │          264.12 / 290.68 ±15.08 / 308.59 ms │ +2.89x faster │
│ QQuery 17 │ 1955.89 / 2254.28 ±218.41 / 2536.58 ms │          414.80 / 500.09 ±60.06 / 570.00 ms │ +4.51x faster │
│ QQuery 18 │ 2048.64 / 2333.05 ±220.05 / 2680.52 ms │          663.33 / 758.69 ±54.41 / 823.14 ms │ +3.08x faster │
│ QQuery 19 │ 2229.85 / 2460.77 ±189.15 / 2718.59 ms │          286.28 / 368.22 ±49.94 / 435.03 ms │ +6.68x faster │
│ QQuery 20 │ 2154.52 / 2295.76 ±108.26 / 2460.23 ms │          349.59 / 487.22 ±76.76 / 567.05 ms │ +4.71x faster │
│ QQuery 21 │  4267.84 / 4363.01 ±81.25 / 4462.45 ms │         796.58 / 871.17 ±80.35 / 1015.84 ms │ +5.01x faster │
│ QQuery 22 │     828.50 / 886.45 ±50.82 / 977.98 ms │          425.58 / 469.48 ±47.29 / 555.21 ms │ +1.89x faster │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 43827.52ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 11322.88ms │
│ Average Time (HEAD)                                        │  1992.16ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   514.68ms │
│ Queries Faster                                             │         22 │
│ Queries Slower                                             │          0 │
│ Queries with No Change                                     │          0 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 953518c (merge-base) | Changed: claude/datafusion-rowgroup-buffering-ec9abc

tpch — tpch_sf1

Query	Base	Changed	Change
Query 1	22.8 MiB	49.2 MiB	+116.2%
Query 2	14.8 MiB	14.8 MiB	+0.0%
Query 3	11.5 MiB	57.9 MiB	+404.6%
Query 4	23.2 MiB	30.3 MiB	+30.8%
Query 5	32.7 MiB	64.2 MiB	+96.4%
Query 6	832 B	44.3 MiB	+5585866.2%
Query 7	97.5 MiB	114.4 MiB	+17.3%
Query 8	31.5 MiB	102.9 MiB	+226.6%
Query 9	151.6 MiB	209.6 MiB	+38.2%
Query 10	52.9 MiB	77.3 MiB	+46.1%
Query 11	40.6 MiB	31.0 MiB	-23.7%
Query 12	20.0 MiB	38.7 MiB	+93.7%
Query 13	52.8 MiB	62.3 MiB	+18.0%
Query 14	9.6 MiB	65.8 MiB	+586.3%
Query 15	12.0 MiB	52.3 MiB	+336.0%
Query 16	45.2 MiB	44.4 MiB	-1.7%
Query 17	7.0 MiB	58.7 MiB	+744.5%
Query 18	85.2 MiB	86.7 MiB	+1.8%
Query 19	187.1 KiB	64.9 MiB	+35395.4%
Query 20	11.0 MiB	50.7 MiB	+361.1%
Query 21	42.4 MiB	78.9 MiB	+85.9%
Query 22	25.0 MiB	25.3 MiB	+1.1%

Pool accounting vs. process RSS

Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpch	base (`953518c (merge-base)`)	151.6 MiB	673.9 MiB	522.3 MiB	4.4×
tpch	changed (`claude/datafusion-rowgroup-buffering-ec9abc`)	209.6 MiB	847.3 MiB	637.7 MiB	4.0×

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	225.1s
Peak memory	673.9 MiB
Avg memory	436.7 MiB
CPU user	35.0s
CPU sys	4.8s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	60.0s
Peak memory	847.3 MiB
Avg memory	519.8 MiB
CPU user	41.1s
CPU sys	3.4s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:47:38Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch10
env:
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃        HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 1  │  3906.43 ms │                                   521.43 ms │  +7.49x faster │
│ QQuery 2  │  3001.68 ms │                                   705.05 ms │  +4.26x faster │
│ QQuery 3  │  5434.39 ms │                                   666.04 ms │  +8.16x faster │
│ QQuery 4  │  4768.16 ms │                                   340.53 ms │ +14.00x faster │
│ QQuery 5  │  5452.50 ms │                                   941.48 ms │  +5.79x faster │
│ QQuery 6  │  3888.98 ms │                                   264.50 ms │ +14.70x faster │
│ QQuery 7  │  5719.59 ms │                                   861.42 ms │  +6.64x faster │
│ QQuery 8  │  5730.18 ms │                                  1038.30 ms │  +5.52x faster │
│ QQuery 9  │  6730.56 ms │                                  1117.41 ms │  +6.02x faster │
│ QQuery 10 │  5132.41 ms │                                   751.85 ms │  +6.83x faster │
│ QQuery 11 │  2634.39 ms │                                   663.03 ms │  +3.97x faster │
│ QQuery 12 │  4972.35 ms │                                   442.18 ms │ +11.24x faster │
│ QQuery 13 │  1724.31 ms │                                   546.11 ms │  +3.16x faster │
│ QQuery 14 │  3995.45 ms │                                   426.48 ms │  +9.37x faster │
│ QQuery 15 │  7563.01 ms │                                   669.91 ms │ +11.29x faster │
│ QQuery 16 │  1624.61 ms │                                   492.52 ms │  +3.30x faster │
│ QQuery 17 │  7974.54 ms │                                   960.10 ms │  +8.31x faster │
│ QQuery 18 │  9429.55 ms │                                  1096.39 ms │  +8.60x faster │
│ QQuery 19 │  4019.56 ms │                                   576.20 ms │  +6.98x faster │
│ QQuery 20 │  5396.18 ms │                                   801.93 ms │  +6.73x faster │
│ QQuery 21 │ 12683.18 ms │                                  1305.28 ms │  +9.72x faster │
│ QQuery 22 │  1902.97 ms │                                   520.63 ms │  +3.66x faster │
└───────────┴─────────────┴─────────────────────────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃             ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 113684.97ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  15708.77ms │
│ Average Time (HEAD)                                        │   5167.50ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │    714.03ms │
│ Queries Faster                                             │          22 │
│ Queries Slower                                             │           0 │
│ Queries with No Change                                     │           0 │
│ Queries with Failure                                       │           0 │
└────────────────────────────────────────────────────────────┴─────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃                                      HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 1  │    3906.43 / 4120.45 ±177.52 / 4414.01 ms │           521.43 / 531.16 ±6.89 / 542.37 ms │  +7.76x faster │
│ QQuery 2  │    3001.68 / 3224.24 ±126.57 / 3339.41 ms │        705.05 / 846.70 ±130.38 / 1049.08 ms │  +3.81x faster │
│ QQuery 3  │    5434.39 / 5747.47 ±213.08 / 5984.72 ms │          666.04 / 755.55 ±53.05 / 822.79 ms │  +7.61x faster │
│ QQuery 4  │    4768.16 / 5138.20 ±253.23 / 5549.77 ms │          340.53 / 445.89 ±59.46 / 514.31 ms │ +11.52x faster │
│ QQuery 5  │    5452.50 / 5682.84 ±176.59 / 5945.60 ms │        941.48 / 1008.05 ±52.50 / 1065.68 ms │  +5.64x faster │
│ QQuery 6  │    3888.98 / 4043.84 ±238.97 / 4519.51 ms │          264.50 / 304.14 ±26.52 / 333.07 ms │ +13.30x faster │
│ QQuery 7  │    5719.59 / 5937.79 ±176.15 / 6202.47 ms │         861.42 / 934.75 ±54.83 / 1031.09 ms │  +6.35x faster │
│ QQuery 8  │    5730.18 / 6011.93 ±225.02 / 6255.91 ms │       1038.30 / 1164.07 ±85.35 / 1269.95 ms │  +5.16x faster │
│ QQuery 9  │    6730.56 / 6995.00 ±216.12 / 7213.75 ms │       1117.41 / 1186.56 ±58.04 / 1268.55 ms │  +5.90x faster │
│ QQuery 10 │    5132.41 / 5663.09 ±342.32 / 6170.09 ms │          751.85 / 823.34 ±58.20 / 928.72 ms │  +6.88x faster │
│ QQuery 11 │    2634.39 / 2843.95 ±124.82 / 3017.75 ms │          663.03 / 746.30 ±66.36 / 827.94 ms │  +3.81x faster │
│ QQuery 12 │    4972.35 / 5346.21 ±218.50 / 5626.17 ms │          442.18 / 508.85 ±41.96 / 553.29 ms │ +10.51x faster │
│ QQuery 13 │    1724.31 / 1938.27 ±139.75 / 2103.69 ms │          546.11 / 591.50 ±28.88 / 620.94 ms │  +3.28x faster │
│ QQuery 14 │    3995.45 / 4237.36 ±176.07 / 4488.89 ms │          426.48 / 527.05 ±57.70 / 581.51 ms │  +8.04x faster │
│ QQuery 15 │    7563.01 / 8094.37 ±405.93 / 8461.99 ms │          669.91 / 781.76 ±66.50 / 867.13 ms │ +10.35x faster │
│ QQuery 16 │     1624.61 / 1716.91 ±87.97 / 1839.11 ms │          492.52 / 565.13 ±45.30 / 635.69 ms │  +3.04x faster │
│ QQuery 17 │    7974.54 / 8175.72 ±229.13 / 8614.58 ms │        960.10 / 1001.09 ±36.51 / 1046.01 ms │  +8.17x faster │
│ QQuery 18 │   9429.55 / 9685.04 ±308.43 / 10122.09 ms │       1096.39 / 1175.49 ±54.40 / 1265.65 ms │  +8.24x faster │
│ QQuery 19 │    4019.56 / 4327.17 ±249.30 / 4644.20 ms │          576.20 / 615.38 ±30.68 / 650.97 ms │  +7.03x faster │
│ QQuery 20 │     5396.18 / 5471.56 ±46.81 / 5523.40 ms │          801.93 / 821.88 ±11.21 / 836.19 ms │  +6.66x faster │
│ QQuery 21 │ 12683.18 / 13541.64 ±542.59 / 14245.79 ms │       1305.28 / 1388.93 ±62.56 / 1484.39 ms │  +9.75x faster │
│ QQuery 22 │    1902.97 / 2155.51 ±226.17 / 2445.87 ms │          520.63 / 581.34 ±56.30 / 658.66 ms │  +3.71x faster │
└───────────┴───────────────────────────────────────────┴─────────────────────────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃             ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 120098.56ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  17304.89ms │
│ Average Time (HEAD)                                        │   5459.03ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │    786.59ms │
│ Queries Faster                                             │          22 │
│ Queries Slower                                             │           0 │
│ Queries with No Change                                     │           0 │
│ Queries with Failure                                       │           0 │
└────────────────────────────────────────────────────────────┴─────────────┘

Resource Usage

tpch10 — base (merge-base)

Metric	Value
Wall time	605.1s
Peak memory	3.6 GiB
Avg memory	695.9 MiB
CPU user	356.4s
CPU sys	37.0s
Peak spill	0 B

tpch10 — branch

Metric	Value
Wall time	90.0s
Peak memory	4.4 GiB
Avg memory	1.3 GiB
CPU user	359.7s
CPU sys	28.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:48:05Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpcds
env:
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │  119.60 ms │                                   186.58 ms │  1.56x slower │
│ QQuery 2  │  528.02 ms │                                   612.05 ms │  1.16x slower │
│ QQuery 3  │  234.70 ms │                                   303.81 ms │  1.29x slower │
│ QQuery 4  │ 1191.25 ms │                                  1093.47 ms │ +1.09x faster │
│ QQuery 5  │  434.22 ms │                                   460.98 ms │  1.06x slower │
│ QQuery 6  │  426.52 ms │                                   487.14 ms │  1.14x slower │
│ QQuery 7  │  480.71 ms │                                   368.27 ms │ +1.31x faster │
│ QQuery 8  │  568.19 ms │                                   339.96 ms │ +1.67x faster │
│ QQuery 9  │ 1242.85 ms │                                  1395.69 ms │  1.12x slower │
│ QQuery 10 │  517.36 ms │                                   602.00 ms │  1.16x slower │
│ QQuery 11 │  854.10 ms │                                   786.50 ms │ +1.09x faster │
│ QQuery 12 │  124.79 ms │                                   193.92 ms │  1.55x slower │
│ QQuery 13 │  557.15 ms │                                   702.40 ms │  1.26x slower │
│ QQuery 14 │ 3532.53 ms │                                  3456.08 ms │     no change │
│ QQuery 15 │  385.27 ms │                                   396.11 ms │     no change │
│ QQuery 16 │  118.13 ms │                                    93.06 ms │ +1.27x faster │
│ QQuery 17 │  680.39 ms │                                   561.77 ms │ +1.21x faster │
│ QQuery 18 │  472.51 ms │                                   445.27 ms │ +1.06x faster │
│ QQuery 19 │  469.66 ms │                                   426.58 ms │ +1.10x faster │
│ QQuery 20 │  205.89 ms │                                   176.12 ms │ +1.17x faster │
│ QQuery 21 │  382.96 ms │                                   423.64 ms │  1.11x slower │
│ QQuery 22 │  422.55 ms │                                   329.95 ms │ +1.28x faster │
│ QQuery 23 │ 2175.68 ms │                                  2166.75 ms │     no change │
│ QQuery 24 │ 1880.72 ms │                                  1771.12 ms │ +1.06x faster │
│ QQuery 25 │  607.01 ms │                                   574.53 ms │ +1.06x faster │
│ QQuery 26 │  330.60 ms │                                   315.70 ms │     no change │
│ QQuery 27 │  136.94 ms │                                   124.22 ms │ +1.10x faster │
│ QQuery 28 │  364.36 ms │                                   351.38 ms │     no change │
│ QQuery 29 │  580.13 ms │                                   708.58 ms │  1.22x slower │
│ QQuery 30 │  442.56 ms │                                   463.26 ms │     no change │
│ QQuery 31 │ 1103.26 ms │                                  1023.34 ms │ +1.08x faster │
│ QQuery 32 │  348.43 ms │                                   303.98 ms │ +1.15x faster │
│ QQuery 33 │  463.98 ms │                                   513.54 ms │  1.11x slower │
│ QQuery 34 │  169.98 ms │                                   204.91 ms │  1.21x slower │
│ QQuery 35 │  520.29 ms │                                   591.99 ms │  1.14x slower │
│ QQuery 36 │   42.70 ms │                                    40.00 ms │ +1.07x faster │
│ QQuery 37 │  169.02 ms │                                   215.06 ms │  1.27x slower │
│ QQuery 38 │  422.58 ms │                                   440.61 ms │     no change │
│ QQuery 39 │ 1506.30 ms │                                  1722.42 ms │  1.14x slower │
│ QQuery 40 │  265.13 ms │                                   340.47 ms │  1.28x slower │
│ QQuery 41 │  274.81 ms │                                   106.35 ms │ +2.58x faster │
│ QQuery 42 │  126.54 ms │                                   240.85 ms │  1.90x slower │
│ QQuery 43 │   37.16 ms │                                   106.46 ms │  2.87x slower │
│ QQuery 44 │  141.03 ms │                                   138.39 ms │     no change │
│ QQuery 45 │  466.09 ms │                                   206.53 ms │ +2.26x faster │
│ QQuery 46 │  258.62 ms │                                   175.74 ms │ +1.47x faster │
│ QQuery 47 │ 1125.31 ms │                                  1059.72 ms │ +1.06x faster │
│ QQuery 48 │  409.58 ms │                                   476.24 ms │  1.16x slower │
│ QQuery 49 │  487.29 ms │                                   446.47 ms │ +1.09x faster │
│ QQuery 50 │  534.37 ms │                                   471.78 ms │ +1.13x faster │
│ QQuery 51 │  369.53 ms │                                   376.07 ms │     no change │
│ QQuery 52 │  243.03 ms │                                   281.66 ms │  1.16x slower │
│ QQuery 53 │  129.21 ms │                                   253.52 ms │  1.96x slower │
│ QQuery 54 │  444.71 ms │                                   590.38 ms │  1.33x slower │
│ QQuery 55 │  177.91 ms │                                   238.51 ms │  1.34x slower │
│ QQuery 56 │  414.12 ms │                                   499.28 ms │  1.21x slower │
│ QQuery 57 │  825.27 ms │                                  1057.51 ms │  1.28x slower │
│ QQuery 58 │ 1002.42 ms │                                  1198.51 ms │  1.20x slower │
│ QQuery 59 │  735.86 ms │                                   860.14 ms │  1.17x slower │
│ QQuery 60 │  474.05 ms │                                   550.77 ms │  1.16x slower │
│ QQuery 61 │  304.36 ms │                                   345.67 ms │  1.14x slower │
│ QQuery 62 │  415.54 ms │                                   434.88 ms │     no change │
│ QQuery 63 │  249.34 ms │                                   267.83 ms │  1.07x slower │
│ QQuery 64 │ 2535.57 ms │                                  2361.67 ms │ +1.07x faster │
│ QQuery 65 │  640.09 ms │                                   575.59 ms │ +1.11x faster │
│ QQuery 66 │  447.00 ms │                                   519.69 ms │  1.16x slower │
│ QQuery 67 │  633.77 ms │                                   538.44 ms │ +1.18x faster │
│ QQuery 68 │  241.92 ms │                                   261.66 ms │  1.08x slower │
│ QQuery 69 │  616.98 ms │                                   591.79 ms │     no change │
│ QQuery 70 │  508.06 ms │                                   516.46 ms │     no change │
│ QQuery 71 │  526.15 ms │                                   462.78 ms │ +1.14x faster │
│ QQuery 72 │ 2341.74 ms │                                  2593.59 ms │  1.11x slower │
│ QQuery 73 │  260.22 ms │                                   142.38 ms │ +1.83x faster │
│ QQuery 74 │  624.27 ms │                                   631.02 ms │     no change │
│ QQuery 75 │  771.11 ms │                                   766.70 ms │     no change │
│ QQuery 76 │  460.69 ms │                                   360.95 ms │ +1.28x faster │
│ QQuery 77 │  543.64 ms │                                   596.29 ms │  1.10x slower │
│ QQuery 78 │  507.68 ms │                                   580.55 ms │  1.14x slower │
│ QQuery 79 │  371.16 ms │                                   262.67 ms │ +1.41x faster │
│ QQuery 80 │  554.29 ms │                                   748.41 ms │  1.35x slower │
│ QQuery 81 │  433.76 ms │                                   550.52 ms │  1.27x slower │
│ QQuery 82 │  353.07 ms │                                   518.62 ms │  1.47x slower │
│ QQuery 83 │  657.72 ms │                                   620.15 ms │ +1.06x faster │
│ QQuery 84 │  396.68 ms │                                   337.17 ms │ +1.18x faster │
│ QQuery 85 │  534.31 ms │                                   583.72 ms │  1.09x slower │
│ QQuery 86 │  171.81 ms │                                   228.00 ms │  1.33x slower │
│ QQuery 87 │  451.36 ms │                                   481.38 ms │  1.07x slower │
│ QQuery 88 │  652.75 ms │                                   572.88 ms │ +1.14x faster │
│ QQuery 89 │  283.23 ms │                                   220.54 ms │ +1.28x faster │
│ QQuery 90 │  337.75 ms │                                   340.64 ms │     no change │
│ QQuery 91 │  366.55 ms │                                   464.04 ms │  1.27x slower │
│ QQuery 92 │  317.67 ms │                                   400.80 ms │  1.26x slower │
│ QQuery 93 │  431.86 ms │                                   286.59 ms │ +1.51x faster │
│ QQuery 94 │  488.04 ms │                                   299.22 ms │ +1.63x faster │
│ QQuery 95 │  790.55 ms │                                   588.61 ms │ +1.34x faster │
│ QQuery 96 │  164.63 ms │                                   272.72 ms │  1.66x slower │
│ QQuery 97 │  372.72 ms │                                   321.19 ms │ +1.16x faster │
│ QQuery 98 │  293.36 ms │                                   339.66 ms │  1.16x slower │
│ QQuery 99 │  414.66 ms │                                   400.18 ms │     no change │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 55621.96ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 56433.71ms │
│ Average Time (HEAD)                                        │   561.84ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   570.04ms │
│ Queries Faster                                             │         37 │
│ Queries Slower                                             │         46 │
│ Queries with No Change                                     │         16 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     119.60 / 165.77 ±29.83 / 206.83 ms │          186.58 / 199.32 ±10.35 / 208.96 ms │  1.20x slower │
│ QQuery 2  │    528.02 / 689.32 ±109.72 / 858.40 ms │          612.05 / 710.09 ±75.77 / 819.31 ms │     no change │
│ QQuery 3  │     234.70 / 376.57 ±94.71 / 526.25 ms │          303.81 / 341.25 ±39.79 / 416.42 ms │ +1.10x faster │
│ QQuery 4  │  1191.25 / 1231.24 ±52.88 / 1334.89 ms │       1093.47 / 1227.74 ±87.83 / 1312.60 ms │     no change │
│ QQuery 5  │    434.22 / 567.17 ±110.07 / 720.26 ms │          460.98 / 526.83 ±50.82 / 612.59 ms │ +1.08x faster │
│ QQuery 6  │    426.52 / 606.47 ±118.94 / 778.00 ms │          487.14 / 607.52 ±70.84 / 679.11 ms │     no change │
│ QQuery 7  │     480.71 / 559.70 ±55.09 / 629.52 ms │          368.27 / 451.39 ±53.31 / 514.54 ms │ +1.24x faster │
│ QQuery 8  │     568.19 / 658.97 ±67.53 / 743.83 ms │         339.96 / 543.92 ±128.65 / 716.69 ms │ +1.21x faster │
│ QQuery 9  │ 1242.85 / 1488.80 ±155.59 / 1703.69 ms │       1395.69 / 1513.52 ±73.58 / 1616.10 ms │     no change │
│ QQuery 10 │    517.36 / 666.11 ±125.98 / 876.80 ms │         602.00 / 754.13 ±110.72 / 874.48 ms │  1.13x slower │
│ QQuery 11 │    854.10 / 929.31 ±60.82 / 1012.22 ms │         786.50 / 948.77 ±93.92 / 1060.54 ms │     no change │
│ QQuery 12 │    124.79 / 303.48 ±147.21 / 520.31 ms │          193.92 / 246.54 ±57.91 / 347.05 ms │ +1.23x faster │
│ QQuery 13 │   557.15 / 818.27 ±177.57 / 1067.88 ms │          702.40 / 767.21 ±45.79 / 845.44 ms │ +1.07x faster │
│ QQuery 14 │ 3532.53 / 4014.01 ±304.80 / 4343.81 ms │      3456.08 / 3726.12 ±237.56 / 4105.98 ms │ +1.08x faster │
│ QQuery 15 │     385.27 / 451.48 ±72.80 / 588.80 ms │          396.11 / 459.83 ±41.35 / 511.99 ms │     no change │
│ QQuery 16 │     118.13 / 143.17 ±15.70 / 161.09 ms │           93.06 / 129.58 ±32.82 / 188.25 ms │ +1.10x faster │
│ QQuery 17 │     680.39 / 754.86 ±46.54 / 816.76 ms │          561.77 / 693.04 ±67.43 / 750.05 ms │ +1.09x faster │
│ QQuery 18 │     472.51 / 584.84 ±82.41 / 706.84 ms │          445.27 / 559.82 ±85.92 / 707.67 ms │     no change │
│ QQuery 19 │     469.66 / 604.12 ±98.93 / 710.01 ms │          426.58 / 470.42 ±30.30 / 510.06 ms │ +1.28x faster │
│ QQuery 20 │    205.89 / 368.32 ±151.21 / 630.05 ms │          176.12 / 305.49 ±72.04 / 389.55 ms │ +1.21x faster │
│ QQuery 21 │     382.96 / 473.66 ±51.36 / 520.70 ms │          423.64 / 491.38 ±82.56 / 651.03 ms │     no change │
│ QQuery 22 │     422.55 / 525.17 ±75.69 / 627.48 ms │         329.95 / 490.36 ±137.36 / 657.81 ms │ +1.07x faster │
│ QQuery 23 │  2175.68 / 2252.81 ±93.47 / 2430.60 ms │      2166.75 / 2351.59 ±129.00 / 2538.86 ms │     no change │
│ QQuery 24 │  1880.72 / 2063.47 ±99.22 / 2153.34 ms │      1771.12 / 2011.53 ±205.65 / 2252.38 ms │     no change │
│ QQuery 25 │    607.01 / 752.68 ±116.84 / 892.54 ms │         574.53 / 774.44 ±112.19 / 890.34 ms │     no change │
│ QQuery 26 │     330.60 / 412.18 ±74.82 / 520.14 ms │         315.70 / 429.75 ±108.28 / 625.56 ms │     no change │
│ QQuery 27 │     136.94 / 228.60 ±91.58 / 390.21 ms │          124.22 / 203.33 ±54.49 / 258.61 ms │ +1.12x faster │
│ QQuery 28 │     364.36 / 385.81 ±24.45 / 431.95 ms │          351.38 / 389.81 ±31.57 / 441.10 ms │     no change │
│ QQuery 29 │    580.13 / 765.78 ±138.11 / 988.45 ms │          708.58 / 760.72 ±55.33 / 854.81 ms │     no change │
│ QQuery 30 │     442.56 / 560.64 ±96.14 / 690.78 ms │          463.26 / 538.30 ±66.37 / 623.57 ms │     no change │
│ QQuery 31 │ 1103.26 / 1354.47 ±181.69 / 1629.19 ms │      1023.34 / 1222.04 ±190.89 / 1561.89 ms │ +1.11x faster │
│ QQuery 32 │     348.43 / 408.43 ±50.62 / 490.55 ms │         303.98 / 427.38 ±129.45 / 622.11 ms │     no change │
│ QQuery 33 │     463.98 / 549.03 ±66.89 / 631.33 ms │          513.54 / 585.57 ±92.46 / 764.40 ms │  1.07x slower │
│ QQuery 34 │     169.98 / 257.61 ±67.12 / 343.55 ms │          204.91 / 272.54 ±60.85 / 381.29 ms │  1.06x slower │
│ QQuery 35 │    520.29 / 656.70 ±112.43 / 814.53 ms │          591.99 / 729.39 ±72.36 / 792.35 ms │  1.11x slower │
│ QQuery 36 │      42.70 / 171.24 ±64.58 / 207.76 ms │            40.00 / 95.13 ±44.59 / 137.81 ms │ +1.80x faster │
│ QQuery 37 │     169.02 / 253.75 ±83.17 / 409.61 ms │          215.06 / 248.46 ±24.75 / 288.89 ms │     no change │
│ QQuery 38 │    422.58 / 568.12 ±130.79 / 789.35 ms │         440.61 / 661.66 ±120.26 / 780.86 ms │  1.16x slower │
│ QQuery 39 │ 1506.30 / 1658.69 ±150.59 / 1927.39 ms │      1722.42 / 1889.29 ±136.14 / 2025.89 ms │  1.14x slower │
│ QQuery 40 │     265.13 / 395.76 ±69.06 / 460.19 ms │          340.47 / 449.85 ±84.35 / 592.75 ms │  1.14x slower │
│ QQuery 41 │     274.81 / 332.30 ±33.87 / 376.14 ms │         106.35 / 225.01 ±131.51 / 475.38 ms │ +1.48x faster │
│ QQuery 42 │     126.54 / 237.81 ±92.10 / 349.57 ms │          240.85 / 398.41 ±92.81 / 522.87 ms │  1.68x slower │
│ QQuery 43 │      37.16 / 112.08 ±61.22 / 186.66 ms │          106.46 / 121.54 ±18.64 / 157.20 ms │  1.08x slower │
│ QQuery 44 │     141.03 / 192.82 ±54.75 / 298.06 ms │          138.39 / 257.03 ±83.53 / 362.10 ms │  1.33x slower │
│ QQuery 45 │     466.09 / 561.53 ±74.39 / 687.89 ms │         206.53 / 429.80 ±143.73 / 597.69 ms │ +1.31x faster │
│ QQuery 46 │     258.62 / 343.48 ±54.82 / 416.00 ms │          175.74 / 336.81 ±84.04 / 410.92 ms │     no change │
│ QQuery 47 │ 1125.31 / 1337.54 ±200.31 / 1665.45 ms │      1059.72 / 1302.99 ±145.24 / 1447.47 ms │     no change │
│ QQuery 48 │    409.58 / 619.33 ±112.50 / 725.69 ms │         476.24 / 642.37 ±117.68 / 838.88 ms │     no change │
│ QQuery 49 │     487.29 / 539.86 ±64.66 / 659.32 ms │          446.47 / 500.95 ±35.79 / 557.40 ms │ +1.08x faster │
│ QQuery 50 │     534.37 / 678.72 ±93.30 / 790.67 ms │        471.78 / 690.58 ±214.70 / 1079.24 ms │     no change │
│ QQuery 51 │     369.53 / 510.40 ±84.08 / 621.74 ms │         376.07 / 558.06 ±100.15 / 664.92 ms │  1.09x slower │
│ QQuery 52 │     243.03 / 356.11 ±91.65 / 454.22 ms │          281.66 / 397.55 ±90.21 / 509.83 ms │  1.12x slower │
│ QQuery 53 │    129.21 / 322.41 ±111.55 / 439.65 ms │          253.52 / 387.15 ±77.83 / 466.19 ms │  1.20x slower │
│ QQuery 54 │    444.71 / 723.32 ±162.18 / 903.55 ms │         590.38 / 799.01 ±123.07 / 977.08 ms │  1.10x slower │
│ QQuery 55 │    177.91 / 352.05 ±122.92 / 528.82 ms │         238.51 / 389.83 ±100.11 / 529.42 ms │  1.11x slower │
│ QQuery 56 │     414.12 / 572.40 ±80.40 / 630.34 ms │          499.28 / 602.56 ±90.40 / 760.16 ms │  1.05x slower │
│ QQuery 57 │  825.27 / 1211.58 ±201.63 / 1374.75 ms │      1057.51 / 1224.77 ±171.53 / 1544.09 ms │     no change │
│ QQuery 58 │ 1002.42 / 1183.66 ±140.38 / 1428.90 ms │      1198.51 / 1345.22 ±110.83 / 1474.50 ms │  1.14x slower │
│ QQuery 59 │   735.86 / 902.38 ±106.01 / 1054.04 ms │         860.14 / 960.85 ±69.97 / 1041.40 ms │  1.06x slower │
│ QQuery 60 │     474.05 / 544.25 ±73.17 / 685.38 ms │          550.77 / 598.06 ±29.47 / 632.90 ms │  1.10x slower │
│ QQuery 61 │     304.36 / 341.32 ±38.60 / 414.19 ms │          345.67 / 384.89 ±24.49 / 417.29 ms │  1.13x slower │
│ QQuery 62 │     415.54 / 503.29 ±53.30 / 581.28 ms │         434.88 / 595.89 ±159.36 / 859.73 ms │  1.18x slower │
│ QQuery 63 │     249.34 / 403.95 ±87.25 / 494.08 ms │          267.83 / 396.87 ±74.17 / 495.64 ms │     no change │
│ QQuery 64 │ 2535.57 / 2769.85 ±234.92 / 3184.92 ms │      2361.67 / 2918.36 ±284.47 / 3164.62 ms │  1.05x slower │
│ QQuery 65 │     640.09 / 772.75 ±74.65 / 856.40 ms │          575.59 / 686.64 ±59.71 / 743.84 ms │ +1.13x faster │
│ QQuery 66 │     447.00 / 623.49 ±97.83 / 716.19 ms │         519.69 / 631.57 ±100.62 / 818.85 ms │     no change │
│ QQuery 67 │     633.77 / 745.24 ±96.50 / 870.08 ms │         538.44 / 766.43 ±146.45 / 943.95 ms │     no change │
│ QQuery 68 │     241.92 / 318.55 ±55.25 / 389.86 ms │          261.66 / 315.69 ±36.02 / 372.53 ms │     no change │
│ QQuery 69 │    616.98 / 770.07 ±121.35 / 942.85 ms │        591.79 / 775.12 ±172.09 / 1078.16 ms │     no change │
│ QQuery 70 │     508.06 / 592.77 ±76.54 / 703.32 ms │         516.46 / 654.46 ±116.29 / 855.74 ms │  1.10x slower │
│ QQuery 71 │     526.15 / 568.81 ±46.37 / 632.50 ms │          462.78 / 540.69 ±58.86 / 636.04 ms │     no change │
│ QQuery 72 │ 2341.74 / 2561.95 ±162.94 / 2772.65 ms │       2593.59 / 2697.94 ±54.03 / 2742.57 ms │  1.05x slower │
│ QQuery 73 │     260.22 / 322.89 ±44.08 / 394.80 ms │          142.38 / 232.82 ±50.57 / 298.32 ms │ +1.39x faster │
│ QQuery 74 │    624.27 / 770.50 ±114.83 / 913.56 ms │         631.02 / 822.57 ±100.26 / 910.85 ms │  1.07x slower │
│ QQuery 75 │    771.11 / 915.64 ±97.70 / 1060.05 ms │        766.70 / 914.49 ±129.19 / 1152.53 ms │     no change │
│ QQuery 76 │     460.69 / 538.09 ±38.87 / 562.71 ms │          360.95 / 513.60 ±94.15 / 635.51 ms │     no change │
│ QQuery 77 │    543.64 / 649.01 ±105.76 / 834.12 ms │          596.29 / 722.36 ±86.16 / 862.87 ms │  1.11x slower │
│ QQuery 78 │     507.68 / 647.64 ±88.16 / 768.92 ms │          580.55 / 673.19 ±65.51 / 779.30 ms │     no change │
│ QQuery 79 │     371.16 / 467.84 ±92.26 / 637.01 ms │         262.67 / 500.39 ±128.29 / 632.02 ms │  1.07x slower │
│ QQuery 80 │    554.29 / 743.62 ±107.88 / 885.81 ms │          748.41 / 789.84 ±28.29 / 822.33 ms │  1.06x slower │
│ QQuery 81 │    433.76 / 545.15 ±105.71 / 728.44 ms │          550.52 / 587.77 ±53.61 / 693.25 ms │  1.08x slower │
│ QQuery 82 │     353.07 / 504.56 ±92.67 / 644.33 ms │          518.62 / 580.88 ±47.30 / 645.30 ms │  1.15x slower │
│ QQuery 83 │     657.72 / 751.25 ±59.46 / 811.00 ms │         620.15 / 766.98 ±122.64 / 933.47 ms │     no change │
│ QQuery 84 │     396.68 / 428.97 ±42.62 / 511.36 ms │         337.17 / 537.25 ±122.76 / 669.66 ms │  1.25x slower │
│ QQuery 85 │     534.31 / 596.01 ±76.27 / 741.46 ms │         583.72 / 782.81 ±121.64 / 911.84 ms │  1.31x slower │
│ QQuery 86 │    171.81 / 339.09 ±149.43 / 563.62 ms │          228.00 / 325.49 ±76.77 / 442.84 ms │     no change │
│ QQuery 87 │    451.36 / 578.11 ±116.22 / 765.02 ms │          481.38 / 579.15 ±92.29 / 742.67 ms │     no change │
│ QQuery 88 │     652.75 / 698.80 ±29.38 / 745.02 ms │          572.88 / 645.34 ±56.83 / 735.26 ms │ +1.08x faster │
│ QQuery 89 │     283.23 / 434.12 ±89.44 / 516.30 ms │         220.54 / 384.18 ±131.81 / 574.98 ms │ +1.13x faster │
│ QQuery 90 │     337.75 / 461.22 ±72.78 / 551.28 ms │          340.64 / 455.50 ±74.48 / 575.22 ms │     no change │
│ QQuery 91 │    366.55 / 651.30 ±180.59 / 872.95 ms │         464.04 / 635.33 ±117.10 / 792.17 ms │     no change │
│ QQuery 92 │     317.67 / 395.69 ±58.76 / 499.84 ms │          400.80 / 462.28 ±36.85 / 504.69 ms │  1.17x slower │
│ QQuery 93 │    431.86 / 552.53 ±125.46 / 789.41 ms │          286.59 / 448.96 ±90.05 / 544.35 ms │ +1.23x faster │
│ QQuery 94 │     488.04 / 555.58 ±43.86 / 620.26 ms │         299.22 / 520.43 ±160.43 / 719.37 ms │ +1.07x faster │
│ QQuery 95 │     790.55 / 870.95 ±65.41 / 952.72 ms │         588.61 / 744.69 ±136.33 / 967.15 ms │ +1.17x faster │
│ QQuery 96 │    164.63 / 309.22 ±108.71 / 438.76 ms │          272.72 / 410.02 ±97.98 / 552.88 ms │  1.33x slower │
│ QQuery 97 │     372.72 / 467.20 ±71.22 / 550.11 ms │         321.19 / 539.47 ±148.03 / 700.01 ms │  1.15x slower │
│ QQuery 98 │     293.36 / 333.83 ±29.47 / 369.98 ms │          339.66 / 462.60 ±69.65 / 548.53 ms │  1.39x slower │
│ QQuery 99 │    414.66 / 582.51 ±198.98 / 899.98 ms │         400.18 / 565.79 ±130.53 / 792.07 ms │     no change │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 68393.98ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 69342.33ms │
│ Average Time (HEAD)                                        │   690.85ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   700.43ms │
│ Queries Faster                                             │         25 │
│ Queries Slower                                             │         37 │
│ Queries with No Change                                     │         37 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	350.1s
Peak memory	1.1 GiB
Avg memory	603.7 MiB
CPU user	225.7s
CPU sys	11.8s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	355.1s
Peak memory	1.0 GiB
Avg memory	584.8 MiB
CPU user	231.8s
CPU sys	11.9s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:48:48Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.24 ms │                                     1.23 ms │     no change │
│ QQuery 1  │ 1788.02 ms │                                   719.58 ms │ +2.48x faster │
│ QQuery 2  │ 2201.73 ms │                                   696.26 ms │ +3.16x faster │
│ QQuery 3  │ 2078.53 ms │                                   793.87 ms │ +2.62x faster │
│ QQuery 4  │ 2191.27 ms │                                   925.60 ms │ +2.37x faster │
│ QQuery 5  │ 2327.95 ms │                                   944.68 ms │ +2.46x faster │
│ QQuery 6  │    1.30 ms │                                     1.29 ms │     no change │
│ QQuery 7  │ 1805.68 ms │                                   652.78 ms │ +2.77x faster │
│ QQuery 8  │ 2321.44 ms │                                  1006.78 ms │ +2.31x faster │
│ QQuery 9  │ 2637.19 ms │                                  1207.11 ms │ +2.18x faster │
│ QQuery 10 │ 2302.19 ms │                                   753.06 ms │ +3.06x faster │
│ QQuery 11 │ 2098.87 ms │                                   760.45 ms │ +2.76x faster │
│ QQuery 12 │ 2302.08 ms │                                   884.48 ms │ +2.60x faster │
│ QQuery 13 │ 2346.50 ms │                                   983.42 ms │ +2.39x faster │
│ QQuery 14 │ 2328.10 ms │                                   924.06 ms │ +2.52x faster │
│ QQuery 15 │ 2272.97 ms │                                   967.62 ms │ +2.35x faster │
│ QQuery 16 │ 2641.15 ms │                                  1304.45 ms │ +2.02x faster │
│ QQuery 17 │ 2501.83 ms │                                  1231.76 ms │ +2.03x faster │
│ QQuery 18 │ 3123.23 ms │                                  1823.97 ms │ +1.71x faster │
│ QQuery 19 │ 1628.20 ms │                                   680.82 ms │ +2.39x faster │
│ QQuery 20 │ 2540.79 ms │                                  1147.82 ms │ +2.21x faster │
│ QQuery 21 │ 2483.35 ms │                                  1268.61 ms │ +1.96x faster │
│ QQuery 22 │ 3152.23 ms │                                  1698.94 ms │ +1.86x faster │
│ QQuery 23 │ 4930.69 ms │                                  4718.62 ms │     no change │
│ QQuery 24 │  469.49 ms │                                   291.72 ms │ +1.61x faster │
│ QQuery 25 │ 2185.44 ms │                                   880.74 ms │ +2.48x faster │
│ QQuery 26 │  470.83 ms │                                   304.71 ms │ +1.55x faster │
│ QQuery 27 │ 2655.20 ms │                                  1204.99 ms │ +2.20x faster │
│ QQuery 28 │ 5000.39 ms │                                  3622.91 ms │ +1.38x faster │
│ QQuery 29 │ 2126.75 ms │                                   764.36 ms │ +2.78x faster │
│ QQuery 30 │ 2327.82 ms │                                   926.05 ms │ +2.51x faster │
│ QQuery 31 │ 2229.60 ms │                                   939.81 ms │ +2.37x faster │
│ QQuery 32 │ 2761.24 ms │                                  1390.95 ms │ +1.99x faster │
│ QQuery 33 │ 3571.38 ms │                                  2311.61 ms │ +1.54x faster │
│ QQuery 34 │ 3575.04 ms │                                  2291.03 ms │ +1.56x faster │
│ QQuery 35 │ 2283.57 ms │                                   979.76 ms │ +2.33x faster │
│ QQuery 36 │  129.89 ms │                                   127.81 ms │     no change │
│ QQuery 37 │   76.74 ms │                                    69.41 ms │ +1.11x faster │
│ QQuery 38 │  109.88 ms │                                    75.16 ms │ +1.46x faster │
│ QQuery 39 │  183.16 ms │                                   215.02 ms │  1.17x slower │
│ QQuery 40 │   74.30 ms │                                    49.05 ms │ +1.51x faster │
│ QQuery 41 │   68.70 ms │                                    46.98 ms │ +1.46x faster │
│ QQuery 42 │  142.40 ms │                                    51.44 ms │ +2.77x faster │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 84448.35ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 42640.79ms │
│ Average Time (HEAD)                                        │  1963.92ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   991.65ms │
│ Queries Faster                                             │         38 │
│ Queries Slower                                             │          1 │
│ Queries with No Change                                     │          4 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.24 / 4.14 ±5.64 / 15.43 ms │                1.23 / 4.10 ±5.58 / 15.26 ms │     no change │
│ QQuery 1  │  1788.02 / 1907.55 ±62.29 / 1956.53 ms │          719.58 / 763.46 ±33.15 / 807.12 ms │ +2.50x faster │
│ QQuery 2  │  2201.73 / 2246.17 ±29.44 / 2282.36 ms │          696.26 / 776.72 ±47.23 / 824.53 ms │ +2.89x faster │
│ QQuery 3  │  2078.53 / 2177.38 ±90.03 / 2340.76 ms │          793.87 / 836.02 ±31.40 / 890.08 ms │ +2.60x faster │
│ QQuery 4  │  2191.27 / 2303.44 ±81.42 / 2389.71 ms │         925.60 / 974.55 ±41.34 / 1050.82 ms │ +2.36x faster │
│ QQuery 5  │ 2327.95 / 2512.66 ±116.65 / 2684.02 ms │         944.68 / 986.73 ±22.73 / 1006.39 ms │ +2.55x faster │
│ QQuery 6  │            1.30 / 1.46 ±0.24 / 1.94 ms │                 1.29 / 1.47 ±0.24 / 1.94 ms │     no change │
│ QQuery 7  │  1805.68 / 1919.20 ±77.79 / 2043.00 ms │          652.78 / 751.62 ±83.66 / 862.51 ms │ +2.55x faster │
│ QQuery 8  │ 2321.44 / 2483.84 ±117.40 / 2662.07 ms │       1006.78 / 1064.50 ±35.80 / 1100.31 ms │ +2.33x faster │
│ QQuery 9  │  2637.19 / 2758.31 ±90.90 / 2916.84 ms │       1207.11 / 1242.50 ±38.22 / 1313.24 ms │ +2.22x faster │
│ QQuery 10 │  2302.19 / 2396.40 ±64.56 / 2490.54 ms │          753.06 / 819.20 ±37.29 / 867.82 ms │ +2.93x faster │
│ QQuery 11 │  2098.87 / 2205.39 ±62.83 / 2275.79 ms │          760.45 / 803.60 ±26.83 / 842.70 ms │ +2.74x faster │
│ QQuery 12 │ 2302.08 / 2441.25 ±127.59 / 2627.65 ms │         884.48 / 978.82 ±79.91 / 1085.80 ms │ +2.49x faster │
│ QQuery 13 │  2346.50 / 2472.02 ±76.72 / 2574.67 ms │        983.42 / 1122.26 ±71.77 / 1178.08 ms │ +2.20x faster │
│ QQuery 14 │  2328.10 / 2397.25 ±39.19 / 2447.97 ms │         924.06 / 990.19 ±46.04 / 1065.42 ms │ +2.42x faster │
│ QQuery 15 │  2272.97 / 2384.84 ±79.69 / 2466.32 ms │        967.62 / 1039.60 ±62.27 / 1129.06 ms │ +2.29x faster │
│ QQuery 16 │  2641.15 / 2736.71 ±71.50 / 2827.78 ms │       1304.45 / 1341.08 ±21.12 / 1364.41 ms │ +2.04x faster │
│ QQuery 17 │ 2501.83 / 2671.19 ±133.54 / 2894.10 ms │       1231.76 / 1315.92 ±63.71 / 1427.27 ms │ +2.03x faster │
│ QQuery 18 │  3123.23 / 3210.10 ±84.13 / 3317.97 ms │       1823.97 / 1930.16 ±74.22 / 2054.71 ms │ +1.66x faster │
│ QQuery 19 │ 1628.20 / 1812.70 ±123.44 / 1972.04 ms │          680.82 / 731.01 ±55.39 / 812.77 ms │ +2.48x faster │
│ QQuery 20 │  2540.79 / 2641.94 ±57.84 / 2709.99 ms │       1147.82 / 1262.95 ±62.20 / 1332.77 ms │ +2.09x faster │
│ QQuery 21 │ 2483.35 / 2727.93 ±147.13 / 2944.81 ms │       1268.61 / 1320.41 ±44.46 / 1373.13 ms │ +2.07x faster │
│ QQuery 22 │ 3152.23 / 3311.68 ±114.31 / 3418.48 ms │      1698.94 / 1894.42 ±143.44 / 2103.36 ms │ +1.75x faster │
│ QQuery 23 │ 4930.69 / 5519.42 ±298.58 / 5726.27 ms │      4718.62 / 5044.42 ±267.63 / 5375.18 ms │ +1.09x faster │
│ QQuery 24 │     469.49 / 507.74 ±32.93 / 562.30 ms │          291.72 / 351.27 ±47.76 / 414.53 ms │ +1.45x faster │
│ QQuery 25 │ 2185.44 / 2298.17 ±119.79 / 2498.06 ms │          880.74 / 914.26 ±21.07 / 943.69 ms │ +2.51x faster │
│ QQuery 26 │     470.83 / 506.18 ±40.88 / 585.79 ms │          304.71 / 355.66 ±41.49 / 409.14 ms │ +1.42x faster │
│ QQuery 27 │ 2655.20 / 2767.89 ±101.52 / 2940.84 ms │       1204.99 / 1286.96 ±73.82 / 1405.13 ms │ +2.15x faster │
│ QQuery 28 │ 5000.39 / 5133.82 ±119.06 / 5345.74 ms │      3622.91 / 3725.94 ±111.71 / 3886.55 ms │ +1.38x faster │
│ QQuery 29 │  2126.75 / 2223.22 ±64.79 / 2312.56 ms │          764.36 / 784.60 ±12.89 / 796.55 ms │ +2.83x faster │
│ QQuery 30 │ 2327.82 / 2483.57 ±136.82 / 2645.73 ms │        926.05 / 1063.67 ±78.18 / 1168.23 ms │ +2.33x faster │
│ QQuery 31 │  2229.60 / 2393.40 ±92.32 / 2513.04 ms │         939.81 / 977.17 ±36.02 / 1037.60 ms │ +2.45x faster │
│ QQuery 32 │  2761.24 / 2839.38 ±95.11 / 3024.39 ms │      1390.95 / 1568.03 ±125.84 / 1740.09 ms │ +1.81x faster │
│ QQuery 33 │  3571.38 / 3614.23 ±35.05 / 3654.87 ms │       2311.61 / 2432.01 ±64.78 / 2495.24 ms │ +1.49x faster │
│ QQuery 34 │  3575.04 / 3628.29 ±48.28 / 3702.97 ms │       2291.03 / 2356.30 ±58.18 / 2463.46 ms │ +1.54x faster │
│ QQuery 35 │ 2283.57 / 2459.59 ±139.03 / 2689.07 ms │        979.76 / 1052.35 ±47.56 / 1121.08 ms │ +2.34x faster │
│ QQuery 36 │    129.89 / 276.86 ±105.66 / 455.49 ms │          127.81 / 207.51 ±58.35 / 287.06 ms │ +1.33x faster │
│ QQuery 37 │      76.74 / 154.79 ±62.70 / 269.21 ms │           69.41 / 163.98 ±71.67 / 257.64 ms │  1.06x slower │
│ QQuery 38 │     109.88 / 215.72 ±60.36 / 284.55 ms │           75.16 / 176.04 ±59.11 / 249.95 ms │ +1.23x faster │
│ QQuery 39 │     183.16 / 253.19 ±64.93 / 367.52 ms │          215.02 / 271.14 ±31.03 / 298.76 ms │  1.07x slower │
│ QQuery 40 │     74.30 / 189.30 ±132.77 / 428.91 ms │           49.05 / 190.48 ±70.92 / 231.68 ms │     no change │
│ QQuery 41 │      68.70 / 164.64 ±53.78 / 215.76 ms │           46.98 / 163.79 ±64.14 / 215.44 ms │     no change │
│ QQuery 42 │     142.40 / 202.99 ±32.28 / 227.51 ms │           51.44 / 119.16 ±50.23 / 198.02 ms │ +1.70x faster │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 89555.91ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 46156.04ms │
│ Average Time (HEAD)                                        │  2082.70ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  1073.40ms │
│ Queries Faster                                             │         37 │
│ Queries Slower                                             │          2 │
│ Queries with No Change                                     │          4 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	450.1s
Peak memory	16.4 GiB
Avg memory	2.5 GiB
CPU user	993.6s
CPU sys	148.6s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	235.1s
Peak memory	17.0 GiB
Avg memory	3.5 GiB
CPU user	1012.5s
CPU sys	133.0s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:50:11Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: "true"
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.27 ms │                                     1.22 ms │      no change │
│ QQuery 1  │ 1785.55 ms │                                   759.01 ms │  +2.35x faster │
│ QQuery 2  │ 1970.92 ms │                                   732.90 ms │  +2.69x faster │
│ QQuery 3  │ 2049.12 ms │                                   732.66 ms │  +2.80x faster │
│ QQuery 4  │ 2307.88 ms │                                   860.87 ms │  +2.68x faster │
│ QQuery 5  │ 2235.77 ms │                                   989.65 ms │  +2.26x faster │
│ QQuery 6  │    1.30 ms │                                     1.29 ms │      no change │
│ QQuery 7  │ 1858.08 ms │                                   648.51 ms │  +2.87x faster │
│ QQuery 8  │ 2474.48 ms │                                   986.26 ms │  +2.51x faster │
│ QQuery 9  │ 2426.87 ms │                                  1269.46 ms │  +1.91x faster │
│ QQuery 10 │ 4360.80 ms │                                   754.84 ms │  +5.78x faster │
│ QQuery 11 │ 4173.81 ms │                                   787.09 ms │  +5.30x faster │
│ QQuery 12 │ 2258.32 ms │                                  1116.24 ms │  +2.02x faster │
│ QQuery 13 │ 4507.09 ms │                                  1076.54 ms │  +4.19x faster │
│ QQuery 14 │ 4426.07 ms │                                  1033.44 ms │  +4.28x faster │
│ QQuery 15 │ 2390.88 ms │                                   938.88 ms │  +2.55x faster │
│ QQuery 16 │ 2464.59 ms │                                  1227.61 ms │  +2.01x faster │
│ QQuery 17 │ 2641.59 ms │                                  1295.27 ms │  +2.04x faster │
│ QQuery 18 │ 3078.39 ms │                                  1875.60 ms │  +1.64x faster │
│ QQuery 19 │ 1712.09 ms │                                   681.82 ms │  +2.51x faster │
│ QQuery 20 │ 2664.01 ms │                                  1195.01 ms │  +2.23x faster │
│ QQuery 21 │ 4873.24 ms │                                  1302.45 ms │  +3.74x faster │
│ QQuery 22 │ 8021.01 ms │                                  1800.83 ms │  +4.45x faster │
│ QQuery 23 │ 1058.51 ms │                                   551.35 ms │  +1.92x faster │
│ QQuery 24 │  659.81 ms │                                   255.45 ms │  +2.58x faster │
│ QQuery 25 │ 2203.68 ms │                                   886.19 ms │  +2.49x faster │
│ QQuery 26 │  784.50 ms │                                   270.73 ms │  +2.90x faster │
│ QQuery 27 │ 4176.65 ms │                                  1496.15 ms │  +2.79x faster │
│ QQuery 28 │ 4926.80 ms │                                  3874.84 ms │  +1.27x faster │
│ QQuery 29 │ 2212.40 ms │                                   676.64 ms │  +3.27x faster │
│ QQuery 30 │ 4167.61 ms │                                  1026.77 ms │  +4.06x faster │
│ QQuery 31 │ 4379.66 ms │                                   897.22 ms │  +4.88x faster │
│ QQuery 32 │ 2569.44 ms │                                  1144.95 ms │  +2.24x faster │
│ QQuery 33 │ 3396.85 ms │                                  2072.22 ms │  +1.64x faster │
│ QQuery 34 │ 3399.00 ms │                                  2142.79 ms │  +1.59x faster │
│ QQuery 35 │ 2313.45 ms │                                   958.31 ms │  +2.41x faster │
│ QQuery 36 │  549.66 ms │                                   134.47 ms │  +4.09x faster │
│ QQuery 37 │  790.19 ms │                                    91.50 ms │  +8.64x faster │
│ QQuery 38 │  569.27 ms │                                    73.08 ms │  +7.79x faster │
│ QQuery 39 │  729.34 ms │                                   181.24 ms │  +4.02x faster │
│ QQuery 40 │  627.17 ms │                                   132.11 ms │  +4.75x faster │
│ QQuery 41 │  471.42 ms │                                   113.29 ms │  +4.16x faster │
│ QQuery 42 │  585.71 ms │                                    42.94 ms │ +13.64x faster │
└───────────┴────────────┴─────────────────────────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃             ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 105254.23ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  39089.70ms │
│ Average Time (HEAD)                                        │   2447.77ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │    909.06ms │
│ Queries Faster                                             │          41 │
│ Queries Slower                                             │           0 │
│ Queries with No Change                                     │           2 │
│ Queries with Failure                                       │           0 │
└────────────────────────────────────────────────────────────┴─────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.27 / 4.22 ±5.72 / 15.65 ms │                1.22 / 4.14 ±5.68 / 15.49 ms │     no change │
│ QQuery 1  │  1785.55 / 1965.41 ±95.10 / 2058.81 ms │          759.01 / 783.14 ±19.68 / 812.01 ms │ +2.51x faster │
│ QQuery 2  │ 1970.92 / 2198.81 ±170.27 / 2494.95 ms │          732.90 / 812.56 ±43.43 / 856.55 ms │ +2.71x faster │
│ QQuery 3  │  2049.12 / 2197.76 ±84.83 / 2303.11 ms │          732.66 / 805.42 ±56.25 / 867.69 ms │ +2.73x faster │
│ QQuery 4  │  2307.88 / 2397.94 ±79.92 / 2520.94 ms │         860.87 / 939.94 ±65.68 / 1017.61 ms │ +2.55x faster │
│ QQuery 5  │  2235.77 / 2347.41 ±76.91 / 2464.48 ms │        989.65 / 1015.61 ±29.07 / 1064.48 ms │ +2.31x faster │
│ QQuery 6  │            1.30 / 1.49 ±0.25 / 1.99 ms │                 1.29 / 1.44 ±0.23 / 1.90 ms │     no change │
│ QQuery 7  │  1858.08 / 1898.96 ±31.10 / 1948.81 ms │          648.51 / 726.68 ±53.89 / 793.72 ms │ +2.61x faster │
│ QQuery 8  │  2474.48 / 2559.04 ±65.86 / 2653.10 ms │        986.26 / 1049.17 ±41.28 / 1104.66 ms │ +2.44x faster │
│ QQuery 9  │ 2426.87 / 2653.19 ±115.48 / 2748.48 ms │       1269.46 / 1298.16 ±17.25 / 1322.26 ms │ +2.04x faster │
│ QQuery 10 │  4360.80 / 4455.22 ±95.21 / 4621.32 ms │          754.84 / 803.51 ±39.13 / 867.17 ms │ +5.54x faster │
│ QQuery 11 │ 4173.81 / 4492.82 ±186.31 / 4741.40 ms │          787.09 / 832.93 ±37.29 / 899.94 ms │ +5.39x faster │
│ QQuery 12 │  2258.32 / 2371.75 ±67.21 / 2460.44 ms │       1116.24 / 1145.26 ±24.44 / 1178.37 ms │ +2.07x faster │
│ QQuery 13 │ 4507.09 / 4653.65 ±109.38 / 4815.62 ms │       1076.54 / 1173.68 ±64.28 / 1250.80 ms │ +3.97x faster │
│ QQuery 14 │ 4426.07 / 4577.57 ±159.81 / 4873.18 ms │       1033.44 / 1102.35 ±35.60 / 1130.92 ms │ +4.15x faster │
│ QQuery 15 │  2390.88 / 2464.16 ±66.03 / 2558.55 ms │         938.88 / 983.84 ±28.74 / 1021.51 ms │ +2.50x faster │
│ QQuery 16 │ 2464.59 / 2701.19 ±138.93 / 2857.46 ms │       1227.61 / 1318.81 ±61.60 / 1392.34 ms │ +2.05x faster │
│ QQuery 17 │  2641.59 / 2791.00 ±96.04 / 2925.88 ms │       1295.27 / 1340.45 ±48.78 / 1429.92 ms │ +2.08x faster │
│ QQuery 18 │ 3078.39 / 3203.08 ±104.98 / 3341.06 ms │       1875.60 / 1931.77 ±37.91 / 1986.72 ms │ +1.66x faster │
│ QQuery 19 │  1712.09 / 1782.47 ±58.10 / 1862.76 ms │          681.82 / 742.95 ±42.19 / 800.93 ms │ +2.40x faster │
│ QQuery 20 │  2664.01 / 2772.19 ±63.00 / 2840.11 ms │       1195.01 / 1335.99 ±75.96 / 1406.79 ms │ +2.08x faster │
│ QQuery 21 │  4873.24 / 4906.72 ±35.38 / 4971.90 ms │       1302.45 / 1392.49 ±46.90 / 1431.51 ms │ +3.52x faster │
│ QQuery 22 │ 8021.01 / 8347.92 ±218.58 / 8634.76 ms │       1800.83 / 1848.25 ±44.19 / 1918.78 ms │ +4.52x faster │
│ QQuery 23 │ 1058.51 / 1272.20 ±113.73 / 1385.31 ms │          551.35 / 652.21 ±65.48 / 724.47 ms │ +1.95x faster │
│ QQuery 24 │   659.81 / 834.68 ±138.76 / 1013.88 ms │          255.45 / 349.39 ±53.44 / 417.94 ms │ +2.39x faster │
│ QQuery 25 │ 2203.68 / 2347.57 ±118.24 / 2522.18 ms │          886.19 / 936.44 ±35.26 / 991.67 ms │ +2.51x faster │
│ QQuery 26 │     784.50 / 888.11 ±61.63 / 975.61 ms │          270.73 / 321.89 ±38.81 / 384.31 ms │ +2.76x faster │
│ QQuery 27 │  4176.65 / 4274.87 ±83.27 / 4407.53 ms │       1496.15 / 1551.28 ±62.20 / 1661.75 ms │ +2.76x faster │
│ QQuery 28 │ 4926.80 / 5094.33 ±142.73 / 5269.30 ms │       3874.84 / 3979.77 ±68.06 / 4065.36 ms │ +1.28x faster │
│ QQuery 29 │  2212.40 / 2297.35 ±52.01 / 2372.14 ms │          676.64 / 788.07 ±67.78 / 882.52 ms │ +2.92x faster │
│ QQuery 30 │ 4167.61 / 4509.64 ±217.42 / 4835.23 ms │       1026.77 / 1081.90 ±63.78 / 1189.09 ms │ +4.17x faster │
│ QQuery 31 │ 4379.66 / 4589.27 ±120.85 / 4756.96 ms │         897.22 / 989.46 ±53.36 / 1054.33 ms │ +4.64x faster │
│ QQuery 32 │  2569.44 / 2615.27 ±50.99 / 2710.97 ms │       1144.95 / 1253.99 ±59.38 / 1306.19 ms │ +2.09x faster │
│ QQuery 33 │  3396.85 / 3500.93 ±74.37 / 3589.76 ms │       2072.22 / 2160.57 ±51.93 / 2229.72 ms │ +1.62x faster │
│ QQuery 34 │  3399.00 / 3483.48 ±73.60 / 3618.07 ms │       2142.79 / 2217.75 ±38.77 / 2249.75 ms │ +1.57x faster │
│ QQuery 35 │  2313.45 / 2374.49 ±40.25 / 2421.68 ms │        958.31 / 1017.85 ±43.25 / 1075.86 ms │ +2.33x faster │
│ QQuery 36 │   549.66 / 830.33 ±224.38 / 1136.24 ms │          134.47 / 193.64 ±57.56 / 268.76 ms │ +4.29x faster │
│ QQuery 37 │   790.19 / 986.48 ±185.78 / 1297.74 ms │           91.50 / 170.21 ±67.35 / 249.87 ms │ +5.80x faster │
│ QQuery 38 │   569.27 / 798.38 ±171.11 / 1022.81 ms │           73.08 / 107.98 ±41.26 / 167.22 ms │ +7.39x faster │
│ QQuery 39 │     729.34 / 764.74 ±37.88 / 834.46 ms │          181.24 / 220.15 ±31.70 / 251.02 ms │ +3.47x faster │
│ QQuery 40 │   627.17 / 882.24 ±237.36 / 1270.54 ms │          132.11 / 174.53 ±27.88 / 216.08 ms │ +5.05x faster │
│ QQuery 41 │    471.42 / 686.73 ±171.30 / 986.37 ms │          113.29 / 166.38 ±45.91 / 219.89 ms │ +4.13x faster │
│ QQuery 42 │   585.71 / 790.66 ±160.63 / 1071.36 ms │           42.94 / 124.63 ±60.61 / 196.23 ms │ +6.34x faster │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃             ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 111565.72ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  41846.60ms │
│ Average Time (HEAD)                                        │   2594.55ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │    973.18ms │
│ Queries Faster                                             │          41 │
│ Queries Slower                                             │           0 │
│ Queries with No Change                                     │           2 │
│ Queries with Failure                                       │           0 │
└────────────────────────────────────────────────────────────┴─────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	560.1s
Peak memory	9.5 GiB
Avg memory	1.7 GiB
CPU user	851.4s
CPU sys	118.1s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	215.0s
Peak memory	11.0 GiB
Avg memory	2.6 GiB
CPU user	923.6s
CPU sys	82.8s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-29T19:53:13Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: "true"
  SIMULATE_LATENCY: "true"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │  105.71 ms │                                   136.23 ms │  1.29x slower │
│ QQuery 2  │ 1041.67 ms │                                   699.84 ms │ +1.49x faster │
│ QQuery 3  │  494.03 ms │                                   349.27 ms │ +1.41x faster │
│ QQuery 4  │ 1780.95 ms │                                  1238.02 ms │ +1.44x faster │
│ QQuery 5  │  750.89 ms │                                   530.55 ms │ +1.42x faster │
│ QQuery 6  │ 1046.28 ms │                                   445.56 ms │ +2.35x faster │
│ QQuery 7  │  897.02 ms │                                   453.11 ms │ +1.98x faster │
│ QQuery 8  │  595.80 ms │                                   303.15 ms │ +1.97x faster │
│ QQuery 9  │ 1149.36 ms │                                  1371.29 ms │  1.19x slower │
│ QQuery 10 │ 1039.61 ms │                                   812.63 ms │ +1.28x faster │
│ QQuery 11 │ 1334.98 ms │                                   767.05 ms │ +1.74x faster │
│ QQuery 12 │  553.64 ms │                                   215.17 ms │ +2.57x faster │
│ QQuery 13 │ 1490.36 ms │                                   657.78 ms │ +2.27x faster │
│ QQuery 14 │ 6815.17 ms │                                  4230.79 ms │ +1.61x faster │
│ QQuery 15 │  561.23 ms │                                   375.50 ms │ +1.49x faster │
│ QQuery 16 │  140.45 ms │                                   116.69 ms │ +1.20x faster │
│ QQuery 17 │ 1358.91 ms │                                   881.42 ms │ +1.54x faster │
│ QQuery 18 │ 1263.76 ms │                                   470.35 ms │ +2.69x faster │
│ QQuery 19 │  887.50 ms │                                   520.83 ms │ +1.70x faster │
│ QQuery 20 │  522.43 ms │                                   318.59 ms │ +1.64x faster │
│ QQuery 21 │  836.69 ms │                                   392.71 ms │ +2.13x faster │
│ QQuery 22 │  639.90 ms │                                   265.29 ms │ +2.41x faster │
│ QQuery 23 │ 2604.49 ms │                                  2456.48 ms │ +1.06x faster │
│ QQuery 24 │ 2501.39 ms │                                  1905.85 ms │ +1.31x faster │
│ QQuery 25 │ 1391.67 ms │                                   740.44 ms │ +1.88x faster │
│ QQuery 26 │  777.27 ms │                                   347.51 ms │ +2.24x faster │
│ QQuery 27 │  189.34 ms │                                   218.62 ms │  1.15x slower │
│ QQuery 28 │  476.11 ms │                                   338.97 ms │ +1.40x faster │
│ QQuery 29 │ 1519.39 ms │                                   787.36 ms │ +1.93x faster │
│ QQuery 30 │  686.32 ms │                                   539.67 ms │ +1.27x faster │
│ QQuery 31 │ 2959.74 ms │                                  1435.78 ms │ +2.06x faster │
│ QQuery 32 │  699.37 ms │                                   371.63 ms │ +1.88x faster │
│ QQuery 33 │ 1471.40 ms │                                   617.30 ms │ +2.38x faster │
│ QQuery 34 │  342.28 ms │                                   207.34 ms │ +1.65x faster │
│ QQuery 35 │ 1476.14 ms │                                   749.06 ms │ +1.97x faster │
│ QQuery 36 │   77.62 ms │                                    90.92 ms │  1.17x slower │
│ QQuery 37 │  318.19 ms │                                   173.92 ms │ +1.83x faster │
│ QQuery 38 │ 1145.97 ms │                                   615.90 ms │ +1.86x faster │
│ QQuery 39 │ 3327.85 ms │                                  1887.16 ms │ +1.76x faster │
│ QQuery 40 │  474.78 ms │                                   426.93 ms │ +1.11x faster │
│ QQuery 41 │  262.42 ms │                                   219.55 ms │ +1.20x faster │
│ QQuery 42 │  731.53 ms │                                   230.55 ms │ +3.17x faster │
│ QQuery 43 │   91.46 ms │                                    40.75 ms │ +2.24x faster │
│ QQuery 44 │  143.67 ms │                                   188.49 ms │  1.31x slower │
│ QQuery 45 │  945.53 ms │                                   302.51 ms │ +3.13x faster │
│ QQuery 46 │  570.62 ms │                                   238.86 ms │ +2.39x faster │
│ QQuery 47 │ 1931.62 ms │                                  1403.91 ms │ +1.38x faster │
│ QQuery 48 │ 1487.46 ms │                                   708.70 ms │ +2.10x faster │
│ QQuery 49 │  989.47 ms │                                   491.07 ms │ +2.01x faster │
│ QQuery 50 │  768.53 ms │                                   614.69 ms │ +1.25x faster │
│ QQuery 51 │  827.66 ms │                                   357.14 ms │ +2.32x faster │
│ QQuery 52 │  521.21 ms │                                   279.29 ms │ +1.87x faster │
│ QQuery 53 │  776.74 ms │                                   336.91 ms │ +2.31x faster │
│ QQuery 54 │ 1224.19 ms │                                   608.98 ms │ +2.01x faster │
│ QQuery 55 │  725.44 ms │                                   385.06 ms │ +1.88x faster │
│ QQuery 56 │  999.22 ms │                                   466.64 ms │ +2.14x faster │
│ QQuery 57 │ 2061.97 ms │                                  1056.39 ms │ +1.95x faster │
│ QQuery 58 │ 1876.56 ms │                                  1122.43 ms │ +1.67x faster │
│ QQuery 59 │ 1084.83 ms │                                   703.28 ms │ +1.54x faster │
│ QQuery 60 │  866.79 ms │                                   551.03 ms │ +1.57x faster │
│ QQuery 61 │  444.72 ms │                                   325.02 ms │ +1.37x faster │
│ QQuery 62 │  559.14 ms │                                   404.22 ms │ +1.38x faster │
│ QQuery 63 │  713.57 ms │                                   331.89 ms │ +2.15x faster │
│ QQuery 64 │ 4889.11 ms │                                  2709.97 ms │ +1.80x faster │
│ QQuery 65 │ 1310.88 ms │                                   805.36 ms │ +1.63x faster │
│ QQuery 66 │ 1134.08 ms │                                   572.86 ms │ +1.98x faster │
│ QQuery 67 │  866.67 ms │                                   649.11 ms │ +1.34x faster │
│ QQuery 68 │  423.68 ms │                                   276.76 ms │ +1.53x faster │
│ QQuery 69 │ 1235.03 ms │                                   611.69 ms │ +2.02x faster │
│ QQuery 70 │ 1088.36 ms │                                   743.64 ms │ +1.46x faster │
│ QQuery 71 │ 1009.64 ms │                                   471.77 ms │ +2.14x faster │
│ QQuery 72 │ 1494.35 ms │                                   974.02 ms │ +1.53x faster │
│ QQuery 73 │  494.68 ms │                                   184.32 ms │ +2.68x faster │
│ QQuery 74 │ 1547.21 ms │                                   748.64 ms │ +2.07x faster │
│ QQuery 75 │ 1680.49 ms │                                  1042.29 ms │ +1.61x faster │
│ QQuery 76 │  726.22 ms │                                   515.79 ms │ +1.41x faster │
│ QQuery 77 │ 1221.65 ms │                                   673.30 ms │ +1.81x faster │
│ QQuery 78 │  873.35 ms │                                   597.01 ms │ +1.46x faster │
│ QQuery 79 │  990.85 ms │                                   494.40 ms │ +2.00x faster │
│ QQuery 80 │ 1297.06 ms │                                   699.46 ms │ +1.85x faster │
│ QQuery 81 │  925.13 ms │                                   278.13 ms │ +3.33x faster │
│ QQuery 82 │ 1002.20 ms │                                   413.57 ms │ +2.42x faster │
│ QQuery 83 │ 1170.46 ms │                                   815.82 ms │ +1.43x faster │
│ QQuery 84 │  851.86 ms │                                   385.36 ms │ +2.21x faster │
│ QQuery 85 │ 1461.10 ms │                                   673.66 ms │ +2.17x faster │
│ QQuery 86 │  532.21 ms │                                   327.25 ms │ +1.63x faster │
│ QQuery 87 │ 1028.45 ms │                                   859.68 ms │ +1.20x faster │
│ QQuery 88 │ 1345.06 ms │                                   727.85 ms │ +1.85x faster │
│ QQuery 89 │  807.35 ms │                                   332.71 ms │ +2.43x faster │
│ QQuery 90 │  833.90 ms │                                   465.09 ms │ +1.79x faster │
│ QQuery 91 │ 1048.92 ms │                                   671.91 ms │ +1.56x faster │
│ QQuery 92 │  742.39 ms │                                   394.78 ms │ +1.88x faster │
│ QQuery 93 │  294.00 ms │                                   275.71 ms │ +1.07x faster │
│ QQuery 94 │ 1074.26 ms │                                   406.29 ms │ +2.64x faster │
│ QQuery 95 │ 1144.34 ms │                                   738.64 ms │ +1.55x faster │
│ QQuery 96 │  663.60 ms │                                   331.27 ms │ +2.00x faster │
│ QQuery 97 │  652.58 ms │                                   467.52 ms │ +1.40x faster │
│ QQuery 98 │  607.17 ms │                                   314.76 ms │ +1.93x faster │
│ QQuery 99 │  924.22 ms │                                   431.75 ms │ +2.14x faster │
└───────────┴────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃             ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 107742.52ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  62912.13ms │
│ Average Time (HEAD)                                        │   1088.31ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │    635.48ms │
│ Queries Faster                                             │          94 │
│ Queries Slower                                             │           5 │
│ Queries with No Change                                     │           0 │
│ Queries with Failure                                       │           0 │
└────────────────────────────────────────────────────────────┴─────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     105.71 / 142.63 ±44.21 / 205.78 ms │          136.23 / 182.62 ±30.28 / 207.33 ms │  1.28x slower │
│ QQuery 2  │ 1041.67 / 1335.53 ±264.79 / 1702.23 ms │          699.84 / 824.06 ±82.81 / 947.63 ms │ +1.62x faster │
│ QQuery 3  │     494.03 / 626.47 ±98.78 / 774.41 ms │          349.27 / 448.66 ±64.85 / 539.24 ms │ +1.40x faster │
│ QQuery 4  │ 1780.95 / 2273.11 ±264.41 / 2528.31 ms │      1238.02 / 1394.42 ±120.85 / 1589.80 ms │ +1.63x faster │
│ QQuery 5  │     750.89 / 796.60 ±50.79 / 892.27 ms │          530.55 / 580.36 ±42.63 / 642.38 ms │ +1.37x faster │
│ QQuery 6  │ 1046.28 / 1189.86 ±107.72 / 1340.35 ms │         445.56 / 689.24 ±164.85 / 926.96 ms │ +1.73x faster │
│ QQuery 7  │  897.02 / 1124.63 ±168.53 / 1335.61 ms │          453.11 / 540.43 ±83.45 / 647.64 ms │ +2.08x faster │
│ QQuery 8  │     595.80 / 634.07 ±31.86 / 689.11 ms │         303.15 / 454.38 ±131.37 / 652.49 ms │ +1.40x faster │
│ QQuery 9  │ 1149.36 / 1683.63 ±358.92 / 2214.98 ms │      1371.29 / 1735.88 ±282.65 / 2130.64 ms │     no change │
│ QQuery 10 │ 1039.61 / 1439.39 ±242.98 / 1751.68 ms │         812.63 / 934.28 ±91.93 / 1060.97 ms │ +1.54x faster │
│ QQuery 11 │ 1334.98 / 1629.44 ±173.21 / 1857.60 ms │        767.05 / 920.90 ±121.81 / 1071.17 ms │ +1.77x faster │
│ QQuery 12 │   553.64 / 703.48 ±197.28 / 1083.83 ms │          215.17 / 270.95 ±55.62 / 375.48 ms │ +2.60x faster │
│ QQuery 13 │ 1490.36 / 1655.46 ±156.56 / 1910.84 ms │          657.78 / 672.88 ±14.58 / 695.31 ms │ +2.46x faster │
│ QQuery 14 │ 6815.17 / 7489.91 ±461.49 / 8118.92 ms │      4230.79 / 4620.71 ±285.36 / 5114.86 ms │ +1.62x faster │
│ QQuery 15 │     561.23 / 678.40 ±95.79 / 838.55 ms │          375.50 / 444.13 ±45.57 / 503.23 ms │ +1.53x faster │
│ QQuery 16 │     140.45 / 183.80 ±29.13 / 207.83 ms │          116.69 / 159.80 ±28.49 / 206.56 ms │ +1.15x faster │
│ QQuery 17 │ 1358.91 / 1615.25 ±237.95 / 2067.34 ms │       881.42 / 1017.40 ±115.10 / 1175.36 ms │ +1.59x faster │
│ QQuery 18 │  1263.76 / 1337.88 ±78.64 / 1446.51 ms │         470.35 / 686.60 ±138.22 / 900.94 ms │ +1.95x faster │
│ QQuery 19 │  887.50 / 1273.24 ±224.90 / 1593.48 ms │         520.83 / 644.93 ±115.90 / 843.13 ms │ +1.97x faster │
│ QQuery 20 │     522.43 / 640.68 ±87.01 / 784.07 ms │          318.59 / 444.84 ±69.10 / 511.24 ms │ +1.44x faster │
│ QQuery 21 │  836.69 / 1035.68 ±135.25 / 1216.59 ms │         392.71 / 578.05 ±110.13 / 715.09 ms │ +1.79x faster │
│ QQuery 22 │     639.90 / 703.35 ±52.82 / 794.34 ms │         265.29 / 485.15 ±143.12 / 674.59 ms │ +1.45x faster │
│ QQuery 23 │ 2604.49 / 3364.00 ±547.01 / 3944.38 ms │      2456.48 / 2708.68 ±245.40 / 3157.72 ms │ +1.24x faster │
│ QQuery 24 │ 2501.39 / 3102.41 ±322.08 / 3406.83 ms │      1905.85 / 2151.20 ±148.96 / 2365.23 ms │ +1.44x faster │
│ QQuery 25 │ 1391.67 / 1582.96 ±190.37 / 1898.01 ms │          740.44 / 839.25 ±88.66 / 953.12 ms │ +1.89x faster │
│ QQuery 26 │  777.27 / 1058.02 ±160.54 / 1271.85 ms │          347.51 / 513.57 ±98.58 / 623.07 ms │ +2.06x faster │
│ QQuery 27 │     189.34 / 261.10 ±68.54 / 368.84 ms │          218.62 / 263.95 ±43.83 / 338.32 ms │     no change │
│ QQuery 28 │     476.11 / 500.84 ±20.70 / 538.61 ms │          338.97 / 383.12 ±49.61 / 459.86 ms │ +1.31x faster │
│ QQuery 29 │ 1519.39 / 1689.16 ±190.74 / 2003.99 ms │       787.36 / 1032.70 ±152.27 / 1190.85 ms │ +1.64x faster │
│ QQuery 30 │   686.32 / 940.43 ±197.39 / 1242.89 ms │          539.67 / 645.33 ±88.25 / 750.77 ms │ +1.46x faster │
│ QQuery 31 │ 2959.74 / 3297.70 ±309.69 / 3876.19 ms │      1435.78 / 1635.65 ±191.76 / 1915.72 ms │ +2.02x faster │
│ QQuery 32 │  699.37 / 1081.22 ±218.07 / 1294.16 ms │         371.63 / 583.89 ±122.03 / 724.36 ms │ +1.85x faster │
│ QQuery 33 │ 1471.40 / 1589.78 ±108.58 / 1781.22 ms │         617.30 / 751.51 ±119.41 / 951.29 ms │ +2.12x faster │
│ QQuery 34 │    342.28 / 535.91 ±112.11 / 672.67 ms │          207.34 / 354.25 ±88.94 / 482.58 ms │ +1.51x faster │
│ QQuery 35 │ 1476.14 / 1663.85 ±177.69 / 1934.04 ms │        749.06 / 985.56 ±208.43 / 1341.35 ms │ +1.69x faster │
│ QQuery 36 │      77.62 / 150.21 ±47.34 / 207.72 ms │           90.92 / 137.38 ±39.43 / 191.06 ms │ +1.09x faster │
│ QQuery 37 │     318.19 / 484.02 ±89.91 / 574.91 ms │          173.92 / 309.23 ±96.38 / 420.77 ms │ +1.57x faster │
│ QQuery 38 │ 1145.97 / 1383.03 ±149.52 / 1618.32 ms │          615.90 / 703.23 ±69.57 / 820.13 ms │ +1.97x faster │
│ QQuery 39 │ 3327.85 / 3749.76 ±280.06 / 4057.19 ms │       1887.16 / 1942.63 ±81.58 / 2104.65 ms │ +1.93x faster │
│ QQuery 40 │    474.78 / 745.95 ±171.27 / 944.26 ms │         426.93 / 544.10 ±101.69 / 672.68 ms │ +1.37x faster │
│ QQuery 41 │    262.42 / 468.38 ±162.69 / 711.97 ms │          219.55 / 297.94 ±74.20 / 416.77 ms │ +1.57x faster │
│ QQuery 42 │     731.53 / 834.64 ±81.06 / 961.42 ms │          230.55 / 355.79 ±76.29 / 462.89 ms │ +2.35x faster │
│ QQuery 43 │      91.46 / 125.84 ±31.93 / 186.48 ms │           40.75 / 113.16 ±53.04 / 206.31 ms │ +1.11x faster │
│ QQuery 44 │     143.67 / 231.94 ±53.75 / 312.02 ms │          188.49 / 273.76 ±65.31 / 373.22 ms │  1.18x slower │
│ QQuery 45 │   945.53 / 1046.81 ±99.37 / 1183.09 ms │          302.51 / 460.34 ±99.37 / 589.75 ms │ +2.27x faster │
│ QQuery 46 │     570.62 / 640.70 ±85.21 / 807.73 ms │         238.86 / 480.73 ±183.72 / 722.45 ms │ +1.33x faster │
│ QQuery 47 │ 1931.62 / 2270.45 ±371.13 / 2992.61 ms │       1403.91 / 1526.74 ±69.58 / 1595.06 ms │ +1.49x faster │
│ QQuery 48 │ 1487.46 / 1657.60 ±124.40 / 1827.99 ms │          708.70 / 813.90 ±61.24 / 891.19 ms │ +2.04x faster │
│ QQuery 49 │   989.47 / 1083.35 ±92.96 / 1255.38 ms │          491.07 / 588.12 ±63.49 / 659.18 ms │ +1.84x faster │
│ QQuery 50 │  768.53 / 1218.25 ±249.52 / 1520.45 ms │         614.69 / 731.49 ±108.37 / 869.41 ms │ +1.67x faster │
│ QQuery 51 │   827.66 / 982.51 ±111.37 / 1173.30 ms │         357.14 / 589.11 ±131.62 / 737.18 ms │ +1.67x faster │
│ QQuery 52 │   521.21 / 879.96 ±240.82 / 1107.93 ms │          279.29 / 346.03 ±60.70 / 435.02 ms │ +2.54x faster │
│ QQuery 53 │    776.74 / 915.87 ±87.05 / 1011.32 ms │          336.91 / 489.07 ±93.53 / 589.88 ms │ +1.87x faster │
│ QQuery 54 │ 1224.19 / 1617.15 ±280.97 / 2049.27 ms │         608.98 / 745.02 ±110.95 / 936.15 ms │ +2.17x faster │
│ QQuery 55 │   725.44 / 882.53 ±135.94 / 1126.19 ms │          385.06 / 431.99 ±61.70 / 548.94 ms │ +2.04x faster │
│ QQuery 56 │   999.22 / 1145.10 ±86.20 / 1267.08 ms │         466.64 / 610.09 ±129.12 / 845.65 ms │ +1.88x faster │
│ QQuery 57 │ 2061.97 / 2312.33 ±255.78 / 2775.55 ms │      1056.39 / 1284.51 ±207.49 / 1626.15 ms │ +1.80x faster │
│ QQuery 58 │ 1876.56 / 2372.76 ±394.12 / 3073.25 ms │      1122.43 / 1343.36 ±151.06 / 1548.69 ms │ +1.77x faster │
│ QQuery 59 │ 1084.83 / 1467.73 ±201.68 / 1677.89 ms │          703.28 / 862.48 ±80.56 / 922.48 ms │ +1.70x faster │
│ QQuery 60 │  866.79 / 1110.94 ±218.29 / 1482.51 ms │          551.03 / 631.21 ±82.72 / 787.56 ms │ +1.76x faster │
│ QQuery 61 │    444.72 / 617.50 ±101.98 / 761.03 ms │         325.02 / 395.07 ±100.78 / 594.70 ms │ +1.56x faster │
│ QQuery 62 │  559.14 / 1020.66 ±299.18 / 1457.77 ms │         404.22 / 547.99 ±136.56 / 777.57 ms │ +1.86x faster │
│ QQuery 63 │   713.57 / 955.30 ±261.70 / 1446.74 ms │          331.89 / 463.68 ±73.95 / 536.77 ms │ +2.06x faster │
│ QQuery 64 │ 4889.11 / 5213.59 ±292.29 / 5661.75 ms │      2709.97 / 3208.38 ±278.48 / 3545.70 ms │ +1.62x faster │
│ QQuery 65 │ 1310.88 / 1632.19 ±225.04 / 1927.97 ms │        805.36 / 985.16 ±132.93 / 1128.57 ms │ +1.66x faster │
│ QQuery 66 │ 1134.08 / 1309.89 ±138.79 / 1497.63 ms │          572.86 / 654.37 ±50.14 / 727.99 ms │ +2.00x faster │
│ QQuery 67 │  866.67 / 1072.22 ±140.51 / 1216.02 ms │          649.11 / 732.04 ±74.32 / 861.95 ms │ +1.46x faster │
│ QQuery 68 │    423.68 / 549.00 ±107.77 / 706.66 ms │          276.76 / 444.46 ±92.29 / 547.91 ms │ +1.24x faster │
│ QQuery 69 │ 1235.03 / 1495.09 ±145.97 / 1642.81 ms │         611.69 / 824.19 ±121.69 / 963.12 ms │ +1.81x faster │
│ QQuery 70 │ 1088.36 / 1254.42 ±137.64 / 1506.45 ms │          743.64 / 805.86 ±66.30 / 924.92 ms │ +1.56x faster │
│ QQuery 71 │ 1009.64 / 1170.39 ±106.31 / 1291.42 ms │          471.77 / 630.95 ±97.14 / 750.09 ms │ +1.85x faster │
│ QQuery 72 │ 1494.35 / 1760.74 ±163.08 / 1997.85 ms │       974.02 / 1104.97 ±182.82 / 1463.72 ms │ +1.59x faster │
│ QQuery 73 │     494.68 / 551.95 ±60.40 / 664.10 ms │          184.32 / 327.58 ±91.34 / 445.70 ms │ +1.68x faster │
│ QQuery 74 │ 1547.21 / 1726.65 ±154.37 / 1900.14 ms │        748.64 / 961.50 ±111.27 / 1075.58 ms │ +1.80x faster │
│ QQuery 75 │ 1680.49 / 1888.19 ±147.46 / 2127.37 ms │      1042.29 / 1206.07 ±101.31 / 1354.41 ms │ +1.57x faster │
│ QQuery 76 │    726.22 / 830.88 ±97.39 / 1000.29 ms │          515.79 / 596.38 ±70.17 / 690.76 ms │ +1.39x faster │
│ QQuery 77 │  1221.65 / 1313.35 ±66.94 / 1393.92 ms │         673.30 / 811.88 ±110.66 / 946.96 ms │ +1.62x faster │
│ QQuery 78 │  873.35 / 1055.49 ±141.94 / 1256.72 ms │         597.01 / 769.17 ±122.61 / 939.12 ms │ +1.37x faster │
│ QQuery 79 │  990.85 / 1232.54 ±188.26 / 1529.71 ms │         494.40 / 635.23 ±120.47 / 845.26 ms │ +1.94x faster │
│ QQuery 80 │ 1297.06 / 1440.67 ±116.64 / 1630.69 ms │          699.46 / 833.97 ±88.18 / 948.10 ms │ +1.73x faster │
│ QQuery 81 │  925.13 / 1116.98 ±188.42 / 1440.50 ms │         278.13 / 518.45 ±131.86 / 636.71 ms │ +2.15x faster │
│ QQuery 82 │ 1002.20 / 1200.56 ±140.98 / 1382.11 ms │         413.57 / 615.37 ±102.19 / 689.31 ms │ +1.95x faster │
│ QQuery 83 │ 1170.46 / 1490.94 ±229.38 / 1717.48 ms │         815.82 / 940.72 ±89.09 / 1048.95 ms │ +1.58x faster │
│ QQuery 84 │    851.86 / 951.26 ±99.29 / 1138.93 ms │         385.36 / 600.68 ±131.35 / 737.38 ms │ +1.58x faster │
│ QQuery 85 │ 1461.10 / 1658.85 ±164.88 / 1947.50 ms │          673.66 / 797.59 ±91.91 / 909.15 ms │ +2.08x faster │
│ QQuery 86 │     532.21 / 590.26 ±47.98 / 658.34 ms │          327.25 / 489.46 ±89.71 / 597.85 ms │ +1.21x faster │
│ QQuery 87 │ 1028.45 / 1281.55 ±194.35 / 1521.39 ms │        859.68 / 946.18 ±111.55 / 1156.41 ms │ +1.35x faster │
│ QQuery 88 │  1345.06 / 1473.61 ±69.64 / 1536.14 ms │          727.85 / 844.30 ±61.02 / 905.76 ms │ +1.75x faster │
│ QQuery 89 │  807.35 / 1001.22 ±190.09 / 1341.39 ms │         332.71 / 522.31 ±111.72 / 677.86 ms │ +1.92x faster │
│ QQuery 90 │    833.90 / 994.64 ±88.59 / 1090.76 ms │          465.09 / 546.67 ±90.71 / 721.24 ms │ +1.82x faster │
│ QQuery 91 │ 1048.92 / 1226.65 ±149.55 / 1429.02 ms │          671.91 / 780.98 ±92.32 / 929.33 ms │ +1.57x faster │
│ QQuery 92 │   742.39 / 937.80 ±153.77 / 1143.46 ms │          394.78 / 520.11 ±88.51 / 669.82 ms │ +1.80x faster │
│ QQuery 93 │     294.00 / 426.42 ±89.77 / 554.97 ms │          275.71 / 407.98 ±72.97 / 492.16 ms │     no change │
│ QQuery 94 │ 1074.26 / 1329.42 ±192.10 / 1556.22 ms │          406.29 / 555.23 ±99.71 / 717.96 ms │ +2.39x faster │
│ QQuery 95 │ 1144.34 / 1580.34 ±353.92 / 2166.38 ms │        738.64 / 892.91 ±162.79 / 1174.46 ms │ +1.77x faster │
│ QQuery 96 │   663.60 / 905.48 ±190.13 / 1123.54 ms │         331.27 / 485.44 ±159.85 / 699.17 ms │ +1.87x faster │
│ QQuery 97 │   652.58 / 843.99 ±140.97 / 1005.43 ms │          467.52 / 583.28 ±68.75 / 664.93 ms │ +1.45x faster │
│ QQuery 98 │    607.17 / 779.41 ±126.15 / 934.08 ms │         314.76 / 432.86 ±118.33 / 651.70 ms │ +1.80x faster │
│ QQuery 99 │  924.22 / 1053.29 ±135.07 / 1273.70 ms │         431.75 / 577.90 ±101.65 / 750.26 ms │ +1.82x faster │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃             ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 129571.05ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │  77187.99ms │
│ Average Time (HEAD)                                        │   1308.80ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │    779.68ms │
│ Queries Faster                                             │          94 │
│ Queries Slower                                             │           2 │
│ Queries with No Change                                     │           3 │
│ Queries with Failure                                       │           0 │
└────────────────────────────────────────────────────────────┴─────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	655.1s
Peak memory	1.2 GiB
Avg memory	652.1 MiB
CPU user	119.5s
CPU sys	15.0s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	390.1s
Peak memory	1.1 GiB
Avg memory	607.9 MiB
CPU user	135.8s
CPU sys	10.0s
Peak spill	0 B

File an issue against this benchmark runner

adriangb · 2026-09-29T22:22:33Z

Benchmark summary: `02429ca` (arrow-rs pin with apache/arrow-rs#11241 in place of apache/arrow-rs#11223)

This revision is rebased on main after #25809, which draws the simulated object store latency at random instead of round-robin. The arrow-rs pin (pydantic/arrow-rs 6bea7c8969) is now 60.0.0 + apache/arrow-rs#11241 + apache/arrow-rs#10555 (current tips). The DataFusion code did not change.

read_ahead_bytes = 100 MB on the branch side only. GKE runs: adriangbot c4a-highmem-16, compared with merge-base 953518c. "Speedup" is the ratio of the per-query time totals. The previous column is the speedup of 4a9f276 from the previous summary. Because #25809 changed the latency model, the two columns are not exactly comparable.

With `SIMULATE_LATENCY`, `pushdown_filters=false`

suite	query total base → branch	speedup	previous	faster / slower / same	peak memory base → branch
tpch_sf1	17.02 s → 9.27 s	1.84x	1.75x	21 / 0 / 1	890 MiB → 1.0 GiB
tpch_sf10	113.68 s → 15.71 s	7.24x	7.03x	22 / 0 / 0	3.6 GiB → 4.4 GiB
tpcds_sf1	55.62 s → 56.43 s	0.99x	1.00x	37 / 46 / 16	1.1 GiB → 1.0 GiB
clickbench_partitioned	84.45 s → 42.64 s	1.98x	1.96x	38 / 1 / 4	16.4 GiB → 17.0 GiB

With `SIMULATE_LATENCY`, `pushdown_filters=true`

suite	query total base → branch	speedup	previous	faster / slower / same	peak memory base → branch
tpch_sf1	38.58 s → 9.52 s	4.05x	3.73x	22 / 0 / 0	677 MiB → 713 MiB
tpcds_sf1	107.74 s → 62.91 s	1.71x	1.72x	94 / 5 / 0	1.2 GiB → 1.1 GiB
clickbench_partitioned	105.25 s → 39.09 s	2.69x	2.75x	41 / 0 / 2	9.5 GiB → 11.0 GiB
tpch_sf1, `DATAFUSION_RUNTIME_MEMORY_LIMIT=1G`	39.53 s → 9.63 s	4.10x	3.82x	22 / 0 / 0	674 MiB → 847 MiB

Notes

The new arrow-rs pin does not change the results. All speedups are within a few percent of 4a9f276. The tpch pushdown runs are a little faster (4.05x and 4.10x, previously 3.73x and 3.82x).
tpcds without pushdown is flat, as before (0.99x). The query counts changed from 9 / 12 / 78 to 37 / 46 / 16 because the random latency draws make each query more variable: the standard deviation on the base side is 30–60% of the mean for many queries. The mean-based total is also flat (68.4 s → 69.3 s).
tpcds Q27 is no longer slower in the no-pushdown run: its mean is 1.12x faster. It is 1.15x slower with pushdown (189 → 219 ms), previously 1.55x.
tpcds Q1 and Q44 are slower in both tpcds runs: Q1 is 1.29x slower with pushdown and 1.56x without, and Q44 is 1.31x slower with pushdown and 1.33x (mean) without. These are small queries (100–250 ms). I will profile them.
The 1 GB pool run has no failures.

Runs: pushdown + latency, latency, 1 GB pool.

adriangb · 2026-09-29T23:55:50Z

run benchmark tpch tpch10 tpcds clickbench_partitioned

env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

adriangbot · 2026-09-30T00:15:53Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5901309623-2932-prwq4 6.12.94+ #1 SMP Wed Aug 19 07:47:20 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch
env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-30T00:15:54Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5901309623-2933-bng8x 6.12.94+ #1 SMP Wed Aug 19 07:47:20 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch10
env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-30T00:15:54Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5901309623-2934-qjj7l 6.12.94+ #1 SMP Wed Aug 19 07:47:20 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpcds
env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-30T00:16:41Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5901309623-2935-lrf5s 6.12.94+ #1 SMP Wed Aug 19 07:47:20 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-30T00:19:41Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch
env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃     HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 41.10 ms │                                    41.46 ms │ no change │
│ QQuery 2  │ 19.05 ms │                                    19.19 ms │ no change │
│ QQuery 3  │ 29.44 ms │                                    29.17 ms │ no change │
│ QQuery 4  │ 17.69 ms │                                    18.03 ms │ no change │
│ QQuery 5  │ 36.39 ms │                                    36.49 ms │ no change │
│ QQuery 6  │ 16.66 ms │                                    16.80 ms │ no change │
│ QQuery 7  │ 43.67 ms │                                    43.13 ms │ no change │
│ QQuery 8  │ 41.47 ms │                                    41.78 ms │ no change │
│ QQuery 9  │ 49.94 ms │                                    50.72 ms │ no change │
│ QQuery 10 │ 41.65 ms │                                    41.44 ms │ no change │
│ QQuery 11 │ 14.19 ms │                                    14.28 ms │ no change │
│ QQuery 12 │ 21.65 ms │                                    21.80 ms │ no change │
│ QQuery 13 │ 41.81 ms │                                    41.00 ms │ no change │
│ QQuery 14 │ 24.99 ms │                                    25.26 ms │ no change │
│ QQuery 15 │ 31.87 ms │                                    32.07 ms │ no change │
│ QQuery 16 │ 14.28 ms │                                    14.32 ms │ no change │
│ QQuery 17 │ 76.92 ms │                                    77.41 ms │ no change │
│ QQuery 18 │ 63.55 ms │                                    64.88 ms │ no change │
│ QQuery 19 │ 32.37 ms │                                    33.08 ms │ no change │
│ QQuery 20 │ 33.10 ms │                                    32.95 ms │ no change │
│ QQuery 21 │ 57.50 ms │                                    57.94 ms │ no change │
│ QQuery 22 │ 14.60 ms │                                    14.55 ms │ no change │
└───────────┴──────────┴─────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 763.91ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 767.75ms │
│ Average Time (HEAD)                                        │  34.72ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  34.90ms │
│ Queries Faster                                             │        0 │
│ Queries Slower                                             │        0 │
│ Queries with No Change                                     │       22 │
│ Queries with Failure                                       │        0 │
└────────────────────────────────────────────────────────────┴──────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 41.10 / 42.16 ±1.03 / 44.08 ms │              41.46 / 42.43 ±1.21 / 44.80 ms │ no change │
│ QQuery 2  │ 19.05 / 19.35 ±0.28 / 19.87 ms │              19.19 / 19.38 ±0.24 / 19.83 ms │ no change │
│ QQuery 3  │ 29.44 / 29.71 ±0.39 / 30.48 ms │              29.17 / 30.06 ±1.30 / 32.63 ms │ no change │
│ QQuery 4  │ 17.69 / 17.87 ±0.13 / 18.01 ms │              18.03 / 18.46 ±0.41 / 19.09 ms │ no change │
│ QQuery 5  │ 36.39 / 37.41 ±1.22 / 39.79 ms │              36.49 / 36.67 ±0.14 / 36.88 ms │ no change │
│ QQuery 6  │ 16.66 / 16.86 ±0.21 / 17.24 ms │              16.80 / 16.96 ±0.15 / 17.25 ms │ no change │
│ QQuery 7  │ 43.67 / 46.54 ±1.89 / 48.65 ms │              43.13 / 45.82 ±1.96 / 49.17 ms │ no change │
│ QQuery 8  │ 41.47 / 43.10 ±1.22 / 44.30 ms │              41.78 / 42.09 ±0.23 / 42.41 ms │ no change │
│ QQuery 9  │ 49.94 / 51.31 ±1.01 / 52.73 ms │              50.72 / 51.30 ±0.52 / 52.24 ms │ no change │
│ QQuery 10 │ 41.65 / 43.10 ±1.41 / 44.92 ms │              41.44 / 42.17 ±0.88 / 43.87 ms │ no change │
│ QQuery 11 │ 14.19 / 14.39 ±0.17 / 14.64 ms │              14.28 / 14.68 ±0.43 / 15.47 ms │ no change │
│ QQuery 12 │ 21.65 / 22.23 ±0.57 / 23.29 ms │              21.80 / 22.13 ±0.23 / 22.49 ms │ no change │
│ QQuery 13 │ 41.81 / 43.82 ±2.21 / 47.84 ms │              41.00 / 44.49 ±3.70 / 51.32 ms │ no change │
│ QQuery 14 │ 24.99 / 25.42 ±0.27 / 25.74 ms │              25.26 / 25.54 ±0.19 / 25.82 ms │ no change │
│ QQuery 15 │ 31.87 / 32.43 ±0.70 / 33.80 ms │              32.07 / 32.31 ±0.16 / 32.51 ms │ no change │
│ QQuery 16 │ 14.28 / 14.58 ±0.17 / 14.74 ms │              14.32 / 14.58 ±0.18 / 14.84 ms │ no change │
│ QQuery 17 │ 76.92 / 77.61 ±0.61 / 78.73 ms │              77.41 / 79.03 ±1.84 / 82.21 ms │ no change │
│ QQuery 18 │ 63.55 / 65.43 ±1.80 / 68.43 ms │              64.88 / 65.71 ±0.75 / 66.93 ms │ no change │
│ QQuery 19 │ 32.37 / 32.87 ±0.73 / 34.30 ms │              33.08 / 33.73 ±0.74 / 35.15 ms │ no change │
│ QQuery 20 │ 33.10 / 33.43 ±0.33 / 34.06 ms │              32.95 / 33.34 ±0.25 / 33.66 ms │ no change │
│ QQuery 21 │ 57.50 / 58.55 ±0.97 / 60.13 ms │              57.94 / 58.65 ±0.76 / 59.76 ms │ no change │
│ QQuery 22 │ 14.60 / 15.11 ±0.81 / 16.71 ms │              14.55 / 14.77 ±0.15 / 14.95 ms │ no change │
└───────────┴────────────────────────────────┴─────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 783.28ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 784.30ms │
│ Average Time (HEAD)                                        │  35.60ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  35.65ms │
│ Queries Faster                                             │        0 │
│ Queries Slower                                             │        0 │
│ Queries with No Change                                     │       22 │
│ Queries with Failure                                       │        0 │
└────────────────────────────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	1.3 GiB
Avg memory	500.9 MiB
CPU user	22.0s
CPU sys	1.9s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	1.3 GiB
Avg memory	529.4 MiB
CPU user	22.2s
CPU sys	1.9s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-30T00:21:06Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpcds
env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.61 ms │                                     5.55 ms │    no change │
│ QQuery 2  │   80.72 ms │                                    83.65 ms │    no change │
│ QQuery 3  │   28.76 ms │                                    29.91 ms │    no change │
│ QQuery 4  │  458.58 ms │                                   464.48 ms │    no change │
│ QQuery 5  │   51.66 ms │                                    52.45 ms │    no change │
│ QQuery 6  │   35.42 ms │                                    37.21 ms │ 1.05x slower │
│ QQuery 7  │   74.39 ms │                                    76.86 ms │    no change │
│ QQuery 8  │   36.39 ms │                                    37.56 ms │    no change │
│ QQuery 9  │   53.17 ms │                                    53.21 ms │    no change │
│ QQuery 10 │   61.75 ms │                                    64.46 ms │    no change │
│ QQuery 11 │  285.61 ms │                                   290.54 ms │    no change │
│ QQuery 12 │   28.98 ms │                                    29.98 ms │    no change │
│ QQuery 13 │  117.29 ms │                                   119.88 ms │    no change │
│ QQuery 14 │  397.44 ms │                                   405.40 ms │    no change │
│ QQuery 15 │   55.54 ms │                                    56.43 ms │    no change │
│ QQuery 16 │    6.40 ms │                                     6.48 ms │    no change │
│ QQuery 17 │   78.53 ms │                                    80.26 ms │    no change │
│ QQuery 18 │  104.83 ms │                                   106.71 ms │    no change │
│ QQuery 19 │   41.58 ms │                                    43.50 ms │    no change │
│ QQuery 20 │   36.25 ms │                                    36.64 ms │    no change │
│ QQuery 21 │   17.19 ms │                                    17.57 ms │    no change │
│ QQuery 22 │   64.32 ms │                                    64.35 ms │    no change │
│ QQuery 23 │  312.19 ms │                                   317.60 ms │    no change │
│ QQuery 24 │  197.66 ms │                                   205.07 ms │    no change │
│ QQuery 25 │  107.28 ms │                                   110.93 ms │    no change │
│ QQuery 26 │   47.70 ms │                                    49.32 ms │    no change │
│ QQuery 27 │    6.13 ms │                                     6.12 ms │    no change │
│ QQuery 28 │   55.83 ms │                                    60.88 ms │ 1.09x slower │
│ QQuery 29 │   95.40 ms │                                    97.38 ms │    no change │
│ QQuery 30 │   32.90 ms │                                    33.14 ms │    no change │
│ QQuery 31 │  109.68 ms │                                   112.63 ms │    no change │
│ QQuery 32 │   20.57 ms │                                    21.71 ms │ 1.06x slower │
│ QQuery 33 │   37.80 ms │                                    38.38 ms │    no change │
│ QQuery 34 │   10.30 ms │                                    10.32 ms │    no change │
│ QQuery 35 │   72.60 ms │                                    75.09 ms │    no change │
│ QQuery 36 │    5.81 ms │                                     5.87 ms │    no change │
│ QQuery 37 │    6.85 ms │                                     7.02 ms │    no change │
│ QQuery 38 │   62.36 ms │                                    63.25 ms │    no change │
│ QQuery 39 │   74.37 ms │                                    75.33 ms │    no change │
│ QQuery 40 │   23.74 ms │                                    24.16 ms │    no change │
│ QQuery 41 │   11.34 ms │                                    11.73 ms │    no change │
│ QQuery 42 │   24.04 ms │                                    25.04 ms │    no change │
│ QQuery 43 │    4.85 ms │                                     4.93 ms │    no change │
│ QQuery 44 │    8.89 ms │                                     9.19 ms │    no change │
│ QQuery 45 │   37.55 ms │                                    38.85 ms │    no change │
│ QQuery 46 │   12.16 ms │                                    12.55 ms │    no change │
│ QQuery 47 │  227.64 ms │                                   235.43 ms │    no change │
│ QQuery 48 │   94.48 ms │                                    97.91 ms │    no change │
│ QQuery 49 │   70.73 ms │                                    72.15 ms │    no change │
│ QQuery 50 │   58.71 ms │                                    59.91 ms │    no change │
│ QQuery 51 │   92.61 ms │                                    94.98 ms │    no change │
│ QQuery 52 │   24.39 ms │                                    25.54 ms │    no change │
│ QQuery 53 │   29.45 ms │                                    30.37 ms │    no change │
│ QQuery 54 │   54.57 ms │                                    55.73 ms │    no change │
│ QQuery 55 │   24.21 ms │                                    24.50 ms │    no change │
│ QQuery 56 │   39.50 ms │                                    39.80 ms │    no change │
│ QQuery 57 │  170.10 ms │                                   173.49 ms │    no change │
│ QQuery 58 │  109.14 ms │                                   113.78 ms │    no change │
│ QQuery 59 │  116.49 ms │                                   120.10 ms │    no change │
│ QQuery 60 │   39.45 ms │                                    40.47 ms │    no change │
│ QQuery 61 │   11.88 ms │                                    11.75 ms │    no change │
│ QQuery 62 │   44.61 ms │                                    46.25 ms │    no change │
│ QQuery 63 │   29.90 ms │                                    30.34 ms │    no change │
│ QQuery 64 │  360.88 ms │                                   371.90 ms │    no change │
│ QQuery 65 │  128.68 ms │                                   134.40 ms │    no change │
│ QQuery 66 │   76.98 ms │                                    78.80 ms │    no change │
│ QQuery 67 │  249.80 ms │                                   256.63 ms │    no change │
│ QQuery 68 │   12.31 ms │                                    12.07 ms │    no change │
│ QQuery 69 │   56.66 ms │                                    58.30 ms │    no change │
│ QQuery 70 │  105.49 ms │                                   106.86 ms │    no change │
│ QQuery 71 │   35.55 ms │                                    36.33 ms │    no change │
│ QQuery 72 │ 1791.99 ms │                                  1800.41 ms │    no change │
│ QQuery 73 │    9.79 ms │                                    10.01 ms │    no change │
│ QQuery 74 │  163.77 ms │                                   167.58 ms │    no change │
│ QQuery 75 │  140.64 ms │                                   143.32 ms │    no change │
│ QQuery 76 │   34.57 ms │                                    35.43 ms │    no change │
│ QQuery 77 │   61.11 ms │                                    62.38 ms │    no change │
│ QQuery 78 │  163.11 ms │                                   166.67 ms │    no change │
│ QQuery 79 │   66.20 ms │                                    68.37 ms │    no change │
│ QQuery 80 │   95.88 ms │                                    98.48 ms │    no change │
│ QQuery 81 │   26.03 ms │                                    26.87 ms │    no change │
│ QQuery 82 │   16.73 ms │                                    17.36 ms │    no change │
│ QQuery 83 │   33.73 ms │                                    34.52 ms │    no change │
│ QQuery 84 │   29.74 ms │                                    30.15 ms │    no change │
│ QQuery 85 │  102.67 ms │                                   107.17 ms │    no change │
│ QQuery 86 │   25.68 ms │                                    26.52 ms │    no change │
│ QQuery 87 │   61.72 ms │                                    64.67 ms │    no change │
│ QQuery 88 │   61.01 ms │                                    62.52 ms │    no change │
│ QQuery 89 │   35.09 ms │                                    36.90 ms │ 1.05x slower │
│ QQuery 90 │   16.82 ms │                                    17.48 ms │    no change │
│ QQuery 91 │   44.19 ms │                                    46.61 ms │ 1.05x slower │
│ QQuery 92 │   29.51 ms │                                    31.05 ms │ 1.05x slower │
│ QQuery 93 │   49.09 ms │                                    51.10 ms │    no change │
│ QQuery 94 │   38.82 ms │                                    40.66 ms │    no change │
│ QQuery 95 │   80.65 ms │                                    82.72 ms │    no change │
│ QQuery 96 │   24.03 ms │                                    24.61 ms │    no change │
│ QQuery 97 │   51.15 ms │                                    52.62 ms │    no change │
│ QQuery 98 │   43.14 ms │                                    43.80 ms │    no change │
│ QQuery 99 │   65.57 ms │                                    68.51 ms │    no change │
└───────────┴────────────┴─────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 9125.10ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 9325.82ms │
│ Average Time (HEAD)                                        │   92.17ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   94.20ms │
│ Queries Faster                                             │         0 │
│ Queries Slower                                             │         6 │
│ Queries with No Change                                     │        93 │
│ Queries with Failure                                       │         0 │
└────────────────────────────────────────────────────────────┴───────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.61 / 6.18 ±0.96 / 8.09 ms │                 5.55 / 6.17 ±0.99 / 8.13 ms │     no change │
│ QQuery 2  │        80.72 / 80.99 ±0.17 / 81.18 ms │              83.65 / 83.83 ±0.14 / 84.06 ms │     no change │
│ QQuery 3  │        28.76 / 29.23 ±0.28 / 29.62 ms │              29.91 / 30.10 ±0.12 / 30.21 ms │     no change │
│ QQuery 4  │     458.58 / 462.15 ±2.55 / 466.35 ms │           464.48 / 470.49 ±4.79 / 478.39 ms │     no change │
│ QQuery 5  │        51.66 / 53.44 ±3.29 / 60.02 ms │              52.45 / 55.08 ±3.60 / 62.09 ms │     no change │
│ QQuery 6  │        35.42 / 36.13 ±0.44 / 36.70 ms │              37.21 / 37.54 ±0.31 / 38.12 ms │     no change │
│ QQuery 7  │        74.39 / 75.12 ±0.49 / 75.91 ms │              76.86 / 77.16 ±0.28 / 77.68 ms │     no change │
│ QQuery 8  │        36.39 / 36.64 ±0.22 / 37.06 ms │              37.56 / 37.86 ±0.15 / 37.98 ms │     no change │
│ QQuery 9  │        53.17 / 56.64 ±3.14 / 62.48 ms │              53.21 / 55.93 ±1.72 / 58.48 ms │     no change │
│ QQuery 10 │        61.75 / 62.14 ±0.30 / 62.51 ms │              64.46 / 64.68 ±0.23 / 65.12 ms │     no change │
│ QQuery 11 │     285.61 / 288.52 ±2.66 / 291.66 ms │           290.54 / 295.79 ±4.65 / 303.21 ms │     no change │
│ QQuery 12 │        28.98 / 29.79 ±0.68 / 30.99 ms │              29.98 / 30.36 ±0.36 / 31.01 ms │     no change │
│ QQuery 13 │     117.29 / 118.10 ±0.74 / 119.47 ms │           119.88 / 120.53 ±0.51 / 121.46 ms │     no change │
│ QQuery 14 │     397.44 / 400.76 ±3.11 / 404.96 ms │           405.40 / 410.55 ±5.54 / 419.85 ms │     no change │
│ QQuery 15 │        55.54 / 56.36 ±0.76 / 57.67 ms │              56.43 / 58.26 ±3.15 / 64.56 ms │     no change │
│ QQuery 16 │           6.40 / 6.54 ±0.19 / 6.91 ms │                 6.48 / 6.69 ±0.21 / 7.05 ms │     no change │
│ QQuery 17 │        78.53 / 79.86 ±1.49 / 82.74 ms │              80.26 / 81.76 ±1.01 / 83.02 ms │     no change │
│ QQuery 18 │     104.83 / 104.99 ±0.10 / 105.10 ms │           106.71 / 111.25 ±3.95 / 118.50 ms │  1.06x slower │
│ QQuery 19 │        41.58 / 42.93 ±2.16 / 47.25 ms │              43.50 / 44.04 ±0.42 / 44.77 ms │     no change │
│ QQuery 20 │        36.25 / 36.74 ±0.45 / 37.44 ms │              36.64 / 37.00 ±0.19 / 37.17 ms │     no change │
│ QQuery 21 │        17.19 / 17.48 ±0.17 / 17.66 ms │              17.57 / 17.75 ±0.16 / 18.00 ms │     no change │
│ QQuery 22 │        64.32 / 65.20 ±0.50 / 65.86 ms │              64.35 / 65.40 ±0.66 / 66.36 ms │     no change │
│ QQuery 23 │     312.19 / 318.19 ±4.80 / 324.22 ms │           317.60 / 324.45 ±4.23 / 329.74 ms │     no change │
│ QQuery 24 │     197.66 / 200.67 ±2.63 / 204.42 ms │           205.07 / 206.87 ±1.86 / 209.70 ms │     no change │
│ QQuery 25 │     107.28 / 109.84 ±3.29 / 116.34 ms │           110.93 / 112.38 ±1.18 / 114.14 ms │     no change │
│ QQuery 26 │        47.70 / 48.98 ±0.93 / 50.35 ms │              49.32 / 49.97 ±0.51 / 50.75 ms │     no change │
│ QQuery 27 │           6.13 / 6.35 ±0.20 / 6.64 ms │                 6.12 / 6.20 ±0.05 / 6.25 ms │     no change │
│ QQuery 28 │        55.83 / 59.06 ±2.20 / 60.99 ms │              60.88 / 63.65 ±3.12 / 69.70 ms │  1.08x slower │
│ QQuery 29 │       95.40 / 97.69 ±2.53 / 102.58 ms │             97.38 / 98.90 ±1.71 / 102.19 ms │     no change │
│ QQuery 30 │        32.90 / 34.08 ±1.70 / 37.39 ms │              33.14 / 33.45 ±0.28 / 33.97 ms │     no change │
│ QQuery 31 │     109.68 / 111.41 ±2.95 / 117.30 ms │           112.63 / 114.07 ±1.70 / 117.40 ms │     no change │
│ QQuery 32 │        20.57 / 20.83 ±0.16 / 21.03 ms │              21.71 / 22.76 ±1.66 / 26.05 ms │  1.09x slower │
│ QQuery 33 │        37.80 / 38.15 ±0.27 / 38.55 ms │              38.38 / 38.93 ±0.37 / 39.52 ms │     no change │
│ QQuery 34 │        10.30 / 10.48 ±0.27 / 11.01 ms │              10.32 / 10.50 ±0.11 / 10.65 ms │     no change │
│ QQuery 35 │        72.60 / 73.41 ±0.69 / 74.54 ms │              75.09 / 76.25 ±1.01 / 78.02 ms │     no change │
│ QQuery 36 │           5.81 / 5.95 ±0.21 / 6.38 ms │                 5.87 / 6.09 ±0.20 / 6.45 ms │     no change │
│ QQuery 37 │           6.85 / 7.01 ±0.11 / 7.17 ms │                 7.02 / 7.10 ±0.06 / 7.20 ms │     no change │
│ QQuery 38 │        62.36 / 62.74 ±0.37 / 63.32 ms │              63.25 / 64.17 ±0.73 / 65.43 ms │     no change │
│ QQuery 39 │        74.37 / 75.41 ±1.27 / 77.90 ms │              75.33 / 76.73 ±0.87 / 77.63 ms │     no change │
│ QQuery 40 │        23.74 / 24.42 ±0.41 / 24.93 ms │              24.16 / 24.47 ±0.33 / 25.06 ms │     no change │
│ QQuery 41 │        11.34 / 11.61 ±0.14 / 11.76 ms │              11.73 / 11.88 ±0.14 / 12.11 ms │     no change │
│ QQuery 42 │        24.04 / 24.26 ±0.20 / 24.62 ms │              25.04 / 25.73 ±0.99 / 27.68 ms │  1.06x slower │
│ QQuery 43 │           4.85 / 5.03 ±0.19 / 5.39 ms │                 4.93 / 5.12 ±0.23 / 5.57 ms │     no change │
│ QQuery 44 │           8.89 / 9.04 ±0.18 / 9.39 ms │                 9.19 / 9.37 ±0.11 / 9.48 ms │     no change │
│ QQuery 45 │        37.55 / 39.30 ±1.73 / 42.24 ms │              38.85 / 40.67 ±2.02 / 44.61 ms │     no change │
│ QQuery 46 │        12.16 / 12.31 ±0.12 / 12.50 ms │              12.55 / 12.77 ±0.19 / 13.09 ms │     no change │
│ QQuery 47 │     227.64 / 230.43 ±1.50 / 231.82 ms │           235.43 / 238.05 ±2.74 / 242.74 ms │     no change │
│ QQuery 48 │        94.48 / 95.90 ±1.17 / 97.55 ms │              97.91 / 98.52 ±0.64 / 99.60 ms │     no change │
│ QQuery 49 │        70.73 / 74.20 ±5.44 / 84.96 ms │              72.15 / 74.36 ±2.30 / 78.73 ms │     no change │
│ QQuery 50 │        58.71 / 59.75 ±1.37 / 62.43 ms │              59.91 / 60.51 ±0.44 / 61.17 ms │     no change │
│ QQuery 51 │        92.61 / 95.17 ±2.47 / 99.70 ms │             94.98 / 97.63 ±4.14 / 105.86 ms │     no change │
│ QQuery 52 │        24.39 / 26.04 ±2.60 / 31.20 ms │              25.54 / 25.97 ±0.63 / 27.22 ms │     no change │
│ QQuery 53 │        29.45 / 29.74 ±0.24 / 30.15 ms │              30.37 / 30.60 ±0.20 / 30.84 ms │     no change │
│ QQuery 54 │        54.57 / 55.71 ±1.60 / 58.77 ms │              55.73 / 56.36 ±0.41 / 56.87 ms │     no change │
│ QQuery 55 │        24.21 / 24.50 ±0.31 / 25.06 ms │              24.50 / 24.70 ±0.12 / 24.84 ms │     no change │
│ QQuery 56 │        39.50 / 39.81 ±0.39 / 40.53 ms │              39.80 / 41.20 ±1.86 / 44.72 ms │     no change │
│ QQuery 57 │     170.10 / 172.19 ±2.57 / 177.23 ms │           173.49 / 176.41 ±1.81 / 178.70 ms │     no change │
│ QQuery 58 │     109.14 / 109.86 ±0.49 / 110.62 ms │           113.78 / 115.49 ±1.80 / 118.36 ms │  1.05x slower │
│ QQuery 59 │     116.49 / 119.27 ±3.14 / 124.72 ms │           120.10 / 120.49 ±0.24 / 120.80 ms │     no change │
│ QQuery 60 │        39.45 / 39.88 ±0.40 / 40.58 ms │              40.47 / 41.93 ±1.62 / 44.98 ms │  1.05x slower │
│ QQuery 61 │        11.88 / 12.96 ±1.92 / 16.80 ms │              11.75 / 11.97 ±0.16 / 12.16 ms │ +1.08x faster │
│ QQuery 62 │        44.61 / 45.61 ±1.01 / 47.13 ms │              46.25 / 46.59 ±0.27 / 47.03 ms │     no change │
│ QQuery 63 │        29.90 / 30.11 ±0.35 / 30.81 ms │              30.34 / 30.60 ±0.24 / 31.03 ms │     no change │
│ QQuery 64 │     360.88 / 365.31 ±4.97 / 372.82 ms │           371.90 / 376.91 ±6.11 / 388.49 ms │     no change │
│ QQuery 65 │     128.68 / 133.62 ±3.76 / 138.88 ms │           134.40 / 136.30 ±1.31 / 137.85 ms │     no change │
│ QQuery 66 │        76.98 / 78.46 ±1.86 / 82.10 ms │              78.80 / 79.71 ±0.72 / 80.76 ms │     no change │
│ QQuery 67 │     249.80 / 256.03 ±5.19 / 265.04 ms │           256.63 / 262.15 ±3.35 / 267.17 ms │     no change │
│ QQuery 68 │        12.31 / 12.44 ±0.17 / 12.78 ms │              12.07 / 12.21 ±0.13 / 12.44 ms │     no change │
│ QQuery 69 │        56.66 / 56.95 ±0.31 / 57.39 ms │              58.30 / 59.62 ±1.69 / 62.89 ms │     no change │
│ QQuery 70 │     105.49 / 108.08 ±3.66 / 115.30 ms │           106.86 / 107.46 ±0.52 / 108.41 ms │     no change │
│ QQuery 71 │        35.55 / 36.41 ±0.98 / 38.33 ms │              36.33 / 40.13 ±6.13 / 52.37 ms │  1.10x slower │
│ QQuery 72 │ 1791.99 / 1845.82 ±46.86 / 1902.59 ms │       1800.41 / 1858.14 ±39.95 / 1904.71 ms │     no change │
│ QQuery 73 │         9.79 / 10.15 ±0.33 / 10.70 ms │              10.01 / 10.59 ±0.55 / 11.61 ms │     no change │
│ QQuery 74 │     163.77 / 166.26 ±1.62 / 168.83 ms │           167.58 / 173.11 ±3.72 / 179.24 ms │     no change │
│ QQuery 75 │     140.64 / 144.06 ±4.73 / 153.33 ms │           143.32 / 146.06 ±3.50 / 152.74 ms │     no change │
│ QQuery 76 │        34.57 / 34.98 ±0.41 / 35.70 ms │              35.43 / 35.67 ±0.26 / 36.15 ms │     no change │
│ QQuery 77 │        61.11 / 61.81 ±0.45 / 62.44 ms │              62.38 / 63.50 ±0.95 / 65.08 ms │     no change │
│ QQuery 78 │     163.11 / 166.49 ±4.23 / 174.71 ms │           166.67 / 171.80 ±4.18 / 177.86 ms │     no change │
│ QQuery 79 │        66.20 / 68.69 ±4.50 / 77.68 ms │              68.37 / 69.44 ±0.68 / 70.51 ms │     no change │
│ QQuery 80 │       95.88 / 99.36 ±2.56 / 103.02 ms │            98.48 / 100.07 ±1.13 / 101.52 ms │     no change │
│ QQuery 81 │        26.03 / 26.58 ±0.35 / 26.93 ms │              26.87 / 27.94 ±1.90 / 31.73 ms │  1.05x slower │
│ QQuery 82 │        16.73 / 16.92 ±0.19 / 17.28 ms │              17.36 / 17.52 ±0.16 / 17.83 ms │     no change │
│ QQuery 83 │        33.73 / 34.30 ±0.39 / 34.75 ms │              34.52 / 34.92 ±0.31 / 35.21 ms │     no change │
│ QQuery 84 │        29.74 / 31.11 ±2.42 / 35.94 ms │              30.15 / 30.53 ±0.30 / 31.01 ms │     no change │
│ QQuery 85 │     102.67 / 103.82 ±0.86 / 104.94 ms │           107.17 / 109.08 ±2.29 / 113.32 ms │  1.05x slower │
│ QQuery 86 │        25.68 / 25.98 ±0.25 / 26.44 ms │              26.52 / 26.92 ±0.26 / 27.27 ms │     no change │
│ QQuery 87 │        61.72 / 64.13 ±2.52 / 68.99 ms │              64.67 / 65.28 ±0.65 / 66.19 ms │     no change │
│ QQuery 88 │        61.01 / 61.62 ±0.42 / 62.24 ms │              62.52 / 65.42 ±4.66 / 74.71 ms │  1.06x slower │
│ QQuery 89 │        35.09 / 35.78 ±0.58 / 36.83 ms │              36.90 / 37.71 ±0.81 / 39.22 ms │  1.05x slower │
│ QQuery 90 │        16.82 / 17.10 ±0.19 / 17.30 ms │              17.48 / 17.68 ±0.13 / 17.89 ms │     no change │
│ QQuery 91 │        44.19 / 44.61 ±0.30 / 44.88 ms │              46.61 / 46.79 ±0.30 / 47.39 ms │     no change │
│ QQuery 92 │        29.51 / 32.23 ±4.69 / 41.59 ms │              31.05 / 31.48 ±0.34 / 31.98 ms │     no change │
│ QQuery 93 │        49.09 / 49.92 ±0.74 / 50.86 ms │              51.10 / 51.37 ±0.47 / 52.31 ms │     no change │
│ QQuery 94 │        38.82 / 39.64 ±1.03 / 41.63 ms │              40.66 / 41.02 ±0.35 / 41.53 ms │     no change │
│ QQuery 95 │        80.65 / 81.29 ±0.53 / 82.13 ms │              82.72 / 85.21 ±2.66 / 90.07 ms │     no change │
│ QQuery 96 │        24.03 / 24.23 ±0.15 / 24.43 ms │              24.61 / 24.88 ±0.21 / 25.24 ms │     no change │
│ QQuery 97 │        51.15 / 52.21 ±0.98 / 54.00 ms │              52.62 / 53.39 ±0.60 / 54.08 ms │     no change │
│ QQuery 98 │        43.14 / 44.03 ±0.57 / 44.86 ms │              43.80 / 45.33 ±0.85 / 46.33 ms │     no change │
│ QQuery 99 │        65.57 / 66.22 ±0.39 / 66.60 ms │              68.51 / 68.74 ±0.19 / 69.03 ms │     no change │
└───────────┴───────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 9309.96ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 9516.18ms │
│ Average Time (HEAD)                                        │   94.04ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   96.12ms │
│ Queries Faster                                             │         1 │
│ Queries Slower                                             │        11 │
│ Queries with No Change                                     │        87 │
│ Queries with Failure                                       │         0 │
└────────────────────────────────────────────────────────────┴───────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	50.0s
Peak memory	2.1 GiB
Avg memory	1.3 GiB
CPU user	201.9s
CPU sys	5.7s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	50.0s
Peak memory	1.9 GiB
Avg memory	1.2 GiB
CPU user	204.7s
CPU sys	5.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-30T00:21:07Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark tpch10
env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃      HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │ 341.09 ms │                                   351.67 ms │    no change │
│ QQuery 2  │  91.03 ms │                                    95.58 ms │    no change │
│ QQuery 3  │ 224.94 ms │                                   241.70 ms │ 1.07x slower │
│ QQuery 4  │ 119.85 ms │                                   127.91 ms │ 1.07x slower │
│ QQuery 5  │ 364.88 ms │                                   400.37 ms │ 1.10x slower │
│ QQuery 6  │ 130.02 ms │                                   144.68 ms │ 1.11x slower │
│ QQuery 7  │ 512.27 ms │                                   540.09 ms │ 1.05x slower │
│ QQuery 8  │ 383.78 ms │                                   389.75 ms │    no change │
│ QQuery 9  │ 545.46 ms │                                   615.95 ms │ 1.13x slower │
│ QQuery 10 │ 305.87 ms │                                   322.28 ms │ 1.05x slower │
│ QQuery 11 │  67.46 ms │                                    64.51 ms │    no change │
│ QQuery 12 │ 160.03 ms │                                   165.79 ms │    no change │
│ QQuery 13 │ 329.85 ms │                                   315.19 ms │    no change │
│ QQuery 14 │ 178.04 ms │                                   190.72 ms │ 1.07x slower │
│ QQuery 15 │ 318.51 ms │                                   344.30 ms │ 1.08x slower │
│ QQuery 16 │  69.49 ms │                                    73.42 ms │ 1.06x slower │
│ QQuery 17 │ 664.66 ms │                                   677.74 ms │    no change │
│ QQuery 18 │ 830.36 ms │                                   917.98 ms │ 1.11x slower │
│ QQuery 19 │ 244.54 ms │                                   283.72 ms │ 1.16x slower │
│ QQuery 20 │ 292.95 ms │                                   307.93 ms │ 1.05x slower │
│ QQuery 21 │ 713.93 ms │                                   770.95 ms │ 1.08x slower │
│ QQuery 22 │  60.45 ms │                                    66.10 ms │ 1.09x slower │
└───────────┴───────────┴─────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 6949.47ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 7408.30ms │
│ Average Time (HEAD)                                        │  315.89ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  336.74ms │
│ Queries Faster                                             │         0 │
│ Queries Slower                                             │        15 │
│ Queries with No Change                                     │         7 │
│ Queries with Failure                                       │         0 │
└────────────────────────────────────────────────────────────┴───────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                               HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │  341.09 / 343.39 ±3.25 / 349.75 ms │           351.67 / 357.04 ±3.60 / 362.61 ms │    no change │
│ QQuery 2  │     91.03 / 94.31 ±1.79 / 96.38 ms │            95.58 / 100.64 ±7.56 / 115.55 ms │ 1.07x slower │
│ QQuery 3  │  224.94 / 225.89 ±0.99 / 227.43 ms │           241.70 / 251.76 ±6.85 / 259.72 ms │ 1.11x slower │
│ QQuery 4  │  119.85 / 121.28 ±1.00 / 122.58 ms │           127.91 / 131.28 ±2.60 / 135.27 ms │ 1.08x slower │
│ QQuery 5  │  364.88 / 374.92 ±8.72 / 387.48 ms │           400.37 / 412.51 ±7.53 / 421.01 ms │ 1.10x slower │
│ QQuery 6  │  130.02 / 132.88 ±2.39 / 136.59 ms │           144.68 / 147.60 ±2.47 / 152.15 ms │ 1.11x slower │
│ QQuery 7  │ 512.27 / 526.94 ±10.21 / 541.24 ms │          540.09 / 559.34 ±15.86 / 585.94 ms │ 1.06x slower │
│ QQuery 8  │  383.78 / 386.32 ±2.30 / 390.26 ms │          389.75 / 403.46 ±18.11 / 437.72 ms │    no change │
│ QQuery 9  │  545.46 / 556.04 ±5.69 / 561.44 ms │           615.95 / 625.10 ±9.31 / 637.47 ms │ 1.12x slower │
│ QQuery 10 │  305.87 / 317.07 ±8.97 / 331.52 ms │          322.28 / 347.62 ±17.73 / 372.57 ms │ 1.10x slower │
│ QQuery 11 │     67.46 / 72.46 ±8.01 / 88.42 ms │            64.51 / 72.51 ±14.42 / 101.27 ms │    no change │
│ QQuery 12 │  160.03 / 170.14 ±7.82 / 182.21 ms │           165.79 / 176.61 ±9.67 / 191.40 ms │    no change │
│ QQuery 13 │  329.85 / 341.79 ±8.88 / 357.05 ms │          315.19 / 333.20 ±14.00 / 356.95 ms │    no change │
│ QQuery 14 │  178.04 / 182.65 ±2.63 / 186.16 ms │           190.72 / 196.50 ±5.62 / 206.28 ms │ 1.08x slower │
│ QQuery 15 │  318.51 / 324.50 ±5.87 / 335.05 ms │          344.30 / 382.64 ±21.12 / 405.97 ms │ 1.18x slower │
│ QQuery 16 │     69.49 / 74.23 ±4.44 / 81.63 ms │              73.42 / 75.72 ±1.51 / 77.78 ms │    no change │
│ QQuery 17 │ 664.66 / 679.17 ±13.16 / 703.92 ms │          677.74 / 742.10 ±46.56 / 796.52 ms │ 1.09x slower │
│ QQuery 18 │ 830.36 / 881.49 ±35.12 / 920.60 ms │          917.98 / 941.71 ±16.13 / 960.46 ms │ 1.07x slower │
│ QQuery 19 │ 244.54 / 261.63 ±17.72 / 295.44 ms │          283.72 / 300.61 ±13.55 / 324.94 ms │ 1.15x slower │
│ QQuery 20 │ 292.95 / 305.20 ±11.79 / 326.16 ms │          307.93 / 334.29 ±20.85 / 356.36 ms │ 1.10x slower │
│ QQuery 21 │ 713.93 / 732.67 ±18.14 / 766.05 ms │          770.95 / 791.60 ±13.61 / 803.64 ms │ 1.08x slower │
│ QQuery 22 │     60.45 / 64.93 ±3.40 / 70.51 ms │             66.10 / 72.25 ±10.20 / 92.62 ms │ 1.11x slower │
└───────────┴────────────────────────────────────┴─────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 7169.88ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 7756.10ms │
│ Average Time (HEAD)                                        │  325.90ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │  352.55ms │
│ Queries Faster                                             │         0 │
│ Queries Slower                                             │        16 │
│ Queries with No Change                                     │         6 │
│ Queries with Failure                                       │         0 │
└────────────────────────────────────────────────────────────┴───────────┘

Resource Usage

tpch10 — base (merge-base)

Metric	Value
Wall time	40.0s
Peak memory	5.0 GiB
Avg memory	1.4 GiB
CPU user	364.4s
CPU sys	21.9s
Peak spill	0 B

tpch10 — branch

Metric	Value
Wall time	40.0s
Peak memory	4.6 GiB
Avg memory	2.0 GiB
CPU user	379.2s
CPU sys	26.0s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-30T00:25:26Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

Comparing claude/datafusion-rowgroup-buffering-ec9abc (02429ca) to 953518c (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  SIMULATE_LATENCY: "false"
changed:
  env:
    DATAFUSION_EXECUTION_PARQUET_READ_AHEAD_BYTES: "104857600"

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.25 ms │                                     1.22 ms │    no change │
│ QQuery 1  │   12.09 ms │                                    12.51 ms │    no change │
│ QQuery 2  │   36.86 ms │                                    38.36 ms │    no change │
│ QQuery 3  │   31.37 ms │                                    32.38 ms │    no change │
│ QQuery 4  │  232.33 ms │                                   238.75 ms │    no change │
│ QQuery 5  │  271.25 ms │                                   271.12 ms │    no change │
│ QQuery 6  │    1.27 ms │                                     1.32 ms │    no change │
│ QQuery 7  │   12.78 ms │                                    14.01 ms │ 1.10x slower │
│ QQuery 8  │  332.91 ms │                                   336.50 ms │    no change │
│ QQuery 9  │  468.45 ms │                                   464.07 ms │    no change │
│ QQuery 10 │   66.88 ms │                                    65.26 ms │    no change │
│ QQuery 11 │   73.62 ms │                                    76.26 ms │    no change │
│ QQuery 12 │  257.64 ms │                                   263.73 ms │    no change │
│ QQuery 13 │  358.18 ms │                                   359.94 ms │    no change │
│ QQuery 14 │  274.26 ms │                                   279.79 ms │    no change │
│ QQuery 15 │  278.09 ms │                                   287.22 ms │    no change │
│ QQuery 16 │  614.65 ms │                                   626.78 ms │    no change │
│ QQuery 17 │  621.51 ms │                                   643.03 ms │    no change │
│ QQuery 18 │ 1284.48 ms │                                  1300.15 ms │    no change │
│ QQuery 19 │   27.52 ms │                                    28.98 ms │ 1.05x slower │
│ QQuery 20 │  521.94 ms │                                   527.45 ms │    no change │
│ QQuery 21 │  505.49 ms │                                   528.28 ms │    no change │
│ QQuery 22 │ 1026.18 ms │                                  1028.47 ms │    no change │
│ QQuery 23 │ 3176.94 ms │                                  3139.55 ms │    no change │
│ QQuery 24 │   40.44 ms │                                    42.62 ms │ 1.05x slower │
│ QQuery 25 │  105.52 ms │                                   106.01 ms │    no change │
│ QQuery 26 │   41.36 ms │                                    42.72 ms │    no change │
│ QQuery 27 │  506.65 ms │                                   528.80 ms │    no change │
│ QQuery 28 │ 2868.81 ms │                                  2893.35 ms │    no change │
│ QQuery 29 │   41.35 ms │                                    43.94 ms │ 1.06x slower │
│ QQuery 30 │  301.66 ms │                                   322.49 ms │ 1.07x slower │
│ QQuery 31 │  276.66 ms │                                   293.39 ms │ 1.06x slower │
│ QQuery 32 │  995.29 ms │                                  1144.70 ms │ 1.15x slower │
│ QQuery 33 │ 1564.90 ms │                                  1621.14 ms │    no change │
│ QQuery 34 │ 1569.58 ms │                                  1641.18 ms │    no change │
│ QQuery 35 │  314.92 ms │                                   306.67 ms │    no change │
│ QQuery 36 │   66.65 ms │                                    72.08 ms │ 1.08x slower │
│ QQuery 37 │   34.60 ms │                                    35.75 ms │    no change │
│ QQuery 38 │   40.62 ms │                                    41.61 ms │    no change │
│ QQuery 39 │  144.27 ms │                                   140.54 ms │    no change │
│ QQuery 40 │   14.01 ms │                                    16.34 ms │ 1.17x slower │
│ QQuery 41 │   13.34 ms │                                    15.59 ms │ 1.17x slower │
│ QQuery 42 │   12.97 ms │                                    15.00 ms │ 1.16x slower │
└───────────┴────────────┴─────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 19441.53ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 19889.04ms │
│ Average Time (HEAD)                                        │   452.13ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   462.54ms │
│ Queries Faster                                             │          0 │
│ Queries Slower                                             │         11 │
│ Queries with No Change                                     │         32 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and claude_datafusion-rowgroup-buffering-ec9abc
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ claude_datafusion-rowgroup-buffering-ec9abc ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.25 / 4.03 ±5.46 / 14.96 ms │                1.22 / 4.07 ±5.59 / 15.25 ms │     no change │
│ QQuery 1  │         12.09 / 12.13 ±0.05 / 12.23 ms │              12.51 / 12.80 ±0.24 / 13.11 ms │  1.06x slower │
│ QQuery 2  │         36.86 / 37.16 ±0.31 / 37.63 ms │              38.36 / 38.63 ±0.22 / 38.98 ms │     no change │
│ QQuery 3  │         31.37 / 32.09 ±0.76 / 33.55 ms │              32.38 / 32.91 ±0.50 / 33.72 ms │     no change │
│ QQuery 4  │      232.33 / 237.92 ±2.98 / 241.08 ms │           238.75 / 241.49 ±2.03 / 244.74 ms │     no change │
│ QQuery 5  │      271.25 / 278.31 ±4.26 / 283.52 ms │           271.12 / 277.97 ±6.67 / 288.65 ms │     no change │
│ QQuery 6  │            1.27 / 1.43 ±0.24 / 1.90 ms │                 1.32 / 1.46 ±0.23 / 1.91 ms │     no change │
│ QQuery 7  │         12.78 / 13.13 ±0.20 / 13.38 ms │              14.01 / 14.40 ±0.33 / 14.86 ms │  1.10x slower │
│ QQuery 8  │      332.91 / 335.51 ±1.44 / 337.03 ms │          336.50 / 347.38 ±10.66 / 364.55 ms │     no change │
│ QQuery 9  │      468.45 / 477.86 ±9.49 / 494.10 ms │           464.07 / 478.46 ±8.02 / 484.64 ms │     no change │
│ QQuery 10 │         66.88 / 68.36 ±0.80 / 69.02 ms │            65.26 / 72.72 ±14.22 / 101.14 ms │  1.06x slower │
│ QQuery 11 │         73.62 / 76.61 ±2.15 / 79.26 ms │              76.26 / 77.73 ±2.17 / 82.05 ms │     no change │
│ QQuery 12 │      257.64 / 263.67 ±4.42 / 269.25 ms │           263.73 / 271.02 ±7.79 / 282.12 ms │     no change │
│ QQuery 13 │     358.18 / 369.94 ±11.98 / 389.68 ms │          359.94 / 386.29 ±18.78 / 413.79 ms │     no change │
│ QQuery 14 │      274.26 / 278.13 ±2.48 / 281.07 ms │           279.79 / 290.53 ±8.51 / 299.88 ms │     no change │
│ QQuery 15 │      278.09 / 288.01 ±7.55 / 301.19 ms │          287.22 / 320.99 ±28.05 / 360.13 ms │  1.11x slower │
│ QQuery 16 │     614.65 / 631.20 ±12.13 / 651.45 ms │           626.78 / 639.28 ±8.44 / 649.49 ms │     no change │
│ QQuery 17 │     621.51 / 638.14 ±12.01 / 651.52 ms │          643.03 / 664.04 ±23.51 / 708.14 ms │     no change │
│ QQuery 18 │  1284.48 / 1315.96 ±22.54 / 1348.79 ms │       1300.15 / 1328.47 ±26.76 / 1375.97 ms │     no change │
│ QQuery 19 │         27.52 / 27.75 ±0.13 / 27.89 ms │             28.98 / 46.30 ±20.84 / 74.23 ms │  1.67x slower │
│ QQuery 20 │      521.94 / 532.68 ±8.48 / 543.37 ms │           527.45 / 532.15 ±4.93 / 541.09 ms │     no change │
│ QQuery 21 │      505.49 / 520.26 ±7.50 / 525.55 ms │           528.28 / 546.83 ±9.91 / 556.82 ms │  1.05x slower │
│ QQuery 22 │  1026.18 / 1053.68 ±23.65 / 1096.16 ms │       1028.47 / 1058.34 ±18.94 / 1086.25 ms │     no change │
│ QQuery 23 │  3176.94 / 3236.91 ±48.03 / 3305.24 ms │       3139.55 / 3300.79 ±85.95 / 3387.31 ms │     no change │
│ QQuery 24 │     40.44 / 159.25 ±203.84 / 565.95 ms │          42.62 / 122.63 ±115.78 / 341.79 ms │ +1.30x faster │
│ QQuery 25 │      105.52 / 107.68 ±1.62 / 110.42 ms │         106.01 / 161.51 ±108.25 / 378.01 ms │  1.50x slower │
│ QQuery 26 │         41.36 / 41.81 ±0.27 / 42.12 ms │              42.72 / 43.45 ±0.82 / 44.93 ms │     no change │
│ QQuery 27 │     506.65 / 522.16 ±11.99 / 538.58 ms │           528.80 / 534.62 ±3.65 / 539.17 ms │     no change │
│ QQuery 28 │  2868.81 / 2942.74 ±59.19 / 3045.99 ms │       2893.35 / 2913.95 ±18.82 / 2945.48 ms │     no change │
│ QQuery 29 │        41.35 / 51.84 ±12.79 / 73.01 ms │             43.94 / 50.99 ±13.14 / 77.26 ms │     no change │
│ QQuery 30 │     301.66 / 319.54 ±15.75 / 343.79 ms │          322.49 / 334.60 ±13.54 / 360.26 ms │     no change │
│ QQuery 31 │     276.66 / 290.86 ±10.80 / 307.63 ms │           293.39 / 304.22 ±9.51 / 315.50 ms │     no change │
│ QQuery 32 │   995.29 / 1055.61 ±54.09 / 1156.42 ms │      1144.70 / 1262.93 ±137.34 / 1497.96 ms │  1.20x slower │
│ QQuery 33 │  1564.90 / 1688.62 ±99.49 / 1841.02 ms │       1621.14 / 1672.45 ±54.32 / 1777.34 ms │     no change │
│ QQuery 34 │ 1569.58 / 1738.94 ±248.02 / 2230.29 ms │       1641.18 / 1664.71 ±30.47 / 1720.77 ms │     no change │
│ QQuery 35 │    314.92 / 429.93 ±208.04 / 845.74 ms │          306.67 / 360.61 ±58.30 / 433.48 ms │ +1.19x faster │
│ QQuery 36 │         66.65 / 72.41 ±4.47 / 79.35 ms │             72.08 / 82.38 ±11.00 / 95.83 ms │  1.14x slower │
│ QQuery 37 │         34.60 / 34.84 ±0.28 / 35.39 ms │              35.75 / 38.58 ±2.49 / 43.06 ms │  1.11x slower │
│ QQuery 38 │        40.62 / 54.98 ±19.61 / 93.26 ms │              41.61 / 48.28 ±5.06 / 56.86 ms │ +1.14x faster │
│ QQuery 39 │     144.27 / 156.99 ±10.73 / 172.76 ms │          140.54 / 155.48 ±11.83 / 173.52 ms │     no change │
│ QQuery 40 │         14.01 / 16.14 ±3.09 / 22.19 ms │              16.34 / 18.59 ±3.27 / 25.08 ms │  1.15x slower │
│ QQuery 41 │        13.34 / 23.30 ±13.60 / 49.42 ms │              15.59 / 18.56 ±4.68 / 27.89 ms │ +1.26x faster │
│ QQuery 42 │         12.97 / 16.66 ±6.67 / 29.99 ms │              15.00 / 16.22 ±1.66 / 19.49 ms │     no change │
└───────────┴────────────────────────────────────────┴─────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                          ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                          │ 20455.17ms │
│ Total Time (claude_datafusion-rowgroup-buffering-ec9abc)   │ 20789.82ms │
│ Average Time (HEAD)                                        │   475.70ms │
│ Average Time (claude_datafusion-rowgroup-buffering-ec9abc) │   483.48ms │
│ Queries Faster                                             │          4 │
│ Queries Slower                                             │         11 │
│ Queries with No Change                                     │         28 │
│ Queries with Failure                                       │          0 │
└────────────────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	105.0s
Peak memory	17.2 GiB
Avg memory	5.9 GiB
CPU user	999.4s
CPU sys	98.3s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	105.0s
Peak memory	18.0 GiB
Avg memory	6.4 GiB
CPU user	1012.3s
CPU sys	108.3s
Peak spill	0 B

File an issue against this benchmark runner

…robin (apache#25809) ## Which issue does this PR close? - No issue. Found while benchmarking apache#24086. ## Rationale for this change With `SIMULATE_LATENCY=true`, a benchmark result can change when a PR changes the number of object store requests in *other* queries, or off the critical path of the same query. `LatencyObjectStore` gives each request the next entry of a 20-entry latency table, with one counter for the whole process: ```rust let idx = self.get_counter.fetch_add(1, Ordering::Relaxed) % GET_LATENCIES_MS.len(); ``` So the latency of a request depends on how many requests came before it, in all earlier queries of the run. The simulator is deterministic, so the same wrong result comes back on every rerun. Example from apache#24086 (TPC-DS SF1, `SIMULATE_LATENCY=true`): the bot reported Q27 1.55x slower (141 → 218 ms) and Q44 1.38x slower. Both queries have only two serial round trips on the critical path. Q44 makes identical requests with and without the change. Local runs, median of 9, ms: | latency model | Q27 base | Q27 PR | Q44 base | Q44 PR | |---|---|---|---|---| | fixed 50 ms per request | 111 | 111 | 114 | 121 | | fixed 100 ms per request | 216 | 211 | – | – | | current round-robin table | 238 | 288 | 212 | 211 | A sweep over the 20 possible start positions of the counter (100 iterations each) gives Q27 medians of 239 ms (base) and 241 ms (PR). The min-of-5 that the bot reports depends on the start position: the base build gets a lucky minimum at 11 of 20 positions, the PR build at 5. ## What changes are included in this PR? `LatencyObjectStore` draws each GET and LIST latency independently at random from the same 20-entry distributions. The distributions do not change. | approach | latency of a request depends on | result | |---|---|---| | shared counter (before) | the number of earlier requests in the process | a change in one query moves latencies in other queries | | hash of the request (not chosen) | the request's path and byte range | a request shape gets the same latency in every iteration, so builds with different request shapes do not average out | | independent random draw (this PR) | nothing | each request samples the same distribution, and iterations average out | ## What is the testing strategy for this PR? This change affects only the benchmark harness. - With this change, Q27 over 99 iterations gives base median 217 ms and PR median 224 ms (mean 223.5 ± 7.5 vs 217.6 ± 6.9), so there is no difference, as with a fixed latency. - Queries that do read data keep their differences: TPC-DS Q3 703 → 353 ms, Q7 1078 → 526 ms, TPC-H Q1 1006 → 209 ms, Q6 1862 → 218 ms (median of 5, same two builds). - `cargo test -p datafusion-benchmarks --lib` and clippy pass. There is no new unit test: the property ("no dependency on earlier requests") is structural, since the store has no state left. ## Are there any user-facing changes? No. Benchmark runs with `SIMULATE_LATENCY=true` are no longer bit-for-bit reproducible between runs. Queries with few serial round trips have more spread per iteration (Q27 SD about 70 ms), so compare them with more iterations or with means. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

EXPERIMENT: pin every arrow crate to pydantic/arrow-rs claude/push-decoder-batch-granular-scan-plan-60.0.0: the 60.0.0 release plus batch-granular decoding in ParquetPushDecoder (apache/arrow-rs#6946) and ParquetPushDecoder::scan_plan (apache/arrow-rs#10555). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

… read-ahead Opt-in with datafusion.execution.parquet.read_ahead_bytes (default unset). If set, the scan builds the same ParquetPushDecoder as the default path with FetchGranularity::Batch and drives it with try_decode. ReadAhead fetches ranges from scan_plan() in the background within that many bytes. datafusion.execution.parquet.read_ahead_conditional (default false) also reads ahead ranges that a pushed-down filter can make unnecessary. Filtered scans, runtime row-group pruning and the per-row-group filter toggle use the same path as main. Squashed from the history in branch claude/datafusion-rowgroup-buffering-history (pydantic/datafusion). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…ranges Skipping ranges that a pushed-down filter can make unnecessary made filtered scans on object storage 2-16x slower. Read-ahead now fetches them, bounded by the read-ahead window. Proto field 40 is reserved. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Each read-ahead stream registers a `ParquetReadAhead[partition]` consumer. The reservation follows the bytes the decoder holds plus the bytes in flight. Bytes the decoder asks for are always reserved (grow). Speculative read-ahead takes only what the pool can grant (try_grow); ranges that do not fit stay pending. `FileSource::create_morselizer_with_context` gives the source the scan's `TaskContext`. The default delegates to `create_morselizer`. `ParquetSource` uses it to get the memory pool. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…time A scan with `read_ahead_bytes` set had the same metrics as a scan without it, so EXPLAIN ANALYZE could not show that read-ahead was active or whether it kept ahead of the decoder. - `read_ahead_bytes_fetched`: bytes fetched before the decoder asked for them. - `read_ahead_wait_time`: time the decoder waited for byte ranges that it needed. Both are registered only for a file that is read with read-ahead.

adriangb · 2026-10-08T19:04:06Z

Possible follow-ups (prototypes, not ready for review):

Eager first fetch. A file starts its first read-ahead fetch when its stream is built. Thus, read-ahead continues across file boundaries when files are smaller than the window. adriangb@88c6304
Prefetch budget. One byte budget for each scan partition limits the data that queued files fetch early. adriangb@64e67f5

github-actions Bot added the datasource Changes to the datasource crate label Aug 4, 2026

adriangb changed the title ~~POC: pipelined (overlapped) row-group prefetch with a bounded byte budget~~ POC: pipelined and streaming (batch-granular) parquet scan fetch policies Aug 4, 2026

This was referenced Sep 29, 2026

[bench] combine apache#24086 (read-ahead) + apache#25752 (optional filters) pydantic/datafusion#82

Closed

[bench] combine apache#24086 (read-ahead) + apache#25752 (optional filters) adriangb/datafusion#17

Draft

adriangb force-pushed the claude/datafusion-rowgroup-buffering-ec9abc branch from 02429ca to 5b7407e Compare October 5, 2026 01:51

adriangb and others added 3 commits October 4, 2026 22:09

adriangb force-pushed the claude/datafusion-rowgroup-buffering-ec9abc branch from 5b7407e to d6d54fd Compare October 5, 2026 02:09

Conversation

adriangb commented Aug 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Changes in this revision

What this does not do

Are these changes tested?

Earlier results

Are there any user-facing changes?

Uh oh!

adriangb commented Aug 4, 2026

Uh oh!

adriangb commented Aug 4, 2026

Uh oh!

adriangb commented Aug 4, 2026

Uh oh!

adriangb commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangb commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Aug 4, 2026

Uh oh!

adriangbot commented Sep 29, 2026

Uh oh!

adriangbot commented Sep 29, 2026

Uh oh!

adriangbot commented Sep 29, 2026

Uh oh!

adriangbot commented Sep 29, 2026

Uh oh!

adriangbot commented Sep 29, 2026

Uh oh!

adriangbot commented Sep 29, 2026

Uh oh!

adriangb commented Aug 4, 2026 •

edited

Loading

Benchmark summary: `02429ca` (arrow-rs pin with apache/arrow-rs#11241 in place of apache/arrow-rs#11223)

With `SIMULATE_LATENCY`, `pushdown_filters=false`

With `SIMULATE_LATENCY`, `pushdown_filters=true`