WebGPU: enable INT64 for Equal/Sub/Where/ReduceSum under enable_int64 flag to support Gemma4#29392

Merged

feich-ms merged 15 commits into

mainfrom

user/feich/webgpu_int64_equal_sub_where_reducesum

Jul 7, 2026

Commits on Jul 3, 2026

webgpu: enable INT64 for Equal/Sub/Where/ReduceSum under enable_int64 flag

feich-ms
and
claude
committed
webgpu: address PR review comments for Equal/Sub/Where/ReduceSum INT64 changes

feich-ms
and
claude
committed
webgpu: add int64 tests and document int32-range limitation for Equal/Sub/Where/ReduceSum

feich-ms
and
claude
committed
webgpu: fix INT64 handling in BinaryElementwise and Where ops

feich-ms
and
claude
committed
webgpu: fix Equal INT64 bool output and ReduceSum test shape

feich-ms
and
claude
committed
Fix INT64+bool OOB reads in WebGPU binary elementwise shader; add bounds checks and update test to size=5

Copilot
authored and
feich-ms
committed
fix lint warnings and style consistency issues
feich-ms
committed
webgpu: fix INT64 Equal broadcast and improve code clarity in BinaryElementwise

feich-ms
and
claude
committed
webgpu: fix where.cc comment explaining INT64 exclusion from fast path

feich-ms
and
claude
committed

Commits on Jul 7, 2026

webgpu: hoist base/element_count declarations for INT64 paths and reuse in output

feich-ms
and
claude
committed