GPU + CPU Miner for CapStash, Xelis, Warthog & Monero (RandomX) Works on NVIDIA and AMD GPUs. CPU mining for any rx/0 RandomX coin. Pool and solo. HiveOS ready. Dual GPU+CPU mining in one process.

Heads-up for v1.3.0a users: v1.4.0 ships a single unified Linux tarball that replaces the old compat + rtx50 split. The new build has CUDA 12.8 fatbins (native sm_120 SASS for RTX 50) but stays linked under CUDA 11.8 so the driver floor remains 525+ — same broad HiveOS compat as the old compat tarball, plus full RTX 50 support, in one file.

Open http://localhost:4068 in your browser while mining. Live hashrate per algorithm, per-GPU shares, GPU temp / fan / power, and the actual CPU model when CPU mining is active. Side-by-side primary/secondary cards in dual mining mode.

CapStash (GPU):

dankminer -a capstash -w YOUR_ADDRESS -p stratum+tcp://1miner.net:3691 --worker rig1

Xelis (GPU):

dankminer -a xelis -w YOUR_ADDRESS -p stratum+tcp://1miner.net:4073

Warthog (GPU + CPU, JanusHash):

dankminer -a warthog -w YOUR_ADDRESS -p stratum+tcp://1miner.net:4200

Monero / XMR (CPU only, RandomX rx/0):

dankminer -a xmr -w YOUR_XMR_ADDRESS -p stratum+tcp://1miner.net:3333

Solo CapStash (against your own capstashd RPC):

dankminer -a capstash -w YOUR_ADDRESS -p http://user:pass@127.0.0.1:33333

Add --xmr-wallet ADDR to any GPU primary command to mine RandomX rx/0 on the CPU at the same time. One process, one console, one dashboard.

dankminer -a capstash -w YOUR_CAP_ADDRESS -p stratum+tcp://1miner.net:3691 \
          --xmr-wallet YOUR_XMR_ADDRESS --xmr-threads 12

• GPU mines CapStash / Xelis / Warthog as the primary algorithm
• CPU mines RandomX rx/0 as a secondary passenger thread
• Each algo keeps its own dev-fee schedule independently
• Dashboard shows both side by side; console prefixes XMR lines with [XMR]

--xmr-pool accepts any rx/0 stratum URL — Monero, Zephyr, Salvium, Townforge, and any other coin using standard RandomX rx/0 parameters. Variants with non-default RandomX parameters (Wownero rx/wow, ArQmA rx/arq) are not supported.

Custom rx/0 pool example:

dankminer -a capstash -w YOUR_CAP_ADDRESS -p stratum+tcp://1miner.net:3691 \
          --xmr-wallet YOUR_RX_ADDRESS \
          --xmr-pool stratum+tcp://your-rx0-pool.example:3333 \
          --xmr-threads 12

Threading rule of thumb: RandomX scales by physical cores, not logical threads. On a 12-core / 24-thread CPU, --xmr-threads 11 or 12 will outperform --xmr-threads 23. Less is more.

  -a ALGO          Algorithm: capstash, xelis, warthog, xmr
  -w WALLET        Wallet/payout address for the primary algo
  -p URL           Pool or RPC URL for the primary algo
  -W NAME          Worker name
  -wl FILE         Wallet list for rotation
  --force-opencl   Force OpenCL (for AMD GPUs if auto-detect doesn't work)
  --no-cuda        Skip CUDA entirely — OpenCL drives every GPU
  --no-ocl         Skip OpenCL entirely — CUDA only
  --cpu-cores N    Pin the miner to the first N logical CPU cores
  --cpu-affinity LIST   Pin process to specific cores (e.g. 0,1 or 0-3)

Dual-mining flags (only valid with a GPU primary):

  --xmr-wallet ADDR    Run RandomX rx/0 on the CPU alongside the primary
  --xmr-pool URL       RandomX pool URL (default stratum+tcp://1miner.net:3333)
  --xmr-worker NAME    Worker name for XMR (default: same as -W)
  --xmr-threads N      CPU threads for XMR (default: hw cores - 1)
  --xmr-light          Light-mode RandomX (256 MiB, ~10x slower than full)

NVIDIA: GTX 1060 through RTX 5090 (CUDA, embedded fatbin) AMD: RX 470/480/570/580, Vega, RX 5000/6000/7000/9000 series (OpenCL) Intel: Arc (OpenCL, experimental) CPU: Any x86-64 with AES-NI for RandomX (XMR / dual mining)

AMD GPUs are auto-detected — no extra config needed. Hybrid NVIDIA + AMD rigs run in the same instance with each card on its native backend.

Examples:

stratum+tcp://1miner.net:3691          # CapStash US pool
stratum+tcp://eu1.1miner.net:3691      # CapStash EU pool
stratum+tcp://sgp.1miner.net:3691      # CapStash Singapore pool
stratum+tcp://1miner.net:3791          # CapStash US solo
stratum+tcp://1miner.net:4073          # Xelis
stratum+tcp://1miner.net:4200          # Warthog
stratum+tcp://1miner.net:3333          # XMR / RandomX

DankMiner runs as a custom miner on HiveOS. From the flight sheet → Miner → "More miners…" → Setup miner config opens this dialog:

Important: the miner name must be exactly dankminer — not dankminerV1.4.0 or anything else. The miner will fail to launch if the name doesn't match what's inside the tarball.

Dual mining on HiveOS — add the XMR flags to the Extra config arguments field:

--xmr-wallet YOUR_XMR_ADDRESS --xmr-threads 12

The same single tarball (dankminer-1.4.0.tar.gz) covers every supported GPU including RTX 50. No separate rtx50 download anymore — Option C unified build.

Use the server closest to you: 1miner.net (US), eu1.1miner.net (EU), or sgp.1miner.net (Singapore).

In dual mining each algo runs its own dev-fee schedule independently — they don't both go to dev fee at the same time.

"CUDA err 999 / unknown error" — GPU driver reset. DankMiner recovers automatically. If it happens frequently on Windows, run install_tdr_fix.bat as Administrator and reboot to raise the Windows GPU watchdog timeout.

"no kernel image available" — Your driver is older than the embedded fatbin needs. Update your NVIDIA driver to 525+ (CUDA 11.8 driver floor).

"CUDA not available — trying OpenCL" — Normal on AMD-only rigs. No action needed.

"GLIBC not found" — Use the HiveOS build (dankminer-1.4.0.tar.gz) instead of the Linux desktop build.

Low GPU hashrate — Check temps, riser cables, and power delivery. Whirlpool is core-heavy — boost core clock, drop memory clock.

Low XMR / RandomX hashrate — RandomX hates SMT and oversubscription. Use --xmr-threads N where N ≈ physical cores (not logical threads). On a 12c/24t CPU try --xmr-threads 12, not 23. Also check that hugepages are enabled (cat /proc/sys/vm/nr_hugepages should be > 0) — RandomX loses 30-50% without them.

"GPU has fallen off the bus" — Hardware issue. Power cycle the rig. If it repeats, replace the riser on that GPU slot.

Duplicate share rejects on big-CPU rigs — Fixed in v1.4.0 (nonce-slice overflow on >64 worker threads). Update from any earlier dual-mining build.

• Monero / XMR support via vendored tevador/RandomX v1.2.1 — the same engine xmrig uses
• Dual mining — GPU primary + RandomX rx/0 on CPU in one process, one dashboard
• Any rx/0 coin — --xmr-pool is not Monero-locked; works with Zephyr, Salvium, Townforge, and any other coin using standard rx/0
• Redesigned web dashboard — flat black palette, embedded DankMiner logo, side-by-side dual-mining cards
• Embedded exe icon — Windows Explorer / taskbar / alt-tab now show the DankMiner logo
• CPU model detection — dashboard and console show the actual chip brand (Ryzen 7 7800X3D, EPYC 9754, Core i9-13900K) instead of a thread-count placeholder
• Single unified Linux tarball — replaces the prior compat + rtx50 split. Same binary works on every supported GPU.
• Bug fixes — dual-mining HR-thread race fixed (no more KH/s flicker into the GPU hashrate slot), nonce-slice overflow fixed for >64-thread CPUs, Monero stratum target endianness corrected to xmrig-compatible LE u32/u64

• Three GPU algorithms in one binary — CapStash, Xelis, Warthog
• Multi-GPU support on all GPU algorithms
• Hybrid CUDA + OpenCL on CapStash for mixed-vendor rigs
• Per-GPU stats — every share logged with which card found it, plus per-GPU accept/reject in the dashboard
• Per-GPU backend selection — --cuda=0,1, --ocl=0, --no-cuda, --no-ocl
• CPU control flags — --cpu-cores, --cpu-affinity, plus matching config.txt support
• Improved reconnect handling and smoother multi-GPU share reporting
• config.txt fields: worker, cpu_threads, cpu_cores, cpu_affinity, device filters

• Pool: 1miner.net
• Discord: discord.gg/YZXGEa9RhK
• CapStash Core: github.com/CapStash/CapStash-Core

(c) 2026 DankMiner / 1Miner.net