Options:
-o, --url URL Pool endpoint: [scheme://]host[:port]
scheme: stratum+ssl:// (TLS only)
--pool URL Alias for --url
-u, --user WALLET Payout wallet
--wallet WALLET Alias for --user
-p, --password PW Pool password
-h, --help Print help and exit
-V, --version Print version and exit
Device selection:
-d, --devices LIST Devices to mine on: comma-separated 0-based
indices, or 'all' (default). e.g. -d 0,2
--devices-pci LIST Devices to mine on, by PCI address: comma-separated
bus:device.function. e.g. --devices-pci 01:00.0,0a:00.0
--amd-igpu Include AMD integrated GPUs (APU/iGPU); excluded by
default. Affects -d and --list-devices indices.
--list-devices List detected devices (index/name/pci) and exit
Backend selection (by default Krig runs both backends and mines every detected NVIDIA and AMD GPU):
--no-cuda Disable the CUDA backend (don't load CUDA DLLs or
enumerate NVIDIA GPUs) — for an AMD-only rig
--no-rocm Disable the ROCm backend (don't load HIP DLLs or
enumerate AMD GPUs) — for an NVIDIA-only rig
--rocm-runtime 6|7 Pin the HIP runtime major; by default 6 is tried
first, then 7
Stats API:
--api-port PORT Enable the Prometheus /metrics exposer on PORT
--api-host IP Bind interface for the exposer (optional, default *)
e.g. --api-port 12000 --api-host 127.0.0.1
