Pearl (PRL) & Quantus (QTC) miner for AMD & NVIDIA GPUs (Quantus also runs on Intel Arc A-series) by pool.kryptex.com.
Quick start:
- Download krig from releases and subscribe for updates.
- Connect to Kryptex Pool:
./krig-miner --url prl.kryptex.network:8048 --user WALLET/WORKER
See all the supported GPUs or check our hashrate database.
These are just examples. Other AMD & NVIDIA GPUs are supported, too. Hashrate values may be outdated.
| GPU Model | Architecture | Arch | Hashrate |
|---|---|---|---|
| AMD Instinct MI355X | CDNA 4 | gfx950 | ~550 TH/s |
| AMD Instinct MI300X | CDNA 3 | gfx942 | ~318 TH/s |
| AMD Radeon RX 9070 XT | RDNA 4 | gfx1201 | ~106 TH/s |
| AMD Radeon RX 9060 XT | RDNA 4 | gfx1200 | ~45 TH/s |
| AMD Radeon RX 7900 XT | RDNA 3 | gfx1100 | ~41.4 TH/s |
| AMD Radeon RX 7600 | RDNA 3 | gfx1102 | ~15.4 TH/s |
| AMD Radeon RX 6750 XT | RDNA 2 | gfx1031 | ~17.8 TH/s |
| AMD Radeon RX 5500 | RDNA 1 | gfx1012 | Supported |
| AMD Radeon VII | GCN 5 | gfx906 | Supported |
| AMD Instinct MI50/MI60 | GCN 5 | gfx906 | Supported |
| GPU Model | Architecture | Arch | Hashrate |
|---|---|---|---|
| NVIDIA B300 | Blackwell | sm_103 | ~431 TH/s |
| NVIDIA RTX PRO 6000 | Blackwell | sm_120 | ~422 TH/s |
| NVIDIA GeForce RTX 6000 | Blackwell | sm_120 | ~402 TH/s |
| NVIDIA GeForce RTX 5090 | Blackwell | sm_120 | ~385 TH/s |
| NVIDIA GeForce RTX 4090 | Ada Lovelace | sm_89 | ~292 TH/s |
| NVIDIA GeForce RTX 3090 | Ampere | sm_86 | Supported |
| NVIDIA L40S | Ada Lovelace | sm_89 | ~259 TH/s |
| NVIDIA RTX 6000 Ada | Ada Lovelace | sm_89 | ~201 TH/s |
| NVIDIA DGX Spark GB10 | Blackwell | sm_121 | ~100 TH/s |
| GPU Model | Architecture | Arch | Hashrate | Power |
|---|---|---|---|---|
| NVIDIA RTX PRO 6000 | Blackwell | sm_120 | 1881.1 MH/s | 600 W |
| NVIDIA GeForce RTX 5090 | Blackwell | sm_120 | 1786.8 MH/s | 575 W |
| NVIDIA GeForce RTX 4090 | Ada Lovelace | sm_89 | 1375.1 MH/s | 449 W |
| NVIDIA RTX 6000 Ada | Ada Lovelace | sm_89 | 989.1 MH/s | 299.7 W |
| NVIDIA L40S | Ada Lovelace | sm_89 | 892.0 MH/s | 348.6 W |
| NVIDIA GeForce RTX 3090 | Ampere | sm_86 | 542.0 MH/s | 305 W |
| NVIDIA DGX Spark GB10 | Blackwell | sm_121 | 460.0 MH/s | 95.7 W |
| NVIDIA CMP 50HX | Turing | sm_75 | 377.9 MH/s | 250 W |
| NVIDIA CMP 90HX | Ampere | sm_86 | 365.9 MH/s | 249 W |
| NVIDIA GeForce RTX 3060 | Ampere | sm_86 | 220.1 MH/s | 169.3 W |
| NVIDIA GeForce GTX 1660 | Turing | sm_75 | 163.9 MH/s | 128.9 W |
| NVIDIA GeForce GTX 1650 | Turing | sm_75 | 96.8 MH/s | 75 W |
| NVIDIA GeForce GTX 1080 | Pascal | sm_61 | 80.8 MH/s | 175.9 W |
| GPU Model | Architecture | Arch | Hashrate | Power |
|---|---|---|---|---|
| AMD Radeon RX 7900 XT | RDNA 3 | gfx1100 | 358.1 MH/s | 257 W |
| AMD Radeon RX 9070 XT | RDNA 4 | gfx1201 | 312.5 MH/s | 304 W |
| AMD Radeon RX 6700 XT / 6750 XT | RDNA 2 | gfx1031 | 199.6 MH/s | 196 W |
| AMD Radeon RX 7600 | RDNA 3 | gfx1102 | 77.3 MH/s | 144 W |
| GPU Model | Architecture | Arch | Hashrate | Power |
|---|---|---|---|---|
| Intel Arc A750 | Xe-HPG (Alchemist) | acm-g10 | 122.8 MH/s | 187.8 W |
All Intel Arc A-series GPUs (A770, A750, A580, A380, A310, mobile and Pro models) are supported since 1.5.10 on Windows, Linux, HiveOS and mmpOS. The Intel GPU driver with Level Zero is required (ze_loader.dll on Windows, libze_loader.so.1 on Linux). The hashrate above is measured on Linux; on Windows the A750 is currently about 25% slower. Use --no-intel to disable Intel GPUs.
usage: krig-miner [options]
options:
-o, --url <url> Pool endpoint: [scheme://]host[:port]
pearl stratum+ssl://
quantus stratum+ssl://
--coin <coin> Coin to mine
pearl default; aliases: prl, pearlhash
quantus alias: qtc
--algorithm <coin> Alias for --coin
--algo <coin> Alias for --coin
--pool <url> Alias for --url
-u, --user <wallet> Payout wallet
--wallet <wallet> Alias for --user
-p, --password <pw> Pool password
Failover pools: add --url --user --password per pool,
or --url per pool and --user --password once
--api-port <port> Enable the HTTP API on <port>
--api-host <ip> Bind interface for the API (optional; default 127.0.0.1)
e.g. --api-port 12000 --api-host 127.0.0.1
-d, --devices <list> Devices to mine on: comma-separated 0-based indices
(as shown by --list-devices), or 'all' (default —
every detected CUDA and ROCm device). 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.
--no-cuda Disable the CUDA backend (don't load CUDA DLLs or
enumerate NVIDIA GPUs). By default both backends
run and all CUDA + ROCm devices are mined.
--no-rocm Disable the ROCm backend (don't load HIP DLLs or
enumerate AMD GPUs)
--no-intel Disable the Intel Level Zero backend (don't load
ze_loader or enumerate Intel Arc GPUs; Quantus only)
--rocm-runtime <6|7> Pin the HIP runtime major; by default 6 is tried
first, then 7
--gpu-cclock <MHz> Lock NVIDIA core clock
--gpu-mclock <MHz> Lock NVIDIA memory clock
--gpu-coffset <MHz> Set NVIDIA core clock offset (signed)
--gpu-moffset <MHz> Set NVIDIA memory clock offset (signed)
--gpu-plimit <watts> Set NVIDIA board power limit
--gpu-fan <percent> Set every fan on each NVIDIA GPU (0-100)
One value applies to all selected NVIDIA GPUs;
lists follow selected-device order, '_' skips.
--gpu-no-reset-oc Keep NVIDIA settings on exit
-l, --log-level <level> Console verbosity: trace|debug|info|warn|error|
critical|none (default info). debug restores the
detailed per-job/per-share diagnostic output.
--log-file <file> log to a file
--no-tui Disable the interactive terminal UI and print logs
line by line (also automatic when redirected)
--list-devices List detected devices (index/name/pci) and exit
-h, --help Print this help and exit
-V, --version Print version and exit
Please contact us if you have any problems, questions, or suggestions:
Also, feel free to open issues on Github.