How to Mine Keryx (KRX) — Mining Setup Guide for suprnova.cc
Step-by-step guide to start mining Keryx (KRX) at krx.suprnova.cc. Works on any modern NVIDIA or AMD GPU with the latest keryx-miner-supr release. VarDiff stratum on ports 4401 (low) and 4404 (recommended), plus SSL on 4405. 1% pool fee, PPLNS payouts, automatic 20% coinbase escrow claiming on your behalf.
Stratum Servers
| Region | Host | Ports | Ping |
|---|---|---|---|
EU (Germany)
|
krx.suprnova.cc |
4404 GPU · 4401 Low · 4405 SSL | Getting Ping |
EU2 (Ukraine)
|
stratum-eu2.suprnova.cc |
4404 GPU · 4401 Low · 4405 SSL | Getting Ping |
US-East
|
stratum-us.suprnova.cc |
4404 GPU · 4401 Low · 4405 SSL | Getting Ping |
Difficulty adapts automatically (VarDiff, ~1 share / 10s). SSL uses stratum+ssl://; plain TCP uses stratum+tcp://. Pool ping explained.
Which port? 4404 for GPUs (normal) · 4401 low-difficulty for CPUs / small or slow cards · 4405 SSL. VarDiff tunes every connection, so no password is needed — and don't set d=<n>: that pins a static difficulty (d=1 = difficulty 1.0, far too hard for a small card, so it looks like "the miner never finds a share"). Leave it unset and let VarDiff work.
Configuration
- Wallet
keryx:address (CashAddr). - Workershort label per rig, e.g.
rig01. - Algorithm
keryx-hash(GPU). - Passwordnot required.
Example command line:
-a keryx:qp0vrxc0k5w0pcyem6vau2pjgztje880tsm239rywtm7l7uv2pcxzq55n8khs.rig01 -s stratum+tcp://krx.suprnova.cc:4404
Miner
The Suprnova build of the reference miner — current release v0.12.7. Download the latest release →
1 · Pick your GPU line
| Line | GPUs | Driver |
|---|---|---|
modern | sm_75+ — RTX 20 / 30 / 40 / 50 (native sm_120 for RTX 50xx) | ≥ 575 |
legacy | sm_61+ — GTX 10-series (Pascal), Tesla V100 / Volta, Turing, CMP 100–210 | ≥ 550 |
amd | AMD (OpenCL) — see the release assets for supported cards | — |
Pascal lives inside legacy — there's no separate pascal download. P100 (sm_60) is not supported (the possession walk needs __dp4a, sm_61+).
2 · Run it on your platform
# <line> = modern (RTX 20xx+) or legacy (GTX 10xx / older)
tar xzf keryx-miner-supr-<line>-v0.12.7-linux-x86_64.tar.gz
cd keryx-miner-supr-*
./keryx-miner-supr -a keryx:YOUR_ADDRESS.rig01 -s stratum+tcp://krx.suprnova.cc:4404
No tier flag needed — it defaults to --tier auto (biggest model your card fits). AMD: keep libkeryxopencl.so next to the binary.
Extract the ZIP for your line, run keryx-miner-supr.exe:
keryx-miner-supr.exe -a keryx:YOUR_ADDRESS.rig01 -s stratum+tcp://krx.suprnova.cc:4404
modern for RTX 20xx and newer, legacy for GTX 10-series and older.
Needs the NVIDIA Container Toolkit. Pool, backup and tier already default to us — only the wallet is required:
docker run -d --name keryx --gpus all --restart unless-stopped \
-e KERYX_WALLET=keryx:YOUR_ADDRESS \
-e KERYX_WORKER=rig01 \
-v keryx-models:/models \
ocminersupr/keryx-miner-supr:latest
Always mount /models or the multi-GB model re-downloads on every recreate. The image is the modern line (sm_75+) — GTX 10-series must use the legacy tarball, not Docker. Pin a version with :0.12.7. Env vars: KERYX_WALLET, KERYX_WORKER, KERYX_TIER, KERYX_MODEL_DIR, KERYX_DEVICES (e.g. 0,1), KERYX_EXTRA_ARGS.
Custom-miner flight sheet:
- Miner name
keryx-miner-supr - Installation URLthe
modernorlegacytarball URL (below) - Hash algorithm
keryx-hash - Walletyour
keryx:address - Pool URL
stratum+tcp://krx.suprnova.cc:4404 - Extra config
--hiveos(keeps models across upgrades; leave the tier on auto)
Asset names: HiveOS keryx-miner-supr-<line>-v0.12.7.tar.gz · MMPOS keryx-miner-supr-<line>-mmpos_v0.12.7.tar.gz · SimpleMiningOS keryx-miner-supr-smos-<line>-v0.12.7.tar.gz · macOS Apple Silicon keryx-miner-supr-macos-arm64-v0.12.7.tar.gz.
⏳ What to expect on first run
Keryx mining serves AI inference, so on first start the miner downloads a model (≈6 GB for the entry tier, up to ≈28 GB for the top) and builds a possession index — this takes several minutes and shows 0 hash/s (preparing). That's normal, not a stall. If no model can be served it logs OPoI: no models ready — mining suspended and won't hash, so pick a tier your VRAM fits. Models are stored in <exe>/models (override with --model-dir <path>); on HiveOS, --hiveos keeps them so they survive miner upgrades.
Model tiers & VRAM — optional, the default is auto
| Flag | Model | VRAM |
|---|---|---|
--tier auto default | biggest model that fits your card | — |
--very-light | Qwen3.5-9B | 8 GB |
--light | GLM-4-9B | 12 GB |
--default | Gemma-4-12B | 15 GB |
--high | Qwen3.6-27B | 24 GB |
--very-high | Kimi-Linear-48B | 30 GB |
Leave it on auto — it runs the largest model your card fits, which earns the highest TiPPLNS tier reward your GPU can. Only pin a tier to force a specific model.
The official reference miner from Keryx-Labs. Version v0.5.4 or newer speaks our stratum-v3 wire and is fully accepted here — mine with the stock binary, no fork required (proven live, side-by-side with our build).
Download keryx-miner (official) →Linux — extract and run (no tier flag; it auto-selects):
tar xzf keryx-miner-<os>-x86_64.tar.gz
cd keryx-miner-*
./keryx-miner -a keryx:YOUR_ADDRESS.rig01 -s stratum+tcp://krx.suprnova.cc:4404
Windows — run keryx-miner.exe:
keryx-miner.exe -a keryx:YOUR_ADDRESS.rig01 -s stratum+tcp://krx.suprnova.cc:4404
Builds older than v0.5.4 are refused at login. For Suprnova-tuned builds (legacy / GTX-10 line, Docker image, VarDiff-optimised) use the keryx-miner-supr tab.