Pearl (PRL) Pool Mining on Idle AI GPUs - Solar Powered, Proof of Useful Work
| Rig | GPU | Power Limit | Tuning | Efficiency | Live Hashrate | Status |
|---|---|---|---|---|---|---|
| ⚙️ Loading rigs... | ||||||
Rigs pause mining whenever their GPUs are serving AI workloads, so live hashrate dips are expected. Tuning figures are from our own benchmarks.
These GPUs run local LLM inference and image generation. A controller watches each workload and stops mining within a second of a request, restoring stock clocks; mining resumes when the GPU is idle again.
The whole operation runs on rooftop solar with battery storage. GPU mining draws about 485 W in total, alongside CPU Monero mining on the same hosts.
Pearl's pearlhash algorithm is built on int8 matrix multiplication - the same math as AI inference. It's compute-bound, so we lock memory clocks low (810 MHz) and roughly double hashes per watt.
Solo, our hashrate would find a block only every year or so. Instead we mine on HeroMiners' PRL pool, which finds blocks every half hour or so and pays proportionally (PROP), with hourly payouts once 1 PRL is earned.
| Date (UTC) | Paid (PRL) | Value (USD) |
|---|---|---|
| ⚙️ Loading payouts... | ||
Daily totals at today's price.
Have an NVIDIA GPU (Ampere or newer)? Get a Pearl wallet from the official Pearl release, then point a pearlhash miner at a pool. Example with OneZeroMiner on HeroMiners:
onezerominer -a pearlhash -o us.pearl.herominers.com:1200 -w YOUR_PRL_ADDRESS --worker rig1
💡 Tip: Lock memory clocks low (e.g. nvidia-smi -lmc 810,810) and cap power around 65-70% for the best TH/W.
Pearl hard-forks often - keep your miner updated. Feeling lucky? Prefix your address with solo: to mine solo on the same pool.