PearlVolt is a CUDA-native miner for the Pearl network — where hashpower is spent on verifiable useful compute, not arbitrary hashes. Auto-tuned kernels, sub-1% dev fee, live telemetry out of the box.
Everything you'd hand-roll for a serious rig — kernel auto-tuning, failover, per-GPU control, a metrics endpoint — shipped in one static binary with no runtime dependencies.
Kernels re-tune per architecture at launch — occupancy, register pressure and memory clocks profiled live, so a 5090 and a 3090 each run their own optimum without config.
A flat 0.75% fee, mined transparently and logged. No hidden second wallet, no fee spikes. Verify it in the telemetry stream at any time.
Define a pool list; PearlVolt fails over on stratum drop in under a second and auto-restarts a hung GPU instead of taking the whole rig down.
A local :4141/metrics Prometheus endpoint plus JSON socket — hashrate, shares, temps and power, ready for Grafana or your own dashboard on day one.
Measured on stock power limits, ambient 24°C, PearlVolt v1.0.0. Efficiency is the number that pays your power bill.
| GPU · miner | Hashrate | Power | Efficiency | Dev fee |
|---|---|---|---|---|
| RTX 5090 · PearlVolt | 390 TH/s | 500 W | 780 GH/W | 0.75% |
| RTX 4090 · PearlVolt | 257 TH/s | 350 W | 734 GH/W | 0.75% |
| RTX 3090 · PearlVolt | 120 TH/s | 300 W | 400 GH/W | 0.75% |
Low-latency PPLNS with transparent payouts, a public API, and stratum endpoints on three continents. Point the miner at it in one line.
One static download. No CUDA toolkit install, no Python, no dependencies.
Paste your PRL address and a stratum endpoint. That's the whole config.
Open localhost:4141 for live telemetry as shares land.
Free, open telemetry, sub-1% fee. Mine on the network where the work isn't wasted.