RGminer v1.0.0

Highlights

• Added new CUDA kernels for NVIDIA SM70, SM90, and SM100 architectures.
• Fixed Kryptex pool connectivity and protocol handling.
• Better compatibility with A100/CMP 170HX

Pearl Performance

GPU	Architecture	OC Profile	Hashrate	Power	Efficiency	Energy Usage
Tesla V100-SXM2-16GB	Volta	0 / 1550 / 0	49.5 TH/s	280 W	0.176 TH/W	5.66 W/TH
RTX 2080	Turing	125 / 1650 / 5000	44.16 TH/s	160 W	0.276 TH/W	3.62 W/TH
RTX 3070m	Ampere	250 / 1650 / 6000	65.3 TH/s	140 W	0.466 TH/W	2.14 W/TH
RTX 3070	Ampere	250 / 1450 / 6000	66.5 TH/s	130 W	0.512 TH/W	1.95 W/TH
CMP 170HX/A100	Ampere	250 / 1450 / 0	160 TH/s	210 W	0.761 TH/W	1.31 W/TH
RTX 4070 Ti	Ada	350 / 2450 / 5000	145.3 TH/s	190 W	0.765 TH/W	1.31 W/TH
RTX 5070 Ti	Blackwell	490 / 2450 / 7000	160.7 TH/s	218 W	0.724 TH/W	1.38 W/TH

Miner Watchdog

The watchdog controls how rgminer responds to critical CUDA errors.

--watchdog=off - disables automatic rig rebooting and persistent GPU disabling.
--watchdog=restart - restart on errors
--watchdog=reboot - reboot on errors

GPU Clock Control

Each option accepts multiple GPUs using:

GPU_INDEX:VALUE,GPU_INDEX:VALUE

./rgminer \
  --cclock 0:125,1:250 \
  --mclock 0:500,1:1000 \
  --lock-cclock 0:1650,1:2450 \
  --lock-mclock 0:7000,1:7000 \
  ...

New Pearl Pool Protocols

Added support for three Pearl pool protocols:

• PearlFortune — --proto pearlfortune
• AlphaPool — --proto alphapool
• f2pool — --proto f2pool

Pool-specific Stratum authorization, job parsing and share submission are supported.

Compressed share submission is also supported where required:

• PearlFortune/herominers — zstd-compressed proofs
• f2pool/kryptex — gzip-compressed v2 shares

Complete Launch Example

./rgminer \
  --algo pearl \
  --stratum POOL:PORT \
  --wallet WALLET \
  --worker RIG_NAME \

Backend policy

• Turing, Ampere and Ada builds are shipped on CUDA 12.4
• Blackwell build is shipped on CUDA 12.8

Performance depends on GPU model, driver, clocks, memory and power limit conditions.