Changelog
Added support for Parano1d(poseidon2b) on all Nvidia gpus(1% dev fee)
pearlhash:
- Efficiency improvement for Ampere, Ada, and Blackwell gpus(4 to 6 % depending on the OC)
- Performance improvement for Hopper and Blackwell datacenter cards(~4%, H100: 807 Th/s)
- *Added support for memory clock 810 on Turing 20xx series(--kernel 2, default auto based on the memory clock).
- Added support for Volta(V100/CMP100-210)
- Changed the default kernel for other generations to kernel 2.
minor bug fixes
Note: Due to the hardware limitations, the 810 memory kernel for Turing gpus is not as good as the others. It is still more efficient and should give you more hashrate or better efficiency but it needs higher core clocks and OC tuning.