Closed benrod3k closed 2 years ago
Beside the two comments about the if
s, it looks good.
I'm wondering if we really want to make it global? We would never want to turn it off in the middle of working miner?
Beside the two comments about the
if
s, it looks good. I'm wondering if we really want to make it global?
In some case, global might not work. eg: Getting the core/mem clocks, temperature and pl, etc in the middle of mining, it might be useful for user who mine without any shell like hiveos to get the statuses.
Should I move it to Worker?
We would never want to turn it off in the middle of working miner?
~Not sure how to turn it on/off in the middle of mining with the miner itself. It is doable with nvidia-smi though.~
Inspired by bzminer (it support set oc per gpu per algo in dual mining mode). So it's potentially could to be changed in the middle of a working miner. Although it won't happen in kaspa-miner.
refs: https://github.com/tmrlvi/kaspa-miner/issues/20
--cuda-lock-mem-clocks 810 --cuda-lock-core-clocks 1950 --cuda-power-limits 200,200,200,170
nvml-wrapper seems to large? and the new changes not accepted by upstream yet.
Tested on hiveos only with:
NVIDIA-SMI 495.46
Driver Version: 495.46
CUDA Version: 11.5