Having free time at the moment, I looked into optimizing my GPUs, on my rigs I managed to reduce my stales by 50% via a few various optimizations.
My RTX 3xxx cards under linux "hiveOS".
On the other hand I have a rig which has more that I still worry about, I managed to reduce the stale but only 35% for him.
When I look at "screen -r miner" I sometimes randomly get a #GPU which displays 1875 Ms ... GPU # 0 | 5.04 G
Then everything becomes correct again, could someone tell me what is really this latency time?
I recall that my other identical rigs same OS / configuration / motherboard I do not have that
It sometimes happens to have more worries if I restart the miner just after the generation of the DAG, sometimes 3 times is enough more latency #GPU blabla.
I try with this: "low-load": 1 no change
Screenshot to illustrate what I'm trying to explain
Is this the time it took him to request a JOB, calculate it and then send it to the pool? who could be involved in the GPU? CPU? a network disturbance
I'm just looking for leads because sometimes the solution is in front of us but we can't see it;)
Thank you