same setting, but i can't get that hash with that wattage, the maximum is 310 sol 100w, what is your core clock?
6x GTX 1070 EVGA FTW - Drivers 376.19
Core +140 (real core ~1860Mhz)
vCore +20%
Mem +685 (real core ~4482Mhz)
TDP 50% (~91w per card)
~320 Sols/card - Total 1927 sols/s with 2 process of EQM (3 cards each) i know you can do 2 threads per cards now, but i found it to be less stable in my experiments (had some crashes, and get 3% less sols/s overall)
~695W @ wall - 1 sol/s per 0.36066W
Very efficient for now ;-), will play a bit more with OC/vCore over the week-end, this configuration being rock stable for 2 days, 0 rejected shares. Would like to see if some settings or code optimizations could max out the GPU's they're not fully maxed (about 96% load in nvidia-smi), i used the usual variables to max out cards, but it's not 100%.
Cheers.
Revision on my post, now i'm able to work with only 1 process and 2 threads by GPU's with the "-ca -ca" switches (the -cd 0 0 1 1 etc... wasn't working crashing), now pushing 50-60 more sol/s (1927 -> 1980-90) than with 2 process 1 threads per card, seems to also have reduced the few reject/stales i could have sometimes (very little about 0.2%, now looking more like 0.1%) stable since ~24h
Still wondering what those "Solver Interrupted" messages are, probably part of the solving process to avoid collisions or something, would love to get some infos
Cheers.