Same question as before but with pictures, miner is hashing away when suddenly the whole OS slows down to a snail pace where i have to wait for 20 seconds for every action to go through and the hash rate is cut by 50% with the GPUs doing this strange hashing pattern:
Please advise, this happens on any pool i try, in both dual mining and ETH only mode, across multiple overclocks..
This is a soft crash; if you have already tried lower OC; it is most likely being caused by too low of a powerlimit.
What GPUs / clocks / powerlimits are you using?
7 x gigabyte 1070 FE, each at 120 pl and clocked +200/+1600, turns out this also happens in ETH only mode, not just dual mining.
any suggestions?
If you want to keep that OC, you will need more power. 120 is not enough to maintain that OC (based on from my experience with my 1070s) I had to give them at least 135-145 to do +200/1600 solo mining ETH.
i see, i was going by UberDaemon's settings since i ended up getting the same parts as his build.
Right now it should be more profitable to use genoil (which unless I am mistaken; can only single mine) as you should be able to net at least +5% from Claymore single mode (with outright higher hashrate and no 2% fee).
I am not sure at all what the best way to avoid soft crashes is with genoil; if anyone can find OPT settings that are stable with high memory clocks let me know.
Currently testing it, but am noticing that it does not have the nice web API to connect too like claymore lol. I will get back with you later today with my settings.
Thanks man,
I just finished testing and for some reason the same overclock (200/1600) @ 125 pl while still soft crashes with genoil, it doesnt cripple the whole system to the point where i have to restart the rig, and for the same clocks it does hash around 10Mh/s faster.
BUT, i've noticed something else, Claymore now hashes
SLOWER with CUDA 8 then with the previous version (218Mh/s vs 222Mh/s), and i was planning on dual mining, and now its slower?
Nothing with CUDA for Claymore has been changed. I have had CUDA 8.0 since v0013. Any changes in Claymore performance are due to your pool or the Network as a whole.
When compiling genoil with the new CUDA implementation you modify 3 files (the new implementation) then compile the mining client.
I have had almost all soft crashes as well when trying to OC too much.