So we have more people with what I believe is the same problem
Hi all:
I have been quiet and doing a bit of experimenting with many of the different versions of Claymore. I just did a test of a card that would run say for 6 hours and then lose connection and drop out. The other card with it would stay mining just fine. Did this 6 times to make sure what I was seeing to be true.
I then took said card to a different miner and one of them to said miner. Now all works fine. I see that some cards just will not work together but will with others. So I feel strongly to say that if you have a rouge card, put it in a different machine and it may correct your problems. May not.
My proof is in the puddin so to speak.
I've been struggling to get my rig to work for two weeks. It was mining fine 24/7 for days at a time. Then it froze, and ever since then it'll mine for a minute, two cards will go to zero hash rate, and Windows freezes. Tried reinstalling drivers. Connecting them in various orders. No one card seems to be causing the problem. I'm at a loss here and this thing is sucking up all my time.
Win10, rx580s (one 480). 1000W evga psu. Biostar mobo.
I am happy to let you know that I think I have found the cause of the problem. I yet have to test it for a few more hours, but it seems that the "Freeze problem" after connection lost and sometimes after a gpu hangs, it is all because of
Power consumptionI myself have 2x750watt PSU and before blockchain drivers I was drawing 1330watts from the wall. With the blockchain drivers I was drawing 1400watts.
Now that I got physical access to my rig, I was able to perform some tests and what I found is that
when the miner starts up, the power consumption skyrockets for a few seconds.
This will happen on the 1st time the miner starts, but
it will also happen after a connection loss (when the connection gets restored I guess) or a miner restart due to a GPU hangI saw that "freeze" happening in front of me and I was amazed to see that half of my GPUs that drew power from the same PSU were turned off while the others kept working!
I have graph data and logs to back this up, I will post later this day probably. For now I have to let it run in order to see if my solution works: lowering voltages
just a tiny bit more, enough to get me back on the 1300watt consumption area.
So if you guys are like me, you like to have your PSU consumption close to its efficiency peak (90-92%) and you are experiencing freezes during startup or after connection loss, a 10% spike might be responsible for switching off your PSUs
FYI:"Parent" post
here