I have a 5x 1080 build.
2x psu
Sometimes (please note sometimes as if i restart the rig there is times where they will all run fine) when running hashcat I get this error :
nvmlInit(): GPU is lost
NvAPI_GPU_GetBusId(): NVAPI_ERROR
NvAPI_GPU_GetBusId(): NVAPI_ERROR
NvAPI_GPU_GetBusId(): NVAPI_ERROR
NvAPI_GPU_GetBusId(): NVAPI_ERROR
NvAPI_GPU_GetBusId(): NVAPI_ERROR
NvAPI_GPU_GetBusId(): NVAPI_ERROR
and either 1 or 2 of my gpus will be no longer run.
Help?
Could be a power issue. Have you audited real-world power draw from the wall for each PSU under various workloads? How is the power balanced across the board and GPUs?
Right, so I have a 1000w psu which support the mobo and 3 cards.
Then another 800w psu which is for the other 2 1080s.
As far as seeing power usage from the wall, I have not done any checking.
Hmm - that seems like enough headroom to me, on each power supply. With ~180W max per 1080 that's only 540W plus 460W for your mainboard, which seems like plenty.
Only other thing I can think of: If you move the GPUs around, does the problem follow the slots, or the GPUs?
Either way, looks less likely to be a problem with hashcat directly, and more a problem of system stability at load.
Classic symptom of bad risers.
https://www.amazon.com/6-Pack-PCI-E-Powe...gpu+risers
These are the risers I have, I plugged 2nd PSU into a new outlet, at first all 5 were working then after a couple session 5th one dropped off again with the same error.
Is it these risers are bad? Are they installed poorly? Or?
With those style risers, the issue is usually with the USB cable. You need really good, high-quality USB cables with those risers. Another common issue is the little x1 card not seated properly in the slot, especially if it's a x4, x8, or x16 slot.
Yikes, now this error has occurred :
clEnqueueWriteBuffer(): CL_OUT_OF_RESOURCES
clEnqueueWriteBuffer(): CL_OUT_OF_RESOURCES
clEnqueueWriteBuffer(): CL_OUT_OF_RESOURCES
clEnqueueWriteBuffer(): CL_OUT_OF_RESOURCES
clEnqueueWriteBuffer(): CL_OUT_OF_RESOURCES