[Ethereum] GPU randomly freezes all the time

linuxminingnvidia

I‘m having trouble with successfully mining Ethereum / Ethereum classic on a 4 GPU rigs consisting of:

1x Intel® Pentium® G4520,
1x Mushkin DIMM 4 GB DDR4-2133
1x GIGABYTE GA-Z170-Gaming K3-EU
4x Palit GeForce GTX 1060 DUAL
2x Kolink ZURC-007 Riser PCIe
1x Enermax Revolution X't II 750W
1x Intenso TOP SSD 128 GB

Running on Ubuntu 16.04 x64 with proprietary Nvidia driver 375.66.

Unfortunately one GPU connected over a riser is freezing all the time, sometimes after 3 h, 7 h, 20 h or even 5 min. I‘ve no idea what is happening. I swapped one GPU from a riser to the problematic slot (Riser energy, USB cable, Riser, PCIe bus, GPU energy) and the “replaced” GPU is freezing as well. I try to run 3 cards by using this slot and it hangs again. By using 3 cards without this “broken slot” it seems to work.

First I thought that I’ve an issue with the power supply but since a PSU calculator confirms that 750 Watt should be enough ( http://www.coolermaster.com/power-supply-calculator/ ) end Enermax guarantees 744 Watt over 12 V (even 850 Watt for seconds) I’m continuing my search. I measured 510 Watt at wall…

Under-clocking doesn’t helped so I’m running the cards on their default settings (except constantly 100 % fan speed). Temperatures are about 62-72° C all the time.

However I tried to use different programs for mining (ethminer, claymore) and I’m still having trouble with freezes. Claymore send “WATCHDOG: GPU 2 hangs in OpenCL call, exit” and ethminer just hangs up. Nvidia-smi doesn’t start up till a reboot.

Now I try to replace the “working” slot with the unstable components (~32 variations) in order to identify the problematic part. Due to the random freezes this could take 10 days or several weeks ( I’ve a job :-/ ).

Can someone help me? I appreciate any helpful suggestion!

Thanks!

Best Answer

The solution was to set the PCIe Port from version 3 to version 2 in the BIOS! Now it's stable for round about 14 days.

Related Topic