Folding Forum

jrweiss wrote:What about voltages? Are you monitoring the 12V rail?

FYI, my readings come from HWiNFO64.

The motherboard has a +12V line that reads 11.808V at the time of the failure. The low/high watermarks for that line are 11.808V to 11.880V. As I'm sitting here it reads 11.880V, and it seems to spend most of its time there. Do you think the 0.072V dip could account for it?

These lines are solid with no fluctuation: +3.3V at 3.305V; +5V at 5.040V; DDR at 1.512V; 3VSB at 3.384V; VBAT at 3.096V.

There's a CPU VRIN line that also seems pretty stable, but has a low/high watermarks of 1.740 to 1.764V. The steady-state seems to fluctuate between 1.754V to 1.764V. It was 1.752V at the freeze.

AVCC3 seems to fluctuate widely. 0.288V to 3.048V. It was 1.464V at the time of the freeze. It currently reads 3.048V. What does this do?

The GPU Core voltage goes up to about 1.044V and is 0.650V at rest. I suspect this rises with load.

For what it's worth, I downgraded my NVIDIA drivers from 445.87 (win10-64bit) to 441.87. Processing went for quite a long time but died at the 21% mark; a full hour after it started. The previous longest run went to 16%, but the duration seems random. Sometimes it only gets as far as 2% before it freezes. The 441.87 is from around January, before the 445.X versions started. I don't know if it's reasonable to suspect the driver at this point though.

LancerDL wrote:...I downgraded my NVIDIA drivers from 445.87 (win10-64bit) to 441.87. Processing went for quite a long time but died at the 21% mark; a full hour after it started. The previous longest run went to 16%, but the duration seems random. Sometimes it only gets as far as 2% before it freezes. The 441.87 is from around January, before the 445.X versions started. I don't know if it's reasonable to suspect the driver at this point though.

If it was a driver issue, we would have gotten a lot of reports about it. However, it could be driver related on your system so there's something unique about yours when compared to the rest.

LancerDL wrote:

jrweiss wrote:What about voltages? Are you monitoring the 12V rail?

FYI, my readings come from HWiNFO64.

The motherboard has a +12V line that reads 11.808V at the time of the failure. The low/high watermarks for that line are 11.808V to 11.880V. As I'm sitting here it reads 11.880V, and it seems to spend most of its time there. Do you think the 0.072V dip could account for it?

The 12V rail is the only one from the PSU important for the GPU. CPU and GPU core voltages will vary with load.

I would try a couple different apps to check that 12V rail, and see if they agree. I see 11.808 as min reported on my rig (HWInfo), but 11.904 nominal and 12.096 max. It is possible that your GPU is sensitive to the lower voltage.

More important, though, is that the voltage is steady. You might try monitoring the 12V rail while you start/run/stop the GPUz render test or your favorite GPU stress test. See if you can detect any significant, repeated voltage drops or variations.

I have had the same problem and did not find a solution yet, any more suggestions?

Sometimes motherboard manufacturers use sub-par capacitors, or on older boards some of those get worn out.

Another one is bad ribbon risers, faulty connection on USB risers (using SATA to 6 pin connectors), or using non-powered risers, using too many GPUs per cable (preferably 1 GPU per PCIE cable), bad contact on the PCIE slot (either from a riser, or GPU), bad contact on one of the PCIE 6+2 pin connectors on the top, power fluctuations on the PSU (less common if you're running a <80% load), faulty hardware (motherboard, GPU, PSU or other plugged in device), (really) dirty power (use a filter), overheating GPU (make sure your heat sinks are clean, fans rotate freely, and you change thermal paste every 4 years.
You could also use thermal pads on the VRM, RAM, and Caps connecting them with the heat sink; to reduce the chance of overheating.
Heat sinks usually stay in a sub 60c region, even if the GPU reads 80C; which I've heard it's better to run VRAM at those temps than totally cold.

santi82 wrote:I have had the same problem and did not find a solution yet, any more suggestions?

What do you have, what do you see, and what have you done?

santi82 wrote:I have had the same problem and did not find a solution yet, any more suggestions?

Can you share your system details? Maybe we can compare and see if we have anything in common?