[solved] overheating after upgrade

[ccole]

I have 3x 7970 in a case. I have two of these machines, both with the same issues.

I upgraded from oclHashcat 1.01 with Catalyst 13.12 (all worked fine) to oclHashcat 1.21 with Catalyst 14.4. Now I can't crack for longer than a minute or two because it overheats.

Previously the GPU temperatures stayed in 60's and 70's while cracking. Now it auto stops as it is supposed to when it hits 90.

Is this a driver issue? Or is something different going on in oclHashcat? I've seen a couple similar posts, but none of them seem to answer this question. My case has pretty good cooling.

[K9]

Try --gpu-temp-retain (f.e. --gpu-temp-retain=70)

[ccole]

Well the default for that is 80, right? That seems like a fine place to have it to me, but my cards blow right past that without even slowing down with this new version of software.

[K9]

I had similar problems. The fan speed was still at 15% while the temp was higher than it should be. With --gpu-temp-retain=70 or even 65 it will set the fan speed at 100% and it won't go over 60°C.

[ccole]

Thanks K9, I see what's happening. I hadn't noticed the fans weren't turning on. I set gpu-temp-retain to 65, and we're working well!

One more question. Does the gpu-temp-retain actually slow down hashing to avoid overheating? Or is the only thing it does to turn on the fans to blow harder?

[ccole]

It's also probably worth noting that since upgrading, my hashing speeds have slowed down slightly on most of the algorithms, comparing the benchmark command from before and now.

[K9]

It does not affect the speed I guess. Speedlose is maybe caused by the new drivers/catalyst.