Folding Forum

I was running normal tgemps nd fan speed till I hit this unit, Project: 10122 (Run 356, Clone 3, Gen 99), now I'm hitting 99% fan and 90-98C on the gpu.

I'm running Project: 10122 (Run 176, Clone 1, Gen 140) onb my other card and its happily chugging as per 2yrs of crunching... 82C and 61% fan.

This heat unit, probably explains why I bsod on the nvidia driver yesterday.

Anyone else getting this?

patonb wrote:I was running normal tgemps nd fan speed till I hit this unit, Project: 10122 (Run 356, Clone 3, Gen 99), now I'm hitting 99% fan and 90-98C on the gpu.

I'm running Project: 10122 (Run 176, Clone 1, Gen 140) onb my other card and its happily chugging as per 2yrs of crunching... 82C and 61% fan.

This heat unit, probably explains why I bsod on the nvidia driver yesterday.

Anyone else getting this?

Yes, my 250 is currently at 87C but it has been as high as 93C recently.

I would suggest a cleaning of the heatsink and manually set the fan to 100%

Sorry for not staying strictly on topic, but I've noticed project-dependent variation on my GPU3 as well. Not that there is much variation these days, since almost all of the assignments are 68xx / 4kPPD instead of, say, 109xx / 8kPPD. I just assume some projects are run more "optimally" than others on hardware X. New or rarely assigned projects or FAHcores may change the picture any moment, for better or worse.

Anyway, if you want something approaching worst case scenario when it comes to power consumption and therefore temperatures, IMHO folding isn't particularly good benchmark. Some torture test results from my setup here, they made folding seem rather tame. Core x temp results might be a bit off, I've checked them against Intel CoreTemp results later on, and the Speedfan results were too low. The rest - including CPU temp - should be real, though.

Sure, folding may be working specific parts of your HW harder than just about anything else, but if you really want to see your HW burn, try running e.g LinX and FurMark concurrently, like I did. You might find out that your cooling or power supply stability isn't quite as adequate as you thought.

CAUTION! If you're already pushing the envelope with FAH, chances are that TAH ( Torture At Home ) might fry your HW, or at least it may reach a point where the HW gets throttled. It'll help if you have a Kill-a-Watt handy and compare folding power consumption vs "torture test" power consumption. If the torture test consumes more power, rest assured that just about all of that additional power will transform into extra heat inside your PSU and/or case. Since the case cooling also comes into play, it might take a long time before reaching maximum temperatures. However, more power == more heat, simple as that. I would not recommend running stuff like this completely unattended.

If I furmarmk kill my card, it shoots way way up past f@h temps.

I'm still gpu2 on my 260's and ppd has been reletively constant. 1 log is showing 9800+ units returned on the card.. the other says only 1500!

But the oddness was same base, 10122 and 2 wildly different temp results. Dropping the shadder oc to default has cooled it back tgo 83C but 70% fan...

Ah well, I'll just soldier on!

My bad, I read the earlier posts too hastily. AFAIK, it really is weird to see that kind of variation between WUs of the same project.
Did you (or someone else) manage to complete that hot WU?

Hzaha I meant units submitted not ppd... I was just showing the stats on the card for my observed temps and fan speeds...

I finished it... on/off on/off... then went to default speeds.

It upp'd the fan to 77% to maintain 83C, where the other card 2inchs away ran 83C at 59%.

C'est la vie... I just need to watch I don't get 100Cs on it..

Good news is that you successfully completed it:

Hi patonb (team 12864),
Your WU (P10122 R356 C3 G99) was added to the stats database on 2011-06-05 12:04:47 for 494 points of credit

Regarding the increase of temperature, what I can suggest that may in the trajectory/WU, the atoms are more closely packed which means that more calculations needs to be done which equates to more heat generated as the GPU is stressed more. The reason that the difference in temperatures in a particular project is that each trajectory chooses a random position of atoms and then calculations are performed. Thus no two trajectories will have the same path. (For further details, please read this http://fahwiki.net/index.php/Runs,_Clones_and_Gens).