severe stuttering problems in SH3. but why?

BilgeMatt · 09-13-10, 11:54 PM

ok so I cracked open the old PC case and went crazy with a vacuum and cleaned it out. it was fairly dusty..not excessively, but dusty. but more importantly I reconnected some fans i had turned off because of fear of noise. well the fans are pretty high quality so they're not that noisy.

anywhooo, i ran the Furmark stress test again, and this time the GPU maxed at 93C and after a minute or so went back down to 91/92C while still under load. So I think ok cool, the video card is running 10C cooler under severe load.

Went back into this SH3 you speak of

and loaded up the same tutorial, and had RealTemp logging (thx to someone's earlier post for the app tip). this time the stuttering wasn't nearly as bad as my previous encounters with it, and I had cranked the particle setting back up to 100, too. During the times of lockup/stuttering, the GPU temp seemed to be around 82-84C. if only I had known about this app before, I would have done logging pre-vacuum and fan reactivation. Now that said, I see in the log during another lockup that my cpu load(s) went to 100 for at least 6-10 seconds (I was logging in 5 second intervals). Now that I chalk up to ... Windows XP's inherent...er how to say this nicely..? But that would definitely aggravate a situation of graphic lockup.

So this begs the question, even though a gpu stress test can draw the fuzzy doughnut just fine at 103C, perhaps moving 'real' data around the gpu and memory is adversely affected by heat, even if we're not at the thermal limits of the hardware? I see some reports to this effect when Googling around a bit.

So before I overhaul my system I'm going to work on some good cooling solutions. Probably not like the guy who posted here with the dual dryer hose tubing going to the window.

I will do some more testing and logging and report back. I know this stuff is riveting!

or at the very least, good for treating insomnia.

manning the pump,
bilgematt

09-13-10, 11:54 PM	#8
BilgeMatt Swabbie Join Date: Sep 2010 Location: SF, CA, US Posts: 7 Downloads: 2 Uploads: 0	the plot thickens? ok so I cracked open the old PC case and went crazy with a vacuum and cleaned it out. it was fairly dusty..not excessively, but dusty. but more importantly I reconnected some fans i had turned off because of fear of noise. well the fans are pretty high quality so they're not that noisy. anywhooo, i ran the Furmark stress test again, and this time the GPU maxed at 93C and after a minute or so went back down to 91/92C while still under load. So I think ok cool, the video card is running 10C cooler under severe load. Went back into this SH3 you speak of and loaded up the same tutorial, and had RealTemp logging (thx to someone's earlier post for the app tip). this time the stuttering wasn't nearly as bad as my previous encounters with it, and I had cranked the particle setting back up to 100, too. During the times of lockup/stuttering, the GPU temp seemed to be around 82-84C. if only I had known about this app before, I would have done logging pre-vacuum and fan reactivation. Now that said, I see in the log during another lockup that my cpu load(s) went to 100 for at least 6-10 seconds (I was logging in 5 second intervals). Now that I chalk up to ... Windows XP's inherent...er how to say this nicely..? But that would definitely aggravate a situation of graphic lockup. So this begs the question, even though a gpu stress test can draw the fuzzy doughnut just fine at 103C, perhaps moving 'real' data around the gpu and memory is adversely affected by heat, even if we're not at the thermal limits of the hardware? I see some reports to this effect when Googling around a bit. So before I overhaul my system I'm going to work on some good cooling solutions. Probably not like the guy who posted here with the dual dryer hose tubing going to the window. I will do some more testing and logging and report back. I know this stuff is riveting! or at the very least, good for treating insomnia. manning the pump, bilgematt