mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware > GPU Computing

Reply
 
Thread Tools
Old 2012-01-13, 04:04   #1
nucleon
 
nucleon's Avatar
 
Mar 2003
Melbourne

5×103 Posts
Default Interesting GPU heat issue...

Interesting GPU heat issue on one of my nodes. This is i7 2600k@4.5GHz with 2x GTX580s.

I have one card here that's instance 0 and 1. (top two mfaktc windows - the gpuz in the foreground/left)

Note that their times are twice that of the bottom 2 instances on the other GPU (gpuz on the right/background)

Looking at gpuz stats the slower card is running 97degC (ish). It seems the card is doing 50% of the work output that it should be doing. If I close the mfaktc windows and restart when it's cooler, iteration times are normal.

Yes I recognize this is running 'hot'. :)

If you look at the fan speed on the problem card - it seems all over the place.

Anyone recommend a windows app to control GPU fan speeds?

-- Craig
Attached Thumbnails
Click image for larger version

Name:	gauss00-heat-slowdown.png
Views:	161
Size:	223.7 KB
ID:	7552  
nucleon is offline   Reply With Quote
Old 2012-01-13, 04:20   #2
kladner
 
kladner's Avatar
 
"Kieren"
Jul 2011
In My Own Galaxy!

2×3×1,693 Posts
Default

Quote:
Originally Posted by nucleon View Post
Interesting GPU heat issue on one of my nodes. This is i7 2600k@4.5GHz with 2x GTX580s.

I have one card here that's instance 0 and 1. (top two mfaktc windows - the gpuz in the foreground/left)

Note that their times are twice that of the bottom 2 instances on the other GPU (gpuz on the right/background)

Looking at gpuz stats the slower card is running 97degC (ish). It seems the card is doing 50% of the work output that it should be doing. If I close the mfaktc windows and restart when it's cooler, iteration times are normal.

Yes I recognize this is running 'hot'. :)

If you look at the fan speed on the problem card - it seems all over the place.

Anyone recommend a windows app to control GPU fan speeds?

-- Craig
You could try MSI Afterburner. It lets you define the fan response curve.

If the frequency shown wasn't so high, I might have thought that the card had throttled because of the high temp. EDIT: But I don't know what frequency it should be running at.
http://event.msi.com/vga/afterburner/download.htm

EDIT2: I use Afterburner even though my GTX 460 is by Gigabyte.

Last fiddled with by kladner on 2012-01-13 at 04:23
kladner is offline   Reply With Quote
Old 2012-01-13, 04:35   #3
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
Jun 2011
Thailand

7·1,373 Posts
Default

Quote:
Originally Posted by nucleon View Post
Interesting GPU heat issue on one of my nodes.
<...>
Anyone recommend a windows app to control GPU fan speeds?
.
That is exactly the issue I was describing, with the GPU card in the middle getting hotter then the other, for the simple reason that airflow is not right, this card is getting all the heat from the CPU (above) and the second GPU (below) and the space in between is not the most "fortunate". I have about the same config, I described somewhere around. What mobo/gpu is that? (brand). If they are Asus (like for example Maximus Extreme Z or Sabertooth with ENGTX580/590/mars) then Asus has very nice "control tools" to deal with fans, overclocking, etc. (SmartDoctor, GPUTweak, iTracker, etc), which I am using, but the best way would be (as I mentioned) going for watercooling for the card in the middle, at least (I was waiting for Santa, but he was not so eager to come this Christmas... or at least, not with that present... maybe in a month or two, he will come for Chinese New Year, or Thai Songkran, depends of what savings we can make...). I said "at least", because once you got the pumps, tubes, radiators, blabla, there is only a small step to buy two more plates and two splitters and have watercooling for all three (both gpu's and the cpu).

Last fiddled with by LaurV on 2012-01-13 at 04:49
LaurV is offline   Reply With Quote
Old 2012-01-13, 04:38   #4
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

3×29×83 Posts
Default

Quote:
Originally Posted by kladner View Post
You could try MSI Afterburner. It lets you define the fan response curve.

If the frequency shown wasn't so high, I might have thought that the card had throttled because of the high temp. EDIT: But I don't know what frequency it should be running at.
http://event.msi.com/vga/afterburner/download.htm

EDIT2: I use Afterburner even though my GTX 460 is by Gigabyte.
Seconded. Though mine is MSI, I've seen in many places that it's more-or-less compatible with everything.

nVidia reports 772 core clock standard, so that's fine.
http://www.geforce.com/Hardware/GPUs...specifications

Last fiddled with by Dubslow on 2012-01-13 at 04:39
Dubslow is offline   Reply With Quote
Old 2012-01-13, 04:47   #5
kladner
 
kladner's Avatar
 
"Kieren"
Jul 2011
In My Own Galaxy!

2×3×1,693 Posts
Default

I think LaurV has really put the finger on it (assuming that this card is sandwiched.) The fact that stopping the app and letting the card cool results in better performance really does suggest some kind of throttling.

These are exotic situations with multiple high-wattage devices packed in so closely. Once again, my airflow instincts are twitching big time. But you have a pretty good idea of such things. Try to get more air in between is all I can say.
kladner is offline   Reply With Quote
Old 2012-01-13, 07:04   #6
nucleon
 
nucleon's Avatar
 
Mar 2003
Melbourne

5·103 Posts
Default

We can see here the fan has pretty much died now.

I've reduced it to one instance for the moment until I can get a fix organized. Like you said more airflow needed.

It's a shame $500 of GPU power boils down to a fan assembly that costs maybe $20-50 all up.

-- Craig
Attached Thumbnails
Click image for larger version

Name:	gauss00-heat-slowdown-3.png
Views:	134
Size:	159.3 KB
ID:	7553  
nucleon is offline   Reply With Quote
Old 2012-01-13, 07:17   #7
nucleon
 
nucleon's Avatar
 
Mar 2003
Melbourne

5·103 Posts
Default

Quote:
Originally Posted by kladner View Post
You could try MSI Afterburner. It lets you define the fan response curve.

If the frequency shown wasn't so high, I might have thought that the card had throttled because of the high temp. EDIT: But I don't know what frequency it should be running at.
http://event.msi.com/vga/afterburner/download.htm

EDIT2: I use Afterburner even though my GTX 460 is by Gigabyte.
Cheers - I've tried afterburner. No luck increasing the fan speed. It looks like the fan has seized. It's happened on another card I had.

-- Craig
nucleon is offline   Reply With Quote
Old 2012-01-13, 13:58   #8
Chuck
 
Chuck's Avatar
 
May 2011
Orange Park, FL

3×5×59 Posts
Default

Quote:
Originally Posted by nucleon View Post
Cheers - I've tried afterburner. No luck increasing the fan speed. It looks like the fan has seized. It's happened on another card I had.

-- Craig
How long did the cards run before you had the fan problem? I'm not looking forward to this happening to me.
Chuck is offline   Reply With Quote
Old 2012-01-13, 15:06   #9
kladner
 
kladner's Avatar
 
"Kieren"
Jul 2011
In My Own Galaxy!

1015810 Posts
Default

Quote:
Originally Posted by nucleon View Post
We can see here the fan has pretty much died now.

I've reduced it to one instance for the moment until I can get a fix organized. Like you said more airflow needed.

It's a shame $500 of GPU power boils down to a fan assembly that costs maybe $20-50 all up.

-- Craig
I failed to notice previously (or put 2 and 2 together) that the fan controller was constantly ~85% while the RPM varied.

Besides seconding Chuck's question as to run time before failure, I am curious who made the card and what approx temp it was running before the fan went south. Is it a single fan design? Is it "in the middle" as LaurV put it? I am wondering now if the ambient temp it was running in helped take down the fan. (Not that that is any excuse.)

It is a damned shame that the manufacturer didn't put a more reliable part in, considering how vital that part is. I'd be pretty surprised if a ~70mm fan (or 2!) cost them even $20 at wholesale. The most charitable thing I can say is that maybe this was just bad luck as to the particular fan you ended up with.

My condolences, and hopes that you get a quick and painle$$ resolution.
kladner is offline   Reply With Quote
Old 2012-01-13, 17:34   #10
KyleAskine
 
KyleAskine's Avatar
 
Oct 2011
Maryland

2·5·29 Posts
Default

Quote:
Originally Posted by kladner View Post
I failed to notice previously (or put 2 and 2 together) that the fan controller was constantly ~85% while the RPM varied.

Besides seconding Chuck's question as to run time before failure, I am curious who made the card and what approx temp it was running before the fan went south. Is it a single fan design? Is it "in the middle" as LaurV put it? I am wondering now if the ambient temp it was running in helped take down the fan. (Not that that is any excuse.)

It is a damned shame that the manufacturer didn't put a more reliable part in, considering how vital that part is. I'd be pretty surprised if a ~70mm fan (or 2!) cost them even $20 at wholesale. The most charitable thing I can say is that maybe this was just bad luck as to the particular fan you ended up with.

My condolences, and hopes that you get a quick and painle$$ resolution.
I think that luck has a lot to do with it.
KyleAskine is offline   Reply With Quote
Old 2012-01-13, 21:54   #11
nucleon
 
nucleon's Avatar
 
Mar 2003
Melbourne

5·103 Posts
Default

I looked up when I purchased 'them'.

I bought 2x of the same model (MSI GTX580 Twin Frozr II OC) - 7/3/11 and 28/3/11. Now both models have fans that aren't working anymore. You could possibly google what they looked like (dual fans). I'm spewing as they advertised that they were pretty good for cooling.

I've been pretty much running them 24*7 since I've bought them. It was running high 90degC for a few weeks now. Looks like the fans have had problems for a while my guess.

If I use my finger to rotate the blades, I can feel obvious friction. They've 'siezed up' somehow.

-- Craig
nucleon is offline   Reply With Quote
Reply



Similar Threads
Thread Thread Starter Forum Replies Last Post
Watts to Heat Ratio of a PSU wildrabbitt Hardware 12 2015-12-12 16:18
Four heat seeking missiles mfgoode Puzzles 51 2007-01-10 10:02
Heat problems michaf Sierpinski/Riesel Base 5 3 2006-07-04 21:11
Keeping the Heat down at Home outlnder Hardware 61 2003-02-15 03:12
heat and errors crash893 Hardware 37 2002-11-12 16:33

All times are UTC. The time now is 09:24.


Sat Jul 17 09:24:59 UTC 2021 up 50 days, 7:12, 1 user, load averages: 1.46, 1.53, 1.59

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.