On 10/22/06, Keith Brown <brownk@usask.ca> wrote:
>
> Hi folks. This isn't really a linux problem but rather a hardware
> problem ... at least, I think it is.
>
> My son bought the components for a gaming machine this past summer and
> we assembled it and he started using it. The problem is that when he plays
> graphics-intensive games the machine freezes ... not immediately but
> usually after some seemingly random time. If he doesn't play games and
> uses his machine for browsing etc. it behaves normally. During one of his
> gaming sessions when it froze I touched the graphics card and it was HOT!!
> So hot that I involuntarily pulled my hand away. The card is an ATI
> Sapphire X800.
>
> We took the card back and had the vendor test it ... no problems. We
> took the whole system back and had them test it ... no problems. Bring it
> home ... problems. We loaded a program called ATI Tools and measured the
> graphics chip temperature at 70 degrees C.
>
> I took the card back yesterday and was told that this temperature is
> normal for the card operating under load. I was also told that the heat
> sink can be hotter than the chip. Now, I don't know what the normal
> operating temperature for the graphics chip is but I do know that if the
> chip is emitting heat and the heat sink is getting rid of it, the
> temperature of the heat sink cannot possibly be higher than the chip.
> Otherwise the heat flow would be in the other direction. This is basic
> physics of which I am more than a little aware.
>
> Do any of you think that there is, in fact, a problem on the graphics
> card, given what I have said? I have no gaming experience and don't know
> what to expect as far as performance of the higer end cards goes. I feel
> like we're getting the runaround from the 'technicians' ala the old place
> on Quebec Ave. (whose name escapes me at the moment) where Cbits is
> (was?).
>
>
> Dr. Keith Brown
> Department of Chemistry/
> Saskatchewan Structural Sciences Center
> University of Saskatchewan
> Saskatoon, Saskatchewan
>
> -----------------------------------------------------------------------
> Life would be so much easier if we could just look at the source code.
> -----------------------------------------------------------------------
>
> --
> "Well, there's SPAM, egg, sausage, and SPAM. That's not got MUCH SPAM in it."
>
> To unsubscribe, send a message with the word "unsubscribe" (without the
> quotes) in the body to linux-request@slg.org
> Archives are at http://list.slg.org/
>
If it helps, I can confirm that 70C is absolutely normal for a video
card. I just ran a few games on both machines that I have reasonable
cards in (ATI 9800pro AGP and Nvidia 7800GT PCIE) and both ranged 70C
up to about 110C.
I've seen them both push as high as 110-120C over the years and never
had any issues with them - and have had it confirmed through various
other online discussions that that's an absolutely normal load temp
for a GPU. I've also accidentally touched my hard heatsink while it
was still hot - and I can relate to it being too hot to touch (nearly
hot enough to burn me) while still operating fine and without issue.
Infact my 7800GT as an example, runs about 51C idle and the
'temperature threshold' setting defaults to 115C at factory settings -
where it won't clock down or throttle the card until it gets to be
that hot. So that seems to confirm 70C being reasonable.
As for the heatsink being hotter than the chip - I'll also call BS on
a technician who simply doesn't know what he's doing - or how heat
transfer works.
Also I think the old quebec ave store was Techtronics, wasn't it?
One thing you can do to see if it is a heat related issue is rig up an
80-120mm fan (zip ties, twist ties, string and chewing gum) so that it
hangs beside the card and will be blowing directly on the
card/heatsink/fan. Leave the side off the case, and put it under load
to see if the jury-rigged cooling is preventing it from misbehaving.
If it does - then I'd point my finger at airflow, or a misreporting
temp sensor on the card (meaning it's actually hotter than ATI Tools
is showing).
I've done the above to provide active cooling to the PWM IC on my DFI
motherboard - which tends to react poorly to the overclocking I have
done to my Opteron 165. I just used zip ties to secure an 80mm fan in
such a way that it wouldn't rattle or hit anything else, but would be
able to provide active airflow over the PWM IC - dropping it from as
high as 70-80C under extreme 'tweaking' down to the 40-50C mark. You
wouldn't likely see the same results on a GPU/Heatsink - but it could
help if it truly is a heating issue with the GPU or even the video
card's ram.
Another option, albeit one that requires you to spend a bit more,
would be to try an after-market cooler on the card. I use the cooler
linked below - as it's significantly quieter than the stock cooler -
but it's also very specific to the model of card you use. The one
linked is for example purposes, but is specific to nvidia 6800/7800GT
models - and may even be PCIE specific. That being said - there's
likely one that will work on an X800 AGP or X800 PCIE.
Received on Sun Oct 22 17:30:27 2006
This archive was generated by hypermail 2.1.8 : Sun Oct 22 2006 - 17:30:39 CST