Graphics cards up over 95 degrees!

shantd

Gawd
Joined
Aug 2, 2008
Messages
665
I've got a pair of radeon 4850s, I've done the fan fix mod and, at the moment, they're set to 80%. Sitting idle, one of the cards is at 98 degrees and the other is 100. I realize it's quite hot in my room, I'm in California and it's in the mid to high 90s outside. It's a little bit cooler than that inside, but not much. Still, that can't possibly explain why the cards are so hot.

I built this computer several months ago (with help from the good members on this board, thanks again btw!) and everytime I did something with heavy processing, my system would shut itself off. I thought it was my GPU at first because it always happened when I tried to play games but I realized it was my CPU. I was only using the stock fan. I got a Coolermaster v8 for the CPU and the problem was solved, no more shutting off out of nowhere. But my cards are now the problem. I've got dual booting with Vista 64 bit ultimate & XP 32 bit. The cards run slightly hotter in Vista, but a degree or two, but both OS are well over 90 degrees idel.

Any ideas?? Thanks in advance,

Shant
 
Last edited:
100F or 100C? If 100F, then that isn't hot at all and isn't anything to worry about. 100F is under 40C, and the 4850s run 80-90C no problem (with the stock fan profile)

I only ask because you say its mid to high 90s outside, which of course means you are talking fahrenheit, but temperatures for CPUs/GPUs are usually talked about in Celsius, and 100C isn't unheard of for the 4850s.

Oh, and how is your case temperature?
 
It's in celsius. I'm not sure how to check my case temperature, but I'm willing to bet it's hotter than average. Still, you're saying I shouldn't be concerned with idle temperatures of 98-100? What happens when I play a game like Crysis and the cards are loaded above 60-70%? I'm guessing they'd hit 120 degrees or higher...Surely that's not safe for the cards, is it?
 
At 100C idle you'd be lucky to have a stable desktop. Gaming wouldn't be possible. Since you are crashing you probably are running hot though (could be power supply but...)

IF you feel up to it get some AS and replace that OEM gunk under your heatsink. You dont say what brand but for my Saphire it looked like the Tin Man made love to my card under there. Terrible runny silver stuff. Knocked 20C of my temp putting AS on.
 
Any idle temp above 75C is not normal. And that's for the main core of a 4870X2. Your 4850 should be no where near that.

Check the cooler mount, maybe it came loose and is not making proper contact, or maybe the fins are clogged by dust.

And definitely check your case's airflow. Given that your CPU was crashing it probably wasn't enough.
 
My old X1950XT ran 80C idle with a clogged HSF, drops to 45-50C when cleaned out. Pretty likely to be your problem as well if you've owned them for several months without cleaning (I live in CA and my room is pretty freaking dusty too).
 
what is up with the 48xx running so freaking hot?

They don't run hot. Its just situational. My 4870 stays under 50 idle and never cracks 70 under load. I live in an air conditioned condo that stays a steady 68-70f. The op stated that its almost 100 in his house, which would account for his high temps.

Theres only so much air can do. Maybe open up the side of the case and have a box fan blowing in until temps cool off around your area.
 
Thanks everybody for your timely and thoughtful responses, that's what I love about this place. So, I'll go point by point because I'd really like to milk your guys' input as much as possible.

First let me address the issue re: temperature in my room. Since I posted this thread, I cranked my swamp cooler and left the room. Upon returning just now, it's 71 degrees in here. My cards are at 88 & 90 degrees respectively. So yes, that's 10 degrees lower than they were a few hours ago when it was probably around 85 - 90 degrees in here but room temperature is clearly not the main problem here.

At 100C idle you'd be lucky to have a stable desktop. Gaming wouldn't be possible. Since you are crashing you probably are running hot though (could be power supply but...)

The crashing was due to my CPU. I picked up a Coolerrmaster V8 for my CPU yesterday and the crashing has stopped (very pleased with the V8 btw, it's damn near silent at full speed). That leads to a question...at what temperature will my cards result in a crash? I've seen them go as high as 103 (idle) with no visible consequences which worries me.

IF you feel up to it get some AS and replace that OEM gunk under your heatsink.

Forgive my noobishness, but what does AS stand for? And what OEM gunk are you referring to, the heatsink on my 4850s (they are sapphire btw)? You can swap those out?

My old X1950XT ran 80C idle with a clogged HSF, drops to 45-50C when cleaned out. Pretty likely to be your problem as well if you've owned them for several months without cleaning (I live in CA and my room is pretty freaking dusty too).

Well you may be onto something because it is quite dusty in my room. Could you elaborate on what a clogged HSF is? Did your cards get hotter and hotter over time or were they 80C idle from the get go? Also, I had to take the cards and all the components out yesterday when I installed the V8. Wouldn't that have loosened at least some of the dust and lowered the temp a bit, assuming that was the issue?

And definitely check your case's airflow. Given that your CPU was crashing it probably wasn't enough.

I've got a Thermaltake MX+ Armor, so the airflow is supposed to be pretty good. But because of the high temperature, I've taken the side door off and I've got an industrial room fan (very powerful, you can hear it clear on the other side of the house) blowing right on the components, so I know airflow is probably not the issue. Incidentally, when the fan is cranking full blast, the cards come down to the low 60s idle. That still seems awfully high, a fan this powerful blowing right on the cards should bring them down way lower than that I would think.

Thanks again guys.
 
Well you may be onto something because it is quite dusty in my room. Could you elaborate on what a clogged HSF is? Did your cards get hotter and hotter over time or were they 80C idle from the get go? Also, I had to take the cards and all the components out yesterday when I installed the V8. Wouldn't that have loosened at least some of the dust and lowered the temp a bit, assuming that was the issue?

The temperature went up in spikes (50ish -> 67 -> 75 -> 80 give or take). What happened to my card was that dust got trapped between the copper fins of the heatsink, and eventually enough dust built up that the fan could not push air through the heatsink anymore, so all the heat from the GPU kinda just built up and couldn't be moved.

I'm assuming you have the reference single slot coolers for these 4850s? You can see the fins underneath the red see-through plastic on the 4850 here, I basically had to blast compressed hair between them to get the dust out.
SAP21132-00-40R.JPG
 
Well, let me give that a shot. Thanks buddy. And yeah, those are the cards I have. So you don't need any after-market cooling on those? How hot do they get with load, say with a game like Crysis?
 
For a single slot cooler, those tempuratures are only a little high.

OEM gunk means the OEM thermal grease.
'AS' typically stands for Artice Silver 5, a high end thermal grease, but compared to a decent OEM thermal grease, not much better (1C-5C difference). However, a bad OEM grease can keep tempuratures about 10C-15C higher then normal. Taking off a heatsink typically viods a warranty, however.

I rigged (in a slimline case, no less [because they have horrible heat flow!]) a 60mm fan to blow on my graphics card and my chipset (main control chip on the motherboard, other then the CPU). That brought tempuratures down about: 15C from 70Cish to 55C. (graphics).
 
At 100C idle you'd be lucky to have a stable desktop. Gaming wouldn't be possible. Since you are crashing you probably are running hot though (could be power supply but...)
Actually, my 4870's quite fine folding up to 105C (cranked the fans up to 100% to cool it down before it got any hotter). It's certainly not good for any sustained period of time, but for a short time? It shouldn't have too much trouble.

what is up with the 48xx running so freaking hot?
My XFX 4870 with reference cooler runs at 33 idle and 58 load under furmark after two hours.
 
I think that both of you must be getting faulty readings. 100C is 'magic smoke' territory and ICs aren't going to be stable at such temperatures. Integrated temp sensors are notoriously inaccurate.

Shantd:

AS= Arctic Silver

Its a brand of high quality thermal interface material (TIM). A smal quantity of TIM is placed on top of the GPU and then the heatsink (and yes you can replace those too FYI) is attached. It enhances the conduction of heat (In theory but perhaps not with what you have now) into the heatsink. I'd read up on how to go about it for the Radeon 4850, but replacing it is cheap (~$5)
 
So I'd only be replacing what's on there now for the same thing, just with better thermal compound? I don't know if that's gonna cut it. What about one of those pci slot coolers? My 4850s need to be close enough for that crossfire cable to connect the cards, so I have only 2 pci slots worth of space between the cards. Would that be enough for 2 single slot coolers to fit in there, 1 for each card? Or do the coolers need to be placed on top of the cards, rather than underneath?
 
You'd be surprised. Those PCI slot coolers might help, but you might change the ambient (GPU air intake) temperature 5-10 degrees at best. The actual change to the GPU temperature would be 3-5 degrees at most.

Replacing poorly installed, low grade, worthless thermal interface material with something half decent on the other hand can easily achieve 10-20C drop. What you have right now could be so bad that it acts as an insulator (applied too thick, low quality, etc)

However above posters had a good reccomendation in telling you to clean the dust out of your cooler first. My GF's video card had some lint acting as a dust dam and practically clogged the whole cooler on her x1600.
 
I took my video card apart the other day (my Diamond 4850), and there was a wall of dust blocking air flow. My temps were near yours, and I should have fried it gaming with it. I then cleaned it out...and all was good.

Fan mod though? Don't know about that...
 
So what's the threshold then? At what temperature does damage begin to occur? And, will my system automatically shut down when it reaches a certain temperature, the way the CPU does?
 
I think that both of you must be getting faulty readings. 100C is 'magic smoke' territory and ICs aren't going to be stable at such temperatures. Integrated temp sensors are notoriously inaccurate.

The 48xx series is rated to run stable up to 104C according to ATI. But if you hit that point the fan should be screaming at 100% and something is probably wrong (either case airflow, clogged heatsink, etc..)

what is up with the 48xx running so freaking hot?

The default fan profile favors low fan speeds and low noise, hence temperatures rise. It is largely a perceived problem and not an actual problem, and isn't an indication of higher power draw or anything like that. The OP's case is different - he has an actual problem.

Actually, my 4870's quite fine folding up to 105C (cranked the fans up to 100% to cool it down before it got any hotter). It's certainly not good for any sustained period of time, but for a short time? It shouldn't have too much trouble.

Yes, but you are talking load temperatures whereas the OP is talking about *idle* temps. Idle shouldn't be anywhere near 100C, especially not with the fan at 70%. My Sapphire 4850 (same cooler) is idling just under 80C with the fan speed at ~20%. If I crank it to 70% temps plummet to like 45C.
 
I was never too concerned about the heat produced by my 4850 however it died after a year and leading up to its death, it used to artifact lots.
 
4850 idles at 80C its normal. under load it can read 95-100 C. There are countless threads about this topic since this card was released.

my last one started artifacting after a year, just got a new one, big deal, lifetime warranty.
 
One of my Visiontek 4850's has issues when the ATI catalyst control panel settings has the GPU fan set at auto, I Have to always manually adjust the fan speed over 80 percent, it's loud but it's the stock cooler.
 
4850 idles at 80C its normal. under load it can read 95-100 C. There are countless threads about this topic since this card was released.

my last one started artifacting after a year, just got a new one, big deal, lifetime warranty.

Jesus I hope you're wrong. My x2's cores idle at like 46-54.
Posted via [H] Mobile Device
 
Well I'll be damned...it was the dust afterall. Tried to take off the stock cooler/heatsink but it wouldn't budge and I didn't want to force it. Didn't see dust anywhere but almost as an after thought or formality I took a can of air and started blasting away. I noticed a lot of dust coming from a particular area so I took a pair of tweezers and dug in there. Sure enough, I pulled out little 'snakes' of dust and lint-like material.

With the fans set at 60%, the cards are now idling at 64 & 65 respectively. Thank goodness for you guys. I reckon I'll still need to get coolers because it's only a matter of time before it happens again but at least now I'll be OK until I then.

Thanks again fellas, wish I could return the help in some way.

Shant
 
4850 idles at 80C its normal. under load it can read 95-100 C. There are countless threads about this topic since this card was released.

my last one started artifacting after a year, just got a new one, big deal, lifetime warranty.
what...

My 4870 idles at 33 using the refence fan at 33%. On auto, even at full load it never goes over 90C. I had to manually set the fan to 0 to get it over 100.
 
Back
Top