Please register or login. There are 0 registered and 289 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s October 18 - 06:36am EDT 
Hardware Analysis
Forums Product Prices

  Latest Topics 

More >>


  You Are Here: 
/ Forums / Videocards /

  gpu temp. too high??? 
 Date Written 
Ben Bernard Oct 16, 2010, 02:44pm EDT Report Abuse
My ATI HD4870 card gets heated up to 72 Celcius when playing games. Is that too high? The game is running alright. Normal temp. when not playing games is 62 Celcius.

Want to enjoy fewer advertisements and more features? Click here to become a Hardware Analysis registered user.
Michael C Oct 16, 2010, 02:57pm EDT Report Abuse
>> Re: gpu temp. too high???
Nope, that's perfectly normal. I have two 4870's and they average anywhere from 55c - 80c.

Plug & Play Oct 16, 2010, 03:12pm EDT Report Abuse
>> Re: gpu temp. too high???

Yeah, As above posted, thats perfectly fine operationg harm to make sure you have no dust etc on the card itself and make sure you have decent airflow in the case itself. Also the card will let you know when its too hot as it will more then likely start to artifact etc.

i5 2500K @ 4.8Ghz- Corsiar H50 WaterCooler- Coolermaster Realpower 1000w- Asus P8P67 Deluxe - Asus 6990 4GB - 8GB Corsair DDR3 2000Mhz - X-Fi Sound - 7.1 Surround Speakers - BenQ 24" TFT - G9x Mouse- G19 Keyboard
G. G. Oct 17, 2010, 06:30pm EDT Report Abuse
>> Re: gpu temp. too high???
Well it might be normal operational temps or within specs.... but for me personally.... it's getting a bit warm...

I have a single HD4870, with fan settings set at a fixed 38% in CCC, I get 41C idle and while doing BFBC2 it runs 59~63C.


john albrich Oct 18, 2010, 05:20am EDT Report Abuse
>> Re: gpu temp. too high???

One needs to be wary of temperature data (and voltage and timing data as well). They can be good guidelines, but they shouldn't be always assumed to be dead accurate, especially if you've never calibrated the sensors with the temperature monitoring program(s) in use. And re-calibration may be required at surprisingly frequent intervals. Such caution is especially warranted when people are comparing readings between two entirely different systems, different programs, different environments, etc.

Even the base temperature sensor data accuracy can be unreliable.

I've seen significant +/-C GPU, CPU, and HDD temperature variations running under the same conditions, depending on which simultaneously running temperature monitoring program was being viewed. The temperature logs from the different programs also show similar variations for min/max values. For example, for a temp reading taken at virtually the same time by several different programs, you might see Everest shows one value, Speedfan shows another, GPU-z shows yet another. One reason this can happen is simply due to differences in the algorithms the program author used to treat the raw sensor data. Another is that the program may not be reading the sensor it claims it is. E.G. instead of the die temp, it may be displaying the case temp...and the algorithms used on the raw data vary depending on which sensor is supposedly being examined. Standards have improved things considerably over the last few years, but that's still a problem that crops up without warning.

I've also seen temperatures (using the same monitoring program) reported as running within a certain range (e.g. 38-43C) one day, and while running under the "same" environmental, program, and etc conditions, report a significantly different range the next day (e.g. 41-50C). Not only a shift up in temp, but a widening of the range. (room ambient temp maintained steady at same temp)...yet nothing was done by the user that would account for such variation.

I've also seen CPU, HDD, and GPU temps reported at several degrees C below steady-state device-adjacent ambient....which is an impossibility with simple air cooling. (multiple adjacent ambients obtained using a lab-grade calibrated temperature monitoring datalogger)

This isn't to say throw-out such data, but be aware of possible limitations that could lead one to make incorrect conclusions and take the wrong actions.

Plug & Play Oct 18, 2010, 07:19am EDT Report Abuse
>> Re: gpu temp. too high???
Without getting into specifics, the question was "Are they normal operating temps" and the short answer is YES....well according to the manufactuer they are. I would be maybe a tad worried about the normal temps.....they are in retrospect a tad high for idle.

Yes you should always aim to get operating temps as low as possible to prolong the life of components. I have found since day one that the CCC panel is not URGENT enough when it comes to temp management.

With my 5970 if left to the CCC panel it only starts to up the fans when temps are very high, where as like G.G said, I force a set fan profile to 45% constantly and my 5970 never goes above late 40's even when gaming for hours......but this has alot to do with my case setup and watercooled CPU which helps to reduce the overall heat in my rig.

i5 2500K @ 4.8Ghz- Corsiar H50 WaterCooler- Coolermaster Realpower 1000w- Asus P8P67 Deluxe - Asus 6990 4GB - 8GB Corsair DDR3 2000Mhz - X-Fi Sound - 7.1 Surround Speakers - BenQ 24" TFT - G9x Mouse- G19 Keyboard
john albrich Oct 18, 2010, 08:35am EDT Report Abuse
>> Re: gpu temp. too high???

I always end up changing my GPU fan profiles to more conservative settings too. As said above, even if the processor is operating within specified limits it is always a good thing to reduce the temperatures when possible. It can only help extend the life of the electronics and I'm more than happy to knock a bit off the fan's lifespan to in order to help the electronics.

In fact, I always design my fan profiles under the assumption that the real temperature is 5C higher than the reported temperature. That way even if there is an error or bias in the reporting I've at least got a 5C margin to work with.

It just ticks me off that even some of the more expensive Graphics cards still use a Cheapo sleeve-bearing fan.

Ian Musial Jul 25, 2012, 11:08pm EDT Report Abuse
>> Re: gpu temp. too high???
Actually that's pretty low for that card, my HD 6850 will go as high as 80 (on 100% load,) but it cools itself down relatively fast.

It could just be that the thermal paste on the GPU is getting old, or is dying down (happens with heat + time.) Also remember due to it getting old dust will crop out of nowhere insanely (though most DECENT pastes don't leave a terrible amount of dust, sometimes it's barely noticeable.) Buy some Artic Silver 5 (my personal favorite,) and apply ASAP.

Also remember, when you're applying new thermal paste it MAY take awhile for it to burn in completely. After it burns in, you'll notice your temperatures are going to be as low as 60, or even lower... (Mine goes as low as 50 depending on the temperature outside and humidity.)

Also be CAREFUL when applying new thermal paste, keep in mind the paste expands out like yeast (though not as dramatically.)

Hopefully this helps.



  Topic Tools 
RSS UpdatesRSS Updates

  Related Articles 

A weekly newsletter featuring an editorial and a roundup of the latest articles, news and other interesting topics.

Please enter your email address below and click Subscribe.