Jump to content

Graphic card overheating.


Recommended Posts

I have an AMD Radeon HD 6870 and it started to overheat around a week ago.

It's gotten to the point that i cannot play any games on the computer.

I tried playing Batman Arkham Asylum as a test and after less than ten minutes the computer crashed.

I cleaned the inside of the computer and the card for dust and removed the old thermal paste and applied a new one (Arctic Silver 5 if it's important) two days ago.

Before cleaning and applying new paste the heat was at 60c when browsing the internet and now it stays at 48-50c.

I tried playing Batman again with the cards fan on 50% speed and played for an half hour with no problems before i quit.

During play the heat rose to around 70-80c if i remember correctly.

The fan speed stays at 24% when on automatic when not playing games so i don't know if theres a problem with just the fan or something more.

Plan B is to buy the new AMD Radeon card that's comming out later this year.

Link to comment
Share on other sites

May want to look at the fan on the videocard to see how easily it spins or if there is any damage to the wires or connections. Taking a look at your power levels might also be an idea. May want to play a reasonably demanding game in windowed mode just to see if the fanspeed ramps up. May want to actually look at your case and see if there is a problem which might limit the airflow to and from the card (sometimes people excessively blow out the inside of the case with compressed air, but neglect the various exhaust and intake grills). 40-45c is about the normal temp that AMD components maintain themselves at, so even on idle something is keeping things hotter than they should be.

 

Frankly, I wouldn't plan to buy the latest and greatest when it comes out. Usually these are overpriced, have numerous driver issues, and even have hardware conflicts with older hardware (particularly for power requirements). The top-line stuff is usually best left to die-hard enthusiasts who can sit down and deal with all the errors that might pop up without having to wait a few months for updates. Your best bang for your buck as far as AMD cards go would probably be among the R7 or R9 series since these will have the greatest amount of power without all the compatibility issues.

Link to comment
Share on other sites

I did go over the exhaust and intake grills while removing dust. I did a dust removal for the first time on my current computer some months ago when it was crashing and giving me blue screen (of death) due to some malware that had taken resident in it and driving up it memory usage.

I only took the card apart enough to remove the old thermal paste and apply the new one as i did not see any more screws on it and was afraid of breaking anything on it.

I already managed to break off the locking mechanism to keep the card in place while trying to remove the card.

 

Funny thing that you should mention the R9 series as the newest AMD card is a R9.

Link to comment
Share on other sites

The AMD 6870 had a fast-degrading set of thermal capacitors, if I remember. A lot of those got returned as well; prone to crashes, even if they were new cards.

 

When you cleaned out your GPU, did you make sure to use compressed air? Another thing to try is using a different PCI-E port for your GPU to be plugged in. As well, see what wattage your PC is actively running at. Finally, if you've had the card for a number of years, maybe it's time to buy another.

 

Note:

The R9 series from AMD (R9 290, R9 290X) usually operate around 75C during gameplay, on average. Apparently, they were designed that way.

Link to comment
Share on other sites

I cleaned the inside of the computer and the card for dust and removed the old thermal paste and applied a new one (Arctic Silver 5 if it's important) two days ago.

How did you apply it and how much did you apply? And since the overheating problem appeared suddenly, try popping the side-panel off while your PC is on and check the fans on the card, see if all of them are spinning like they should. Sounds to me like one of them might have stopped or slowed down.

 

By the way, I don't recommend using AS5 but since you already used it, stick with it. For future reference though, AS5 has a burn-in period of roughly 10 days after which it gets to peak performance (3-4C lower) but even at that point, it's not all that great compared to modern compounds. It was king 4 years ago, nowadays you have cheaper pastes like MX-4 and TX-2 that do the job as well or even better and have no burn-in period.

 

The R9 series from AMD (R9 290, R9 290X) usually operate around 75C during gameplay, on average. Apparently, they were designed that way.

All modern GPU dies are designed to handle constant operating temperature of 95C so unless you can boil water on it, it's fine. :wink:

Link to comment
Share on other sites

The card came with my computer that i bought three or four (maybe five) years ago.

I used compressed air while cleaning the computer both times.

Can't imagine the damage i would have done if i had used the vacuum cleaner instead. :laugh:

The card has been stable until now, never caused any problems for me before.

Il order a replacement fan for the card and see if i can't open it some more and get a better look at it.

I have no idea what a PCI-E port is or how i can check the wattage.

 

Werne: I put the paste on after removing the old and used a q tip to cover the rest of the chip and put some extra in the middle.

I don't know how much is too little or too much for the card so i can't answer that.

After using the AS5 paste it already was at a lower temperature so i don't know if i was just lucky with it.

I bought the paste off ebay for a low price. I don't remember the price right now so il have to check later.

I checked on the cards fan and on the other fan after cleaning and applying the paste and they were both spinning.

Link to comment
Share on other sites

The R9 series from AMD (R9 290, R9 290X) usually operate around 75C during gameplay, on average. Apparently, they were designed that way.

All modern GPU dies are designed to handle constant operating temperature of 95C so unless you can boil water on it, it's fine. :wink:

 

Um... 95c would be 203f and be well above the boiling point of water. Even the quoted ~60c is above boiling point.

Link to comment
Share on other sites

Werne: I put the paste on after removing the old and used a q tip to cover the rest of the chip and put some extra in the middle.

I don't know how much is too little or too much for the card so i can't answer that.

After using the AS5 paste it already was at a lower temperature so i don't know if i was just lucky with it.

I bought the paste off ebay for a low price. I don't remember the price right now so il have to check later.

I checked on the cards fan and on the other fan after cleaning and applying the paste and they were both spinning.

Oh dear god... Never spread the thing with a q-tip, you got q-tip cotton all over your thermal compound which makes it unable to conduct heat properly. Actually, don't spread thermal paste at all. You also have air bubbles in it because of uneven spreading, making it further unable to conduct heat.

 

Basically, start from scratch.

 

I'd strongly suggest getting Arctic MX-4 or MX-2, you'll likely short-circuit the card with AS5 if you use too much since it's electrically conductive and the 6870's GPU has ICs close to the die, MX-4 and MX-2 aren't conductive.

 

MX-4/MX-2 are dirt-cheap, buy either in a store. After you get it, take the card out, take the cooler off, clean out the thermal paste with toilet paper lightly dipped in pure alcohol, make sure the die and cooler base are both as clean as possible. Don't use wet wipes. Let them dry for a minute after cleaning them, don't touch the die or cooler base with anything other than the aforementioned toilet paper.

 

After they're dry, put a bit of MX-4 on the die, like so:

 

http://www.legitreviews.com/images/reviews/822/TRad2_GPU.jpg

 

You can use a bit more if you don't feel safe with just a little bit, you can't do any harm by using a bit more if the paste is non-conductive, just don't overdo it. Don't spread the thing this time, just screw the cooler in, cooler's mounting pressure will spread the paste. Then pop the card in and see what temps you get. I'm 217% sure temps will be a lot better after doing it properly.

 

EDIT: I forgot to ask, did you by any chance redid the power delivery heatsink? If you didn't, good.

 

Um... 95c would be 203f and be well above the boiling point of water. Even the quoted ~60c is above boiling point.

Normal boiling point of water is 99.97C (211.9F).

Edited by Werne
Link to comment
Share on other sites

Thermal paste should be a very thin layer. Basic recommendation is a drop about half the size of a pea, but I would use even less. The paste is intended to fill in any small imperfections in the surface for better heat transfer. Too much will actually cause the temp to be higher. If both of the surfaces were perfectly flat, no paste would be needed and you would get an even better heat transfer. But no normal device, either chip or heat sink is anywhere near perfectly flat, so we use a thin layer of thermal paste to make up for microscopic imperfections.

 

A long time ago I was at a GE sponsored school on electronic devices (hockey puck sized industrial SCRs) where the proper use of thermal paste was covered. According to the GE factory rep, that pic in the post above shows about twice as much thermal paste as they would recommend for a device that size. The optimum amount would be a very thin layer and very little if any would be squeezed out when the heat sink is screwed down using the recommended torque setting. Every device has a recommended torque setting, but very few people have the proper torque wrench needed to actually measure it - so - very snug without stripping the screws is usually considered sufficient.

Link to comment
Share on other sites

 

The R9 series from AMD (R9 290, R9 290X) usually operate around 75C during gameplay, on average. Apparently, they were designed that way.

All modern GPU dies are designed to handle constant operating temperature of 95C so unless you can boil water on it, it's fine. :wink:

 

Um... 95c would be 203f and be well above the boiling point of water. Even the quoted ~60c is above boiling point.

 

 

The boiling point of water is 100c or 212f, the Celsius scale is based on the freezing point and boiling point of water, zero being freezing and one hundred being boiling.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...