TR Forums

Good evening all,


I am having some serious PC problems and it's making me one hell of a sad panda.
The PC is either freezing or flat out turning off and I believe the ATI 4870 is to blame.

My theory is that ATI have set the maximum temperature threshold to be excessively high before the fan kicks in, so they get good reviews for having 'such a quiet card' at the sacrifice of amazingly close to max thermal limit products
When my machine is running full pelt in a game, the card wouldn't be on more than 40 or 50% fan at most, yet the heat that is coming out,.. well it's nuts.

Now the problem is, I don't want to manually set my fan to be 100% all the time (or have to manually change it before / after games) - yet the auto mode is causing the machine to crash.
Hell even my E8500 is underclocked to 2.55ghz, the GPU is running at 600mhz and memory 700mhz (I don't know what stock is, but I do know I damn well dragged it a fair bit to the left!)

So what's the deal here? Is the card faulty? Can I get software which makes the thermal threshold lower, so the fan turns on earlier?
Can I make the fan automatically go to 80 or 90% as soon as certain executables are run perhaps?

Any suggestions?
I'm als running this thread regarding my PSU, as I suspect it could also be the problem.
viewtopic.php?f=16&t=63608

AbRASiON wrote:

Can I get software which makes the thermal threshold lower, so the fan turns on earlier?

Yes.
RivaTuner
Set up some fan profiles and launcher items.
You can also use it to set fan speeds by temp.

I will scope that out, thank you.

Has anyone else had similar issues? It's a bit frustrating to have some nice hardware and I have to underclock it.
I'm a bit worried that even with 100% fan speed, it'll crash if I set speeds back to normal :/

I have a better theory, it seems your card isn't properly placed on the heatsink module. I would take the card apart then clean the heat sink and place it properly over the card. Make sure your system is well ventilated.

shank has a good point... what is your case ventilation like? Please post with specifics about your case and fans (size, brand, speed, and location) -- that'll help a lot.

Indeed, if you're worried about crashing even at 100% fan speed then there's something else going wrong. I've got my 4870 set to 35% constantly (anything higher than that is too loud for me) and I hit around 65C under load.

Does not crash at 100% load, only crashes at 'auto' fan load.

Also I doubt the heatsink is lined up wrong.
Reasoning, if the heatsink was lined up wrong, logically the GPU would be overheating because it's not passing enough heat to the GPU cooler and therefore it's getting 'stuck' at the CPU (makes sense to me)
Unfortunately, in reality the entire system, PSU included is becoming amazingly hot, so the CPU or GPU is building crazy amounts of heat and passing to the PSU OR the PSU is so heavily strained it's generating that much heat.
(to summarise, if the GPU heatsink wasn't on right, the actual heat in the machine wouldn't be that high, it'd be 'stuck' down near my GPU)


On to the specs.
Antec SX-635-II case (same as the performance one series if I recall)
80mm fans at front x2, 1 exhaust fan back, dual slot standard ATI cooler on 4870, standard intel heatsink and finally PSU has 120mm coolermaster fan in it.

The case DOES have foam padding inside now, I'm thinking perhaps this is screwing up the airflow
http://www.pccasegear.com/index.php?mai ... ts_id=6141

AbRASiON wrote:

When my machine is running full pelt in a game, the card wouldn't be on more than 40 or 50% fan at most, yet the heat that is coming out,.. well it's nuts.

Changing your fan setting won't appreciably change heat output. If the card's putting out that much heat, it's putting out that much heat. Fans change temperature, not heat output.

mattsteg wrote:

AbRASiON wrote:

When my machine is running full pelt in a game, the card wouldn't be on more than 40 or 50% fan at most, yet the heat that is coming out,.. well it's nuts.

Changing your fan setting won't appreciably change heat output. If the card's putting out that much heat, it's putting out that much heat. Fans change temperature, not heat output.

Well based on the design, perhaps the heat is becoming trapped in the case at a low speed fan and then rising up to the PSU, rather than being quickly blown ou tthe back.

I would give an estimation that the PSU is at least 176' f with low fan speed on the GPU and that it would be 140 or less with a high speed GPU fan.

AbRASiON wrote:

I will scope that out, thank you.

Has anyone else had similar issues? It's a bit frustrating to have some nice hardware and I have to underclock it.
I'm a bit worried that even with 100% fan speed, it'll crash if I set speeds back to normal :/

My x1950xt will throttle down 25-50 mhz or so if i run overdrive, other than that it runs fine though. It's not unreasonable to call it faulty and get a replacement. Is it under warranty?

Heiwashin wrote:

AbRASiON wrote:

I will scope that out, thank you.

Has anyone else had similar issues? It's a bit frustrating to have some nice hardware and I have to underclock it.
I'm a bit worried that even with 100% fan speed, it'll crash if I set speeds back to normal :/

My x1950xt will throttle down 25-50 mhz or so if i run overdrive, other than that it runs fine though. It's not unreasonable to call it faulty and get a replacement. Is it under warranty?

It's a lot more difficult to get support in Australia.
They will likely honour it but it'll be sent back to a factory and I'll be without a video card for at least a month at a guess - no 'over the shelf replacements' here :/

So if I do send it back, I wanna be damn sure it fails and it's due to the card and not the PSU / system cooling around it.

Have you tried removing the side panel while keeping the fan setting at auto?

AbRASiON wrote:

Heiwashin wrote:

AbRASiON wrote:

I will scope that out, thank you.

Has anyone else had similar issues? It's a bit frustrating to have some nice hardware and I have to underclock it.
I'm a bit worried that even with 100% fan speed, it'll crash if I set speeds back to normal :/

My x1950xt will throttle down 25-50 mhz or so if i run overdrive, other than that it runs fine though. It's not unreasonable to call it faulty and get a replacement. Is it under warranty?

It's a lot more difficult to get support in Australia.
They will likely honour it but it'll be sent back to a factory and I'll be without a video card for at least a month at a guess - no 'over the shelf replacements' here :/

So if I do send it back, I wanna be damn sure it fails and it's due to the card and not the PSU / system cooling around it.

How heavy is your concern of noise. You could always by some vantec tornadoes to slap on the case, but the 80mm ones have a whine about the volume of a hairdryer.

I can't stand noise - hence the low noise (expensive!) fans I purchased.
Not sure what to do, even considering moving to an Antec P150 / Solo, I believe they have 120mm fans in them, which are quieter AND move more air :/

GPU is now 71celcius (160'f) in windows, use is 0% GPU, fan still locked at 26%

I had a SX635 for a number of years (purchased when they first came out - 2001ish?).

Putting a 2008 heat monster video card AND three hard drives in that thing AND insulating the case (which of course trades off noise for higher temps) AND running low-speed 80mm fans does seem like a bad combination in that case.

If you've got two optical drives in there, you should have one open 5.25 drive bay that you can take the front off and see if the extra air helps drop the temps at all. That case doesn't have a lot of airflow coming in to begin with, IIRC (the second front fan is the one right in front of the HDDs, no?), and you're stacking the deck even more against it.

True, true but it is a high quality well built case with amazing features for something 5 years old.
Slidable optical design like the modern ones, removable HDD 'cage' - air filter, dual front fans etc.


The second fan is going through the HDD's yes.
I wish I had a spare video card, then I could just see if it's the ATI or not straight away.


I think I'm just going to have to bit the bullet, do the Antec Solo and some high end power supply, plus 2 extra 92mm fans, hopefully should keep noise at the same levels it's at now at least but much cooler.

As I recall, it was an extremely well-built case (very nice fit & finish, HEAVY steel) - I got rid of it when I replaced it with a Coolermaster Centurion5, which was replaced by a P182 at the beginning of the year - so I'm going off memory here.

The lower bezel wasn't that open, IIRC, and unless your design is slightly different than mine was, the fan in front of the HDs isn't going to help with system airflow, since you've got the drive cage filled up, if you placed three drives in there.

You may indeed want to pop off the side or one of the free 5.25 drive bay openings, and see if that does make a difference.

If you decide on a different case, you can also try your existing PS first before deciding to replace that. It may very well work properly again, once you get it and the rest of your parts out of the sauna I suspect they currently reside it.

Mine kept displaying artifacts after a while with the "auto" setting.
Using the Overdrive utility within the Control Center, I boosted the fan speed up to 45%. No problems since.
Heat problems with this card are not uncommon.
~CC5

what click said; if it was a gpu heat issue you would see graphical artifacts first.

if this happens when the system is stressed (games, etc), it could well be a psu issue. you could always try swapping in old parts if you have some laying around (diff video card, diff psu).

ClickClick5 wrote:

Mine kept displaying artifacts after a while with the "auto" setting.
Using the Overdrive utility within the Control Center, I boosted the fan speed up to 45%. No problems since.
Heat problems with this card are not uncommon.
~CC5

Interestingly, I never get artifacts, it works completely fine - then I get looping audio (freeze) or black screen re-boot or finally, least often the whole machine just turns OFF!

Frustrating indeed.

Cyco-Dude wrote:

what click said; if it was a gpu heat issue you would see graphical artifacts first.

if this happens when the system is stressed (games, etc), it could well be a psu issue. you could always try swapping in old parts if you have some laying around (diff video card, diff psu).

Yeah but @ 100% fan, no issues at all last night.

Oh bloody great! I finally found some more information.

Google for "113-AB68800-XXX" - there's a bios out which fixes problems for SOME of the ATI cards, but the fix won't apply to my card because it's the wrong P/N - yet I'm having an identical fault to other people.

AbRASiON wrote:

Oh bloody great! I finally found some more information.

Google for "113-AB68800-XXX" - there's a bios out which fixes problems for SOME of the ATI cards, but the fix won't apply to my card because it's the wrong P/N - yet I'm having an identical fault to other people.

If you're feeling adventurous, you could give the BIOS editor a shot and try to flash that BIOS onto your card.

Managed to force the bios on and it worked ok last night - will update thread on how it goes over the coming week.

Card hasn't crashed in ages, does still run hot but summer is over now.

AbRASiON wrote:

Card hasn't crashed in ages, does still run hot but summer is over now.

What's "hot" for you? GPU routinely over 90ºC and VRMs above 115ºC is hot.

morphine wrote:

What's "hot" for you? GPU routinely over 90ºC and VRMs above 115ºC is hot.

Hah. I remember that. Did you ever find a solution for that?

mortifiedPenguin wrote:

morphine wrote:

What's "hot" for you? GPU routinely over 90ºC and VRMs above 115ºC is hot.

Hah. I remember that. Did you ever find a solution for that?

Nope. Putting the stock cooler back in when I can be bothered (and writing a rant thread here). I haven't been playing much lately.

Thought I'd chime in. Now 6 months old, my ati4870 has been doing the same horrible thing for months. It took me months to discover that it is the 4870, and not the power supply.

turns out, a few things are weird. the gpu fan doesn't actually kick in at all when the card is inactive. so the heat slowly builds up, until the gpu is 110C, then the fan goes 100%, and in ten seconds it's back down to 50C. that cycle works fine. in-game or under load, the fan works normally and everything is fine. but not under load, the machine can restart at any moment.

'turns out that the overheating isn't the final problem -- which is why it was so hard to find, the machine would reboot when the gpu was at 60C too. The problem was that after the card heats up, it draws more power, to the point where 750W PSU isn't enough. I discovered this after running with the case open for a month, and noticing the diagnostic lights on the board saying power-disconnect, not heat problem.

So currently, I have my desk fan blowing into the case permanently. Which is not an actual solution. I'd like to get some kind of third-party cooler.

The real problem, obviously, is my case structure. I've got the 4870 in the second pice16 slot, so I don't lose my pcie1 slot. I'm maxed-out on boards in this thing. So the 4870 is pretty low, and with zero fan, it hasn't a chance. I guess the cards presuming that 0% gpu activity means it won't heat up too much is quite incorrect.

Sounds more like a cheap PSU. 750W is way overkill. What brand and model of PSU are you using? Your symptoms sound more like a CPU not regulating power well.

If it is the PSU, check this link on what you really need for a PSU: http://www.anandtech.com/casecoolingpsus/showdoc.aspx?i=3413