GPU or CPU crashing?

[justin4pack]

So today is the first time a have been able to game in a while. I decided to play some star forge. its a pretty graphic intense game. So after about oh 30 min or so i get a black screen with grey vertical lines. Now after overclocking the cpu (i7-920) and the memory running at close to 1600, i ran a mem test as well as load test for around an hour and no issues. After checking the GPU-z log it shows the gpu ( Diamond 7870) was only at like 60 % load and Temp was around 58-60C. So i dont think thats hot enough for the GPU to overheat but i also dont think its the cpu/mem so im puzzled.

 

[cmdrdredd]

It could just be a game crash unrelated to your hardware. It happens.

 

[justin4pack]

It could just be a game crash unrelated to your hardware. It happens.

I should be more specific, when it did this, the system was unresponsive. i had to reboot. I wouldn't think a game crash would make it that way?

 

[cmdrdredd]

I should be more specific, when it did this, the system was unresponsive. i had to reboot. I wouldn't think a game crash would make it that way?

It can, it happened to me a lot in buggy games like Fallout 3. I could loop memtest indefinitely, run prime95 for a week, and loop any 3d benchmark indefinitely with no errors or crashes and zero WHEA errors. Yet some games lock everything up.

 

[2is]

It sounds like a GPU crash to me. Screen corruption usually is. I had (still have actually) a 4850 that would do that in certain games (Dirt 2 mainly) and the cause was the actual card was going bad. Not saying your card is bad, but if it were my system, I'd certainly look at troubleshooting the GPU first with that sort of instability. First and foremost I would make sure the drivers are up to date and that the card is running at it's default clocks.

 

[justin4pack]

It sounds like a GPU crash to me. Screen corruption usually is. I had (still have actually) a 4850 that would do that in certain games (Dirt 2 mainly) and the cause was the actual card was going bad. Not saying your card is bad, but if it were my system, I'd certainly look at troubleshooting the GPU first with that sort of instability. First and foremost I would make sure the drivers are up to date and that the card is running at it's default clocks.

Yea its a new to me card, stock clocks and most recent drivers, haven't had it happen with any other games yet. its possible its common ati issues? i had a 5870 that i had issues with and i sold it and got a gtx 480 at the time. no more issues.

 

[2is]

I've had a lot of AMD/ATI cards (probably more so than NVidia actually) and I've had similar corruption with all of them at one point or another. That said, it was always due to an unstable card because it was overclocked. With the exception of my 4850 that started failing, I have not had that happen to a stock clocked AMD GPU.

How many monitors do you have connected to the card?

 

[justin4pack]

How many monitors do you have connected to the card?

Only 1 tv, at 1360x768 via hdmi

 

[BigChickenJim]

I should be more specific, when it did this, the system was unresponsive. i had to reboot. I wouldn't think a game crash would make it that way?

Oh yes, bad crashes can and will hard lock your system. It happens all the time. If you can consistently replicate the issue in numerous games I'd say you have a dying GPU, but if it's only happened once in one game it's almost certainly a software issue.

I believe you are fine, my friend.

 

[cmdrdredd]

Oh yes, bad crashes can and will hard lock your system. It happens all the time. If you can consistently replicate the issue in numerous games I'd say you have a dying GPU, but if it's only happened once in one game it's almost certainly a software issue.

I believe you are fine, my friend.

This is what I am thinking. You can run numerous tests that are more demanding than any game like 3dmark and heaven benchmark. Loop heaven for a couple hours and it won't crash on a good system. Try other games too

 

[Piotrsama]

Go back to stock on CPU, Memory and GPU and try again.
Probably an unstable OC.

 

[hackerballs]

the 920 cannot be pushed too far in OC.................but I agree that it could just be a game issue, happens all the time if you game regularly................

 

[justin4pack]

So i put all settings back to stock on cpu/mem. Ran bf3 for around 10 min and it happened again. Possible driver issues? GPU load was 100% and Temp was 72c. Possible heat issue? will have to test more after work.

 

[Termie]

Is the GPU overclocked?

Which Diamond 7870 do you have? This one? http://www.newegg.com/Product/Produc...scrollFullInfo

That cooler isn't particularly robust, but even 72C should be OK. Unless you have a driver conflict, the issue is not the drivers. As a first step, I'd uninstall the drivers entirely and then reinstall after a restart, to make sure it's not a conflict.

 

[justin4pack]

The drivers are fresh as this is a new build. is there a way to force the fan rpm 100 percent

 

[BigChickenJim]

So i put all settings back to stock on cpu/mem. Ran bf3 for around 10 min and it happened again. Possible driver issues? GPU load was 100% and Temp was 72c. Possible heat issue? will have to test more after work.

72C is nowhere near the thermal failure threshold of a 7870. Most cards are rated to 90-100C. My 7870s often run at 65-70C for hours at a time in games like Metro 2033 and Crysis 3 with no issues and I've pushed them to almost 80C in synthetics with no failures. Your issue isn't temp-related unless your GPU VRMs are overheating, and that seems unlikely. Download HWInfo64 and check your VRM temps under load just to be sure.

Since the issue is now occurring in different games it isn't a software issue. The issue recurring at stock rules out a bad OC, and I'm confident you don't have a temp issue. I'd run DriverSweeper and reinstall your video drivers just to be sure you don't have a conflict going on. If that doesn't work, you've likely got a bad card or PSU. I'd bet it's the card.

 

[2is]

Sounds like the card to me as well. Pretty much identical to what was happening to my 4850. There was certain games it would always crash on within a few minutes of playing, others where it would last a little longer on, others completely random.

 

[justin4pack]

Its this card http://www.newegg.com/Product/Produc...82E16814103208

Guy who i bought it from said that he never had it go over 59C, he thinks its the temp. my current case is not the best so i have a better one on the way. Inturn i setup some temp case fans to reduce heat and also set manual control of the fan to 60%. was idle at 40c is now idle at 33C. will play some bf3 and ill post back.

 

[BigChickenJim]

Its this card http://www.newegg.com/Product/Produc...82E16814103208

Guy who i bought it from said that he never had it go over 59C, he thinks its the temp. my current case is not the best so i have a better one on the way. Inturn i setup some temp case fans to reduce heat and also set manual control of the fan to 60%. was idle at 40c is now idle at 33C. will play some bf3 and ill post back.

With all due respect, I think the guy who sold you the card is wrong. No modern, properly functioning GPU is going to fail completely at 72C. You'd see throttling first, and even that wouldn't start until somewhere past 80C. If it is hitting a thermal failure threshold at those temps there's something very wrong.

I'm all about you buying a better case and adding some fans--cooler is always better--but I don't think it's going to fix your problem. If the card is failing no amount of airflow will save it. I could be wrong, but I'm still banking on the fact that your card is dying.

How'd your BF3 session go with the extra cooling?

 

[2is]

My 680s are at a pretty consistent 78C when I'm gaming. And that's with an aggressive fan profile to try and keep them below 80. Not that 80 is dangerous for a GPU, but at that temp, the boost clocks start to throttle.

Even if you are able to mitigate the issue with lower temps, you shouldn't have to. A stock clocked card should be able to play any game you want for as long as you'd like with the default fan profile.

 

[Symcy94]

My Graphic card GTS450 always goes to 100% usage and temperature 70°C, but never crashed. As far as I know, graphic cards can survive extreme temperatures (even up to 100°C) before start dammaging itself.

And how many FPS do you have ingame?

I had like 400 FPS in KalOnline, so I decited to limit it on 60, since there is no point to have Frame rate over 60 ... my game didn't had function to limit it, so I downloaded d3doverrider and everything is now limited on 60FPS. Card usage doesn't go above 30% and 60°C.
But it seems you are running much stronger game.


You also mentioned that it's new build. You reinstalled windows completly and formated HDD before start playing with new build? It could be the problem sometimes, when you just change few components in PC and don't delete everything on the HDD (in this case settings and drivers from previous components are still on PC and may couse trubles). You might want to take this in concider, and I hope you will fins solution to fix that problem

 

[justin4pack]

So since the fans, i have had no issues. does that still point to a bad card if 70c was causing lock up?

 

[BrightCandle]

The problem with instability problems is they are hard to track down. Always get rid of any overclocks when trying to track them down. You need to run a series of tests to isolate particular components first:

1) Memory check with memtest, possibly for a day. Let it run multiple times completely.
2) Prime95 and LinX running at the same time across all your cores. You aren't looking for heat so don't try and optimise the work down just overutilising the cores. You will want at least 8 hours of this, I would again aim for about 24 hours.
3) GPU test. In the past I have used looped 3D mark and if you do it third then you can be pretty confident its not the CPU and not the memory.

If 1 fails its your RAM. 2 almost certainly wont fail because CPUs don't fail but we need to be sure, if it fails its the CPU. If 3 fails it could be the GPU or it could be the PSU or even the motherboard. In my experience memory is the most likely component to be the cause, in about 60% of these cases its been RAM that doesn't run in its spec. Then its been the PSU and finally I had one case where it was the motherboard. I have never had a GPU failure manifest in this way so my advice is starting thinking and swapping components more widely.

 

[blackened23]

Go back to stock clocks on everything (cpu, gpu, memory) before asking such a question.

Crashes happen sometimes. A lot of times due to software and not necessarily hardware. But GO BACK TO STOCK if you want to rule things out.

 

[Termie]

Its this card http://www.newegg.com/Product/Produc...82E16814103208

Guy who i bought it from said that he never had it go over 59C, he thinks its the temp. my current case is not the best so i have a better one on the way. Inturn i setup some temp case fans to reduce heat and also set manual control of the fan to 60%. was idle at 40c is now idle at 33C. will play some bf3 and ill post back.

If he says his card never went above 59C, he is either lying or he never used it for games. There is no way a reference 7870 could stay below 70C during intense gaming.

So since the fans, i have had no issues. does that still point to a bad card if 70c was causing lock up?

I game on my 7870 at 70C-72C all day long. I don't think that is your problem, especially now that I see it's the reference cooler, which should adequately cover VRMs and other important components.

But if you are in fact able to game at 100% fans without a single crash, well, then, color us all surprised!