|
Details of problem:
Running games that are graphically demanding such as Witcher 2, BF3, Metro results in a crash, in my case, meaning a black screen, no BSOD, audio constantly repeating, cannot Ctrl+Alt+Delete, must force shutdown by holding power button. Prior to the crash, the game will boot up, run perfectly for 2-3 min with no visual lag or anything indicating a crash is about to occur.
Solutions and measures conducted:
1. Set BIOS settings to default. (I never overclocked or changed anything to begin with since I had no real need to.)
2. Ran MemTest for 4 Passes. All successful, no errors.
3. Ran Prime95, In-place large FTT for 2 hours, and Blend test for 2 hours. No crashes, excess temperatures, errors.
4. Ran Furmark, 1280x720 Burn in Test. Crashed when temps reached approximately 60-65, repeated 2 more times, crashed under similar temperatures
5. Ran Witcher 2, Metro 1920x1080 Low Settings. Monitored temps with HW Monitor. CPU around 40-45C, GPU climbed from 35 (idle) to 60-65 ish, then crashed. Repeated for both games, crashed at similar conditions.
6. Ran SC2 (less GPU demanding) 1920x1080 High settings, no crash (GPU never reached 60-65).
7. Updated BIOS drivers, AMD GPU, Intel iGPU drivers.
8. Repeated steps 2-6, all produced same results
Relevant screenshots of CPUZ during Prime95, HWMonitor during Prime95, and MemTest
http://imgur.com/a/6R0Uk
Recent changes to hardware/software
Ordered a replacement front I/O panel for my Corsair Carbide 500R case due to my dog tripping over my mouse cable that was connected to a front USB port. The USB port was no longer working due to a broken pin inside of the port. At this point, I could still play games with no issues. When I replaced the panel, I made sure everything was connected to where it was previously, I did not change or move around anything else, and there were no cut wires, no bent things that are not supposed to be bent, or issues that I saw that could cause the issue.
System specs
OS: Windows 7 Pro 64 SP1
RAM: Gskill 8GB (4GBx2) 1600 9-9-9-24 1.5V
CPU: Intel i5 2500K 3.30 Ghz
Mobo: ASRock Z68 Extreme3 Gen3
iGPU: Intel HD Graphics 3000 (ASRock)
GPU: ASUS HD7950-DC2T-3GD5 Radeon HD 7950 3GB 384-bit GDDR5 PCI Express 3.0 x16 HDCP
Audio: RealTek HD Audio
PSU: Rosewill CAPSTONE-650 650W Continuous @ 50°C 80 PLUS GOLD, ATX12V v2.31 & EPS12V v2.92
Miscellaneous Info and Questions
I know the most probable cause is the GPU but my temperatures are not even close to what would be considered dangerous.
1. Is it possible for the temperature increase from 60C to something considered damaging that quickly, to where HWMonitor cannot report a higher temp due to taking too long to post newer readings because of the time intervals it operates on? (Hope this made sense.)
2. I know this card has had bad reviews (I bought it when 7000s first came out, never had any issues except for coil whining while games are first loading up and even then that might be the PSU). Can the GPU processor become detached from the heatsink even while screwed on? Reason I ask is the heatsink+fan setup seems kind of flimsy, and don't want to start unscrewing GPU as it will possibly void warranty.
Sorry if I posted too much unnecessary info just tried to be as thorough as I could, thanks for your help.
|
|
|
|
|
On June 23 2013 20:09 JonIrenicus wrote:Try the following step: 1- PSU: Did you try changing it? 2 - Update your directx via this tool : http://www.microsoft.com/en-us/download/details.aspx?id=35Test. See if nothing changes. 3- Use CCleaner and clean the registry. 4- videocard: I wonder, if, by changing it, you could fix everything.
1. Don't have access to an additional PSU.
2. Already have most up to date DirectX
3. Cleaned registry, had a few issues detected but most of them were from old MS updates, tested, failed.
4. I've thought about this too, but I only have access to a GTX 560 Ti in a few days, would I just remove the AMD drivers if I were to swap cards in order to avoid a driver conflict?
Downloaded Afterburner and underclocked, my games were stable longer, and so was Furmark but once temps reached around 65C, my PC crashed again.
|
United Kingdom20322 Posts
How easy is it to reach 65c? What do your idle temperatures read as? Did you try adjusting fan curves to hold your temps below that?
My GPU goes over 65c no problem (plateau's at 92c within a minute or two of max load in a game) so if it takes you a bit of effort to get over 65c you might have a miscalibrated sensor or something, like if it reads as 65c when the GPU or parts of the card (vrms etc?) are reaching far hotter temps, you could have problems
What clock speeds did you try when you underclocked?
|
On June 24 2013 08:22 Cyro wrote:
How easy is it to reach 65c? What do your idle temperatures read as? Did you try adjusting fan curves to hold your temps below that?
So I timed how long it takes to reproduce same problem, using Furmark at 1280x720 (default) Burn in test, it took 49 sec until crash, running at 800 Mhz / 1250 Mhz.
Used Metro 2033 to see if game would crash as well with new clock settings, took approx 2 min 35 sec until crash.
My GPU goes over 65c no problem (plateau's at 92c within a minute or two of max load in a game) so if it takes you a bit of effort to get over 65c you might have a miscalibrated sensor or something, like if it reads as 65c when the GPU or parts of the card (vrms etc?) are reaching far hotter temps, you could have problems
Here is screenshot of my idle temps using HWMonitor: http://imgur.com/egLlegu
Temps before crash on Furmark (had to use camera could not use anything on comp)http://imgur.com/iyIV6kv:
Temps before crash on Metro 2033: http://imgur.com/VD6bUUl (My GPU has not melted, nor do I smell any smoke)
Fan curve setup: http://imgur.com/y8YFwzc
What clock speeds did you try when you underclocked?
Used 900/1250 (these were factory settings), and then underclocked to 800/1250.
|
United Kingdom20322 Posts
900mhz is the factory overclock. 800mhz is stock settings for the gpu, try 700/1100, change your fan profile so that it spins up to 80% by 55-60c, if it takes 2 and a half minutes to reach 65c at full load then something's probably up
It's a weird issue though i'm not really sure what to say here, just blind adjustments. It's a bad bandaid solution and you should be able to run GPU intensive games with the factory overclock (try to see if you can do that with the fan spun up enough?) but you should be able to see if you only crash when GPU reads higher than a certain temperature, etc
Read your temperature and GPU stats from MSI afterburner
|
Sounds like a possible temp issue. Take the heatsink off your GPU if youre comfortable doing that. Make sure its not clogged with dust. Make sure the fan on your PSU isnt seized up and causing your psu to overheat.
ASRock boards are not the best quality but lets assume its doesnt have to do anything with that yet.
|
So I've decided just to RMA the card, contacted Asus this morning, card is on its way to their facilities. Will try and use a different card (have access to a GTX 560 TI) and see if it gives me problems. If it doesn't, then it was the card. If it's still a problem, then its probably the mobo, correct?
On June 24 2013 20:59 Cyro wrote:
900mhz is the factory overclock. 800mhz is stock settings for the gpu, try 700/1100, change your fan profile so that it spins up to 80% by 55-60c, if it takes 2 and a half minutes to reach 65c at full load then something's probably up
It's a weird issue though i'm not really sure what to say here, just blind adjustments. It's a bad bandaid solution and you should be able to run GPU intensive games with the factory overclock (try to see if you can do that with the fan spun up enough?) but you should be able to see if you only crash when GPU reads higher than a certain temperature, etc
Read your temperature and GPU stats from MSI afterburner
When I went down from 900/1250 to 800/1250, it just took longer to reach failure. I guess I don't want to have to go down to 700/1100, because I may not reach failure, but what's the point of having a 7950 if you have to downclock so much. I understand what you're getting at though.
As far as the fan curve, I did increase to a more aggressive fan curve (just took longer to reach failure, or didn't reach failure but was reaching 55-60C so I felt I was on the cusp of crashing.) You're right though, it would be a bandaid solution, and the card should be working properly. In addition, looking at the Newegg page from where I bought it from last year, it's been discontinued and has had poor reviews, so I figured I'd just go thru the RMA rather than screw something else trying to fix this issue.
|
|
|
|
|
|
EPT
