New system failing during gameplay (BSOD, APPCRASH)

imbavafan

Honorable
Dec 10, 2013
3
0
10,510
Hello!

I completed a new build last week, but I've been having some struggle with it while playing games, especially BF3/BF4

System
OS: Win 8.1 Pro N 64 (6.3, build 9600)
Mobo: MSI Z87-GD65
BIOS: Ver 1.7 (04.06.05)
CPU: i7-4770k
Memory: Corsair Vengeance Low Profile DDR3 PC12800/1600MHz CL10 2x8GB
GPU: Asus Radeon R9280X-DC2T-3GD5 (latest 13.11 9.5 beta driver)
PSU Corsair RM850 850W
SSD: Samsung 840 EVO Series MZ-7TE250 250GB (latest firmware)

Everything is running in stock and no overclocking has been done. All drivers are up to date and I've used driver-sweeper (safe mode) when re-installing them. I have also made a clean re-install of windows 8.1 and only installed essential drivers. I also made sure being properly grounded to earth during the build by using a ESD-bracelet connected to earth.


The issue at hand is that while playing BF3 (has been my main test game) it either crashes with either BF3.exe, sometimes its the atidxx32.dll as the cause. More often it results in a BSOD with ntoskrnl.exe as cause. I have been playing some Black ops, which has resulted in one crash of the game, it took a whole lot longer then the BF3 crashes.


BluesScreenView gives me these dump files

U3Hy1Eh.jpg

This bug check code has occurred 5 times

Wiy8MdZ.png

This has occurred once


I've ran memtest86 for about 3 hours with no error (maybe not long enough)
I do however had some error occurrences while running CPU-OCCT with errors in either core 1 or 2 after about 30 min. This has happened twice and been running for 4 hours straight with no errors on other occasions, the maximum temperature when running CPU-OCCT has been 70C as a maximum. Also, I've ran prime 95 for 5 hours straight with no errors. I read somewhere that OCCT might cause errors on "healthy" haswell-chips, so I don't know how to trust those results.

Also, when running the OCCT-PSU test the application crashed after 30 min with a direct X failure.

I have tried using only 1 mem stick while playing, also I've tried switching to the other two memory sockets (1-3 2-4) on the mobo, I have even tried using another R9280x card (have two), all with no results. Also I've done diagnostics on the SSD and come up healthy, I've switched SATA cable and port with no results.

I've ran whocrashed and it did not find any offending third party drivers.

Something I should mention is that the mobo is off by about 2mm to the right (not all the way back to the IO-shield) in the chassi. The specific chassi is a Fracal Design Define R4. The result is that the screw-holes for the GPU does not fit perfectly to the chassi, but it is possible to get the screws in. The gpu is firmly in place, but I do not know if it is completely inserted in the PCI-express slot. It is however inserted properly in the back with the pin pulled up and in place, in the front however it might be off by 1 mm or so but I cant really tell if it is inserted all the way or not. It is at least inserted as far as the chassi allows it to. It might be far fetched with the chassi theory due to the fact that crashes only occur during gameplay

FeE5CDz.jpg

Picture of the mobo offset

So what is my conclusion? I don't know, either it's the amd beta driver which is causing the problems or it is my hardware failing on me. Or problems with the insertion of the GPU due faulty chassi construction.

What I need is input from people who know more than I do, which is why come here asking for help. I hope the content is not all over the place and that I gave enough information regarding the issue.

I am happy to answer further questions which (most likely) will arise.



 
Solution
you mentioned it could be the beta amd drivers, have you tried rolling the driver back? i doubt that's the solution, but, couldn't hurt to try. are you overclocked at all?

iatemyelf

Honorable
Jun 14, 2013
48
0
10,560
you mentioned it could be the beta amd drivers, have you tried rolling the driver back? i doubt that's the solution, but, couldn't hurt to try. are you overclocked at all?
 
Solution

imbavafan

Honorable
Dec 10, 2013
3
0
10,510


The Asus R9 280x comes factory overclocked, but I have tried downclocking it to AMD stock settings without any result. Also, the GPU temperature runs around 70C while playing BF3.

No, I have not tried rolling back with the drivers, I did not know it was possible since the 280x is brand new and there are only beta-drivers available.

Would it be possible to go with 7970 drivers since 280x is based on the same card?
 

Kurdain1

Distinguished
Nov 30, 2007
154
2
18,715
Just a note you should ground yourself to the chassis you are working on. You want the parts, yourself and your installation all at the same 'potential'.

It seems that you have enough power supply, and you already went down the path of testing memory and CPU. Have you looked at temperatures on the motherboard itself? Perhaps a chipset problem...?

Also may want to try seating the video card all the way in the slot and forego screwing it in just for testing purposes.
 

imbavafan

Honorable
Dec 10, 2013
3
0
10,510


I was grounded to earth using this technique
https://www.youtube.com/watch?v=0bUghCx9iso

Regarding mobo temps, they are around 45C in high load, and about 30C in idle. I may look into unplugging the GPU screwes from the chassi and see if it gives me any result