EVGA GeForce GTX 660 SLI: black screen sound loop crash on games, bad GPU?

I have a build using two GTX 660's in SLI. I bought one 660 in August last year, and another in February. Both of these GPU's were purchased through Amazon Warehouse, so there's no chance of return. The problems I am experiencing seem to stem from a failing GPU, and I was hoping you guys could help me either confirm or disprove my suspicion. So here are some symptoms I have been experiencing:


Note: This doens't happen in all games. Usually more demanding games trigger the crashes, but less demanding games sometimes cause a crash too. This never happens while doing low maintenance stuff like browsing, watching videos or listening to music.

SLI
1. Black screen, sound loop. Monitors shut off. Requires hard reset and doesn't always make a .dmp, Video (Crash occurs around 1:50)
2. Bad skipping in some games, especially Battlefield 4
3. Driver has stopped responding and has recovered message, very rare.

Newer GPU ONLY
1. DirectX function "GetDeviceRemovedReason" failed with DXGI_ERROR_DEVICE_HUNG
2. Sporadic driver crashes/recoveries
(never requires a hard reset, crashes don't cause system lockups)

Old GPU ONLY
PC works fine, no issues.

I've tried using different power cables, rearranging GPU's, a different SLI bridge, complete removal and reinstallation of latest and older drivers, one monitor instead of two, undervolting and underclocking GPU's, uninstallation of all GPU performance value editing programs such as MSI Afterburner, RivaTuner, EVGA Precision X, etc., but no dice.

This is what the dumps say, all of them:

(short version)

On Tue 4/8/2014 1:51:43 AM GMT your computer crashed
crash dump file: C:\Windows\Minidump\040714-14570-01.dmp
This was probably caused by the following module: nvlddmkm.sys (nvlddmkm+0x9BBE2C)
Bugcheck code: 0x116 (0xFFFFFA800B04A010, 0xFFFFF88011E09E2C, 0xFFFFFFFFC000009A, 0x4)
Error: VIDEO_TDR_ERROR
file path: C:\Windows\system32\drivers\nvlddmkm.sys
product: NVIDIA Windows Kernel Mode Driver, Version 335.23
company: NVIDIA Corporation
description: NVIDIA Windows Kernel Mode Driver, Version 335.23
Bug check description: This indicates that an attempt to reset the display driver and recover from a timeout failed.
A third party driver was identified as the probable root cause of this system error. It is suggested you look for an update for the following driver: nvlddmkm.sys (NVIDIA Windows Kernel Mode Driver, Version 335.23 , NVIDIA Corporation).
Google query: NVIDIA Corporation VIDEO_TDR_ERROR

(long version)


Since I've been requesting help elsewhere, people have been suggesting that the PSU could very well be at fault. Either my system draws to much power in SLI or the PSU is faulty. The reasons I disagree with this is:
- Happens with only new GPU and with SLI, but not with only the old GPU
- SLI with FurMark + Prime95 for 10+ hours miraculously doesn't crash the system, but both cards reach just over 85*C and don't end up crashing.


Anyone have any ideas? If you could help me I would be so grateful. Even if it is confirming that I have a paperweight. Anyone?


Specs:

CPU: AMD FX-8320 4.4GHz @1.4v (having no OC on the CPU doesn't change the effect the GPU's have, so I keep it on)
RAM: Kingston HyperX Blu 2x4GB DDR3 1600MHz
GPU's: EVGA GeForce GTX 660 2GB (believed to be faulty card is temporarily taken out)
MB: Gigabyte GA-990FXA-UD3 rev 4.0
HDD: Seagate Barracuda 2TB 7200RPM SATA III 64MB Cache
HDD2: Hitachi Deskstar 1TB 7200RPM SATA II 32MB Cache
PSU: OCZ ModXStream PRO 700W, estimated peak system draw: 680W
OS: Windows 7 Ultimate 64bit
3 answers Last reply
More about evga geforce gtx 660 sli black screen sound loop crash games bad gpu
  1. can you do a warranty through evga
  2. i had a fair amount of issues with my 660's, gigabyte ones. sell them and get a better single card, that is my 100% solution. Things that did help me is use the older 314.xx drivers, and use nvidia inspector to force a constand clock speed, rather than have a boost clock. doiong those two things increased my stability 10 fold and removed stuttering, but still ocassional crash to desktop's. Final solution was to sell both cards and get a 7970 which is rock solid stable, and less fps fluctuations.
  3. If the card turns out to be dying, I can't sell it for much, but the other leaves me $170ish towards a 780. I'll just have to wait until my next paycheck to get them. And I cannot do an EVGA return, as the card was purchased through Amazon Warehouse, which I believe does not count as one of their authorized sellers.
Ask a new question

Read More

EVGA Games SLI Geforce Graphics Gtx GPUs Build