Why do I have to underclock my GPU to avoid TDR errors?

MegaHamster

Reputable
Feb 11, 2016
3
0
4,510
I am trying to work out why I have to underclock my graphics card in order to get it to run stable.

Here is my system:
Operating System
Windows 10 Pro 64-bit
CPU
AMD FX-4100
RAM
8.00GB Dual-Channel DDR3 (9-9-9-24)
Motherboard
ASUSTeK COMPUTER INC. M5A97 EVO R2.0 (Socket 942)
Graphics
SyncMaster (1920x1080@59Hz)
L70S+ (1280x1024@60Hz)
1023MB NVIDIA GeForce GTX 550 Ti (Gigabyte)


For a number of months the card was crashing with this TDR error whilst playing games.

Display driver nvlddmkm stopped responding and has successfully recovered.
Error 4101


Not just one particular game, a huge range of games. Too many to put it down to one particular game. Most recently I got this error playing Skyrim.

I have tried several things in order to fix or even properly diagnose this issue.

Have updated to the latest drivers. This has been an ongoing issue and I always have the latest drivers. The issue has persisted through many driver updates.
I have run my RAM through multiple tests. It's fine.
I have cleaned inside my PC, the card is not dusty and was doing this when it was new anyway.
In the NVidia control panel in "Manage 3D settings" I have changed "Power Management mode" from "Adaptive" to "Prefer maximum performance".
I have tried a different power supply which did not fix the issue. My current power supply is stable and does not fluctuate even under load.
Different operating systems (was happening under Win 8 & 8.1 as well)
Other things I cannot remember but will add in here when I remember/am reminded.

In the end the only thing I can do to maintain any stability and stop these "crashes" is to underclock my GPU with MSI Afterburner to these settings:

Core Clock (MHz)
Original: 970
Stable setting: 899

Memory Clock (MHz)
Original: 2050
Stable setting: 1904

I'd really like to get to the bottom of what is actually causing this to happen. I've tried several times over quite a few months and always hit a dead-end, got bored of reading endless articles/posts or got frustrated with trying different solutions that don't work when all I want to do is play games in peace! I have no idea where to go from here to work out why this is happening. Can anyone give me some insight?
 
If you tried a different power supply, that points to a heat issue or a failing video card. One thing I don't see you trying is a clean Windows setup. Although I don't think that will help, at least it's something you can try before replacing the card. Did this start happening after Windows 10 setup or have you always had 10 on that system?
 

MegaHamster

Reputable
Feb 11, 2016
3
0
4,510


The card was originally installed with Win 8. It's been through Win 8.1 and Win 10. Win 10 was a totally fresh install on a new HD.

I've actually had the card tested by Nvidia and they couldn't find anything wrong with it and sent it back to me.

So it must be something in my system, but what? I've been through everything I've read about and this is the only solution but it feels like a band aid and obviously I'm not getting the performance I'd like from my card.

I have tried a new power supply, I upgraded to a better one to see if that would fix it. It didn't. =/

 

MegaHamster

Reputable
Feb 11, 2016
3
0
4,510
It's not something I can do easily. I have upgraded mobo in the time I've had this card though so I'm leaning towards it being the card that is the problem but then since it has been tested by Nvidia that doesn't seem likely either.

I have read that the RAM could be the issue. I have tested the RAM several times and never found it to be an issue. I did read something this week about checking that the RAM was set up as per manufacturers instructions in the BIOS but I have no idea how to do that. Seems like a long shot.

Given that I have upgraded both mobo and power supply it seems likely it's either the card or the RAM but I have no idea how to isolate it either way.
 


You isolate the part by testing it on it's own. Which is why you need to either replace it and see if the other one works fine, or test yours in another system and see if the issue follows it. No other way.
 

Tushar_94

Commendable
May 9, 2016
3
0
1,510
I am facing the same issue with my GPU. Got a used zotac 560ti a week back. Got it for 70$. Crashes within 5 mins while playing crysis and nfsmw12 when run at stock core clock 822 MHz memory clock 2004 MHz. However no crashes after I underclock to 720mhz and 1900 MHz respectively. It's not too bad considering the price I paid for such performance but I am really interested to know as to how I can make it stable at stock clocks.
 


It's a bad card, you can't make it stable. If it was working properly you'd be able to run it overclocked never mind running normally at stock speed.