Red boxes followed by graphics driver failure

amazingbrian

Honorable
Nov 23, 2013
20
0
10,510
So whenever running a furmark test or just playing a game (BF4 In this case) my graphics card would crash and i would get these red boxes all over my screen then the game would just stop responding. This is happening in temps that are about 60c-65c is this normal?

Whenever i run a benchmark test with EVGA OC Scanner X (http://i.imgur.com/BcLhjjH.jpg picture of results) it would go just fine, but after i run a game or furmark after a few seconds the card would just fail and stop working for a few seconds. Anything i need to do?

Here are my specs,

i5 3570k @ 4.2Ghz (OC)
660Ti +70 MHz Gpuclock / +500MHz Memclock
8GB Ram
Mobo - MSI z77 G45
PSU - 650W LSP Ultra lifetime series


P.S I had reseted my clock speeds back to default and the red boxes/crashing had stopped.
 
Solution
I would guess the scanner is not as hard on the card as either of those.

It is also important to know that Nvidia cards do strange things since they monitor power consumption. For instance, if it run OCCT's video card stress test, my card will clock to about 1045 while games can push it as far as 1215(as long as it stays within the power limit)

This makes testing stability a bit more interesting on these cards as every program can push the card in a different way.

I would start by testing memory and gpu clocks separately as I am not sure how much the oc scanner tests the memory.

Personally I had a video card(and this was all just stock boost settings) that would NOT crash under furmarks(1006mhz), but would die in Bioshock...

I think you solved your own issue.

Sounds like the overclock is not stable.
 

amazingbrian

Honorable
Nov 23, 2013
20
0
10,510


Well i'm still confused to why my Overclock worked with EVGA's OC Scanner benchmark but then Running Furmark or a video game ended up crashing the driver :/
 
I would guess the scanner is not as hard on the card as either of those.

It is also important to know that Nvidia cards do strange things since they monitor power consumption. For instance, if it run OCCT's video card stress test, my card will clock to about 1045 while games can push it as far as 1215(as long as it stays within the power limit)

This makes testing stability a bit more interesting on these cards as every program can push the card in a different way.

I would start by testing memory and gpu clocks separately as I am not sure how much the oc scanner tests the memory.

Personally I had a video card(and this was all just stock boost settings) that would NOT crash under furmarks(1006mhz), but would die in Bioshock infinite(1162). Simply put, you have to test the overclock more than one way since the cards will limit them selves under certain conditions.

You can use a tool like GPU-Z to check in real time the clock rates your card is running at.
http://www.techpowerup.com/downloads/SysInfo/GPU-Z/

Cpus are similar when overclocking while it may be prime stable it may crash under lynpack or AVX loads.

Personally, I find folding@home stresses a cpu just the right way to make it crack under any circumstances.

Refine the clocks and see what you can get to in games is the best advice I can offer for now.
 
Solution

Latest posts