Graphic card driver crashs and recovers during stability test

[sub-80]

Hey guys;

Am overclocking my graphics card, I run occt and evga oc sanner x to double check that there are no errors (artifacts). I have overclocked it and when I run evga scanner it finishs whith no errors and the same with occt but the driver crashs and recovers within a second. I reinstalled the driver again and it still crashed with occt. Now am running occt at graphic card default setting to see if driver crashs. Even though I think its instability that caused driver to crash and recover, I still would like to hear your opinion.

Update1: ran occt at stock setting no driver crash going to overclock again.

 

[tweakboy]

Increase the fan speeds. open room door and open window,, youll be set then.

 

[hawtdawg]

Its your overclock. That's exactly what my 780's do when pushed too far or arent given enough voltage. I never get artifacts, just a momentary lockup and a driver crash.

 

[sub-80]

Yeah thats what I fingered. Reached 1140/1192 (base/boost), 6208 (memory), 0.925-1.188v (0.025V increase in max volt range) and power target 101%.

 

[sub-80]

This is what am getting when I run occt and I got it once when I stopped stress testing the gpu and started a video to give the gpu a 30 min break.

But the strange part is that the driver is not crashing on evga scanner and I also ran it once on furmark for 15min.

All the tests I did on occt and evga are 1hr duration.

 

[sub-80]

This has happened with only occt or when i tried to open a ext hdd a few seconds after a stress test and when playing a 10 bit 720p a few seconds after a stress test. Tried to do the stress tests again and levels that I had cleared with occt with no errors and with no driver crashing, had the gpu driver crash and recover and had errors in the thousands, so I thought this must be occt problem as I have noticed Furry E GPU Memory burner from Evga pushed the core clock reading an extra 130mhz and the core voltage was close to the max and I had read from a couple of forums stating similar problems with the error checking not the driver crashing.

To be abit at ease and to provide some info to you guys I ran some test which the results and settings are written below.

I have run all the following tests in 1920x1080p with 8x MSAA for 1 hour.
EVGA OC Scanner X v3.4.0.1
4M Particles:

1. Lasted Test Duration: Yes
2. Driver Crashing: No
3. Artifact Detection: None

Furry Donut V2:

1. Lasted Test Duration: Yes
2. Driver Crashing: No
3. Artifact Detection: None

Furry E (GPU Memory Burner: 3072MB)

1. Lasted Test Duration: Yes
2. Driver Crashing: No
3. Artifact Detection: None

Tessellation + DoF:

1. Lasted Test Duration: Yes
2. Driver Crashing: No
3. Artifact Detection: None

• I ran Tessellation + DoF test before at the same setting before I figured that driver crashing might be instability which currently am not sure of, but I had it run for 15 min and during that time the driver did not crash but there was a presence of artifact 1 to 4 artifacts every min, but during the last test as you can see there were none.

Greek 3D Furmark 1.12.0.0
3D Test options checked: Dynamic Background, Burn-in, Xtreme burn-in,and Post-FX

1. Lasted Test Duration: Yes
2. Driver Crashing: No
3. Artifact Detection: N/A

The highest GPU temp in all tests was 75C.



My card is Gigabyte GTX 780 GHz edition.
Core clock: +120MHz
Memory Clock: +250 MHz
VMin:+-0V
VMax:+0.0375V
Target Power:+2% (Power and temp not linked)


What do you think?

 

[rgallant]

What do you think?
Update1: ran occt at stock setting no driver crash going to overclock again.

none of those gpu stress testers are valid any more in the real world game stable oc for testing on a gk110 ,with power limits ,boost tech, temp down clocking and what ever

-use fire strike x, heaven , if the system does not crash , then use a demanding game for final oc testing imo

 

[sub-80]

Tessellation + DoF:

1. Lasted Test Duration: Yes
2. Driver Crashing: No
3. Artifact Detection: None

• I ran Tessellation + DoF test before at the same setting before I figured that driver crashing might be instability which currently am not sure of, but I had it run for 15 min and during that time the driver did not crash but there was a presence of artifact 1 to 4 artifacts every min, but during the last test as you can see there were none.

Greek 3D Furmark 1.12.0.0
3D Test options checked: Dynamic Background, Burn-in, Xtreme burn-in,and Post-FX

1. Lasted Test Duration: Yes
2. Driver Crashing: No
3. Artifact Detection: N/A

Turns out I ran these stress tests with stock settings when I tried the overclock settings for the DOF test it detected 4-6 artifacts every 30 seconds then it crashed after 10 min giving me a popup telling me my graphic card can't handle the work load.

Will try that rgallant