• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

OCCT GPU Error Testing

Caporegime
Joined
8 Sep 2006
Posts
39,152
Location
On Ocuk
I've always ran my GTX 260 at 730/1485/1209 because it passed furmark and did many loops of vantage and other benchmarks. However today i thought i'd try the occt gpu tester , within 5mins i got 1 error :o. Did another test and as soon as the card hit 70c the tester shut off and a nvidia error came up 40+ errors (100% fan speed).

Had to drop it down to 720/1494/1200 and now temps are 56c max with 90% fan speed. Will test more but , it's supprising how just a different tester can show things others can't
 
thats not surprising really, occt was designed from the ground up to tbe a gpu testing tool, thats why it has an artifact tester built in. also as seen on xtreme systems occt is the best gpu tester around, its like the linpak of the gpu world.
do note that the load occt puts out is insane and its been known to show weakness in ati cards due to thier shoddy power supply implementation triggering the overcurrent protection.

your card seems to clock better than my gtx260, i get max temps of 76c on my gainward gtx260 216, clocked at 700 core, 1440 shader, and 1161mhz memory. and thats 12 hour occt stable. (occt ran at fullscreen 1920x1080, NO AA or AF forced in the nvidia control panel).

do note that memory testing is very hard, occt doesnt use much gpu memory at all, so a lot of gaming is best way to test the gfx memory stability.
 
Try EVGA OC Scanner - I've found that the GPU tester in OCCT isn't that stable itself and errors out more due to bad programming than actual GPU problems.

Also bare in mind due to the dividers on the 200 series clocks 1485 and 1494 will probably be pushing the shaders to 1512* (check with GPU-z or similiar to see the actual shader clock). I've found 720/1440 seems to give the best performance in that clock area but again check as some boards are made a little differently.


* An average GTX260 will top out at either 1458 (or 1456 can't remember off the top of my head) or 1512 actual clocks.
 
^ Interesting Cyber , didn't know that Thanks :)

It seems to be that from my testing , clocking the memory is what makes the temps sore , it's giving me no errors at 725/1494/1200 and temps remain at 55-57C. But clocking the memory beyond 1200 and those temps sore


EDIT : Thanks Rroff will do
 
I haven't actually tried OCCT GPU tests in awhile but last time I did they used bad methods with the CUDA implementation which resulted in crashing or errors that weren't actually the GPU failing - it could be fixed however in newer versions.
 
Try EVGA OC Scanner - I've found that the GPU tester in OCCT isn't that stable itself and errors out more due to bad programming than actual GPU problems.

iv ran occt on many cards, nvidia and ati, only time occt errored out was when there was an actual gpu error. then again you could have other system stability issues, occt causes major gpu power draw so its possible your voltage regulators gave up or psu was drooping.

does this EVGA oc scanner have artifact tester in it?

i could run it on my machine and check with the wall plug meter to see what pulls more power, occt or evga oc, gpu temps would also be uaual indicator of heat load produced.
 
Yeah the EVGA OC Scanner is similiar to the old ATI Tool it scans for artifacts (including non visible ones) as well as stress tests.
 
Can someone upload the EVGA OC Scanner ? , i've just gone to register to get the file and it says " this page cannot be found " after i entered all my details

EDIT : You must have a registered evga card to download the scanner file :mad::mad:
 
Last edited:
Its game stable vs Stability program stable.

My 470 is 810 core occt stable with temps going mad and 850 game stable with temps relatively cool.

So if game is stable with your clocks just stick with them.
 
I dropped my clocks and the temps sored again 73C then blue screen but this time with furmark!.I had to hard reboot to get the system back. Not sure why this is going on as i remember doing furmark loops early in the year and the temps never hit 70 and no blue screens. I wonder if it's these beta 260.* drivers?

Having no problems with games but i just wanted to stability test it
 
what were the clocks you used when it blue screened on you?

720/1483/1200 , this is lower than i game at!. For example 6hrs of left 4 dead 2 last night lol

EDIT : This isn't a BSOD but a total blue screen , like all colour no text ;)
 
Last edited:
sure you not got a loose monitor cable?

those clocks still seem quite high to me, then again i always assumed i has a duff clocker of a card, but in actual reality since i perform very stringent overclock stability testing my clocks are not bad at all.

why not lower you clocks down more, try the clocks i use as a starting point and see how they fair for a 12 hour occt/furmark run.
 
My 5870 will get thousands or errors within a minute of OCCT, at stock, overclocked or underclocked, so I wouldn't say its perfect video card stress tester in my experiance. I would try a good combination of other stress testers and demanding game benchmarks to get the best test of a GPU's stability. I doubt the 73c temps are the problem, the gpu may just need more juice.
 
Last edited:
one thing to mention, occt uses direct3d, furmark uses opengl. i had issue before with furmark not working but i figured it out, my 3 year old nephew had pratted with nvidia control panel and turned open gl extension limit on which stopped furmark working on the pc.

also what version of furmark were you using? was it the latest version?
 
I just read on another forum of a number of people having issues with gainward and palit cards and the beta 260 drivers. Will try these quadro ones instead :)
 
Back
Top Bottom