Stress Testing Multiple GPU's Simultaneously

[outlw6669]

Hey all!

I am looking for your input on the best way to simultaneously test a pair of 4850's.
I have been having occasional random crashes while gaming/folding over the past few months and have had some difficulties tracking down the cause.
All components test fine individually which leads me to believe the issues only arise when every component is being loaded fairly heavily.

I have been testing my GPU's one at a time with ATI Tool while running prime95 in the background.
As I have a display attached to each GPU, it is no trouble to switch a card to primary and individually test it in this method.
I have tried to add the '-gpu0' and '-gpu1' switch to a pair of shortcuts to no avail.
If anyone has any suggestions on how to get dual instances running, one for each GPU, please let me know.

Also, if anyone has any suggestions on other testing apps, feel free to suggest them.
Unfortunately my trusty old RaptorX bit the dust over the weekend and took my cache of testing software with it

My full system specs can be found in my profile.
Here are the most important parts:

ASUS P5Q-E
Q6600 @ 3.2Ghz
6Gb DDR2 1066 5-5-5-18
Corsair 1000HX
2x 4850 512Mb's overclocked to 690/1100Mhz through a BIOS flash, brought back down to 625/995 with the CCC.
All drivers are up to date.

Any and all ideas are welcome!

Thanks for reading
/outlw

 

[TerminatorXT]

all of you on this forum are cruel
you never helped him

First of all why are you running Prime 95 during the stress test,
use furmark for the stress test, it will max out your first CPU core, then run prime95 on all cores, then disable one

I would start off by updating all drivers
then test it,
then I would try without prime95

then down the overclock, (if they still are)

and if all else fails, try one card at a time, to see a dud

 

[jsc]

I, along with about 50 others, saw this thread when outlw first posted it. I didn't have anything to add so I didn't post gratuitously just to run my post count up. Outlw knows this. In spite of his "apprentice" rating, he has been around here several years.

all of you on this forum are cruel
you never helped him

First of all why are you running Prime 95 during the stress test,

All components test fine individually which leads me to believe the issues only arise when every component is being loaded fairly heavily.

And that means after loading the GPU's, load the CPU cores.

 

[outlw6669]

Yeah, it is really not that big of an issue.
If it was not a somewhat tough question, I never would have needed to make the post.
As it is something I was unable to locate a solution to, I did not necessary expect anyone else to have the answers either.

TX, thanks for the thoughts.
Unfortunately they will not work to good in my case.

The procedure you suggest would work for a short stability testing.
In my case, however, my system is stable nearly all of the time but randomly crashes once in a while when it is under a heavy load (both GPU's and the CPU loaded).
Since I already know that there is an issue, I need to locate exactly what is causing it.

When I test my components individually, overclocked or not, there are no issues up to ~16 hours of testing (when I decide to call it quits).
What I need to do is to fully load all components in a way that can not only be sustained for quite some time (not a quick 10 min single shot test) but also will alert me to just what is the component that fails (as compared to just failing).

What I like about prime95 and ATI Tool (or any similar programs) is that they allow me to load my system for an indefinite period (as it is a random issue, I am not sure how long this might take) and give me a wall time for failure for which ever CPU core or GPU fails first.
If I can figure out a way to do the tests as described above, I will have a repeatable way to confirm exactly what component is causing my issues.
Personally, I would prefer to know for sure just what is failing before I decide to shell out cash for new parts.

jsc, speaking of being here for quite a while, they finally promoted me to Admiral lol!
Fetch me some swabbies to scrub down the head :lol:

 

[unclefester]

I use Video Card Stability Test 1.0.0.1 http://www.sharewareconnection.com/download-video-card-stability-test-from-sharecon.html But not running dual video cards I don't know if you can run 2 apps. at the same time. I do know that while it's running I can use EVGA Precision Tool and OC it while it's running.

Outlw6669 over 2200 posts and still an Apprentice!!! Gotta love this new rating system. LMAO

 

[outlw6669]

Wow does that load your system!
Not only does it slow my pointer to a slide show on the primary display, it also chews up ~4Gb of extra RAM and a good amount of CPU power
Unfortunately I am not able to get it to run on my secondary display

 

[unclefester]

Ya, I'm really surprised more people don't use that little program, it taxes thee ol' system in a hurry. You probably notice it had your video card up to load temps in about 30 seconds. Plus with Precision Tool you can move the sliders, OCing on the fly.

Although I'd didn't think it would run 2 apps. at once (dang it ) But thanks for letting me know!!
UF

 

[outlw6669]

It does tax the system at that

The real problem I am having is not running two apps at once, rather running any app on the secondary GPU when CrossFire has been dissabled