Issue with gpus v ram

[bigbeasteric]

Okay so I'll try to make this short as I can, so this has been happening recently, just when playing any game for example battlefield 4 after what I would say maybe 20 minutes my fps drops dramatically. So as of today I tried playing but my fps dropped from about 100fps all the way down to 20. I used evga's program for gpus so I can try to see what was going on, the only thing I was able to uncover is that the gpus v ram almost doubles on both cards. Any knowledge on this would be much appreciated thank you.

Specs:
Cpu: intel core I7 6700k
Motherboard: gigabyte z170x gaming 7
Gpu: gtx titan 2 way sli
Ram: G Skill Ripjaws v 32gb
Storage: Samsung 850 evo
Psu: Corsair 1000i

 

[Cycloid Torus]

If "gpus v ram almost doubles on both cards" means what I think it means then these folks would respond by questioning if you have an effective SLI..... http://www.tomshardware.com/answers/id-2868995/video-cards-double-vram.html

 

[LNCPapa]

I believe he means his VRAM usage almost doubles, not that he has twice the capacity.

 

[Cycloid Torus]

I believe he means his VRAM usage almost doubles, not that he has twice the capacity.

Maybe so, it was a bit unclear to me. Not sure that merely doubling usage in a powerful system like his would hobble FPS, but it seemed if the SLI was not functioning properly that this could be a sufficient issue for the symptom.

 

[bigbeasteric]

Maybe so, it was a bit unclear to me. Not sure that merely doubling usage in a powerful system like his would hobble FPS, but it seemed if the SLI was not functioning properly that this could be a sufficient issue for the symptom.

yeah I meant the usage doubles, Sorry for being unclear

 

[Cycloid Torus]

Either way, still got to figure out how to fix...drop to 20 FPS is staggering. Perhaps it is a card / hardware issue... What happens if you only run with one card? Can you test each card by itself? How do they compare? If one card had a heat problem then tracking temp to FPS would isolate an issue.

 

[bigbeasteric]

Either way, still got to figure out how to fix...drop to 20 FPS is staggering. Perhaps it is a card / hardware issue... What happens if you only run with one card? Can you test each card by itself? How do they compare? If one card had a heat problem then tracking temp to FPS would isolate an issue.

yeah I was thinking it was a hardware problem, but the heat at least last time I checked, shouldn't be a problem but I'll check again. Next time I can probably tomorrow I will test both card at once, and then separate and tell you tell you the results. Thanks for the suggestions.

 

[bigbeasteric]

Either way, still got to figure out how to fix...drop to 20 FPS is staggering. Perhaps it is a card / hardware issue... What happens if you only run with one card? Can you test each card by itself? How do they compare? If one card had a heat problem then tracking temp to FPS would isolate an issue.

Okay so I got what you asked for, I tested each one by itself and I also tried a different SLI bridge. Each test was done on the same game with same graphics. All temps are at the point of fps drop.

On my first test I did it as my system was. The drop happened after a little over 12 minutes of game time. Gpu 1 ran at 52 degrees Celsius, Gpu 2 ran at 58 degrees Celsius.

Next test was done with a different Sli Bridge. The fps droped happened a bit over 13 minutes. Gpu 1 ran at 51 degrees Celsius, gpu 2 ran at 54 degrees Celsius.

On this test I ran only one gpu, took again a pinch over 12 minutes for the drop to occur, and it ran at 50 degrees Celsius.

And the last test, on the other gpu. It took about 13 minutes for it to drop and ran at 55 degrees Celsius.

I really hope these will help you thanks again.

 

[Cycloid Torus]

No easy answer.

What did we (pretty much) prove? I think we proved that it isn't either of the cards and it isn't gpu heat.

Does the FPS stay permanently at 20 or does it yoyo? Is CPU temp below TjMax - or could we be seeing CPU throttling?

I would inspect the attachment of the heatsink on the of chance that a push-pin has failed (I have replaced all of mine with 1/4-20 nylon bolts & nuts). A loose heatsink on the Northbridge (if you have one) could do the same thing.

Speccy and HWInfo utilities might identify any runaway heat conditions.

Lastly, I would make sure that the PSU fan is working (my PSU exhaust reaches 120F with the fan working - I'ld hate to consider what it would do without). I stuck a thread on the grate to have visual confirmation.

If this doesn't result in problem identification, then I would go to remove, download fresh again and reinstall drivers.