7900GT artifacting....

[Highland3r]

I know this is a known issue, with many GT's "going bad" after a few days not sure if this is the same issue or not (hence the thread!)

Noticed artifacting in the "deep freeze" test of 06 the other day (after fitting sinks to the memory regs) downclocked the cards to stock and still had the same issue.

Pulled the EVGA from PCI-e slot 1 (closest to CPU) and ran the tests again with the XFX card only. Same thing, artifacts at stock and overclocked (stock = 520/750, overclocked = 600/850)

Pulled that card and removed the ramsink from the regs. Stuck it back in (this time in PCI-e slot 1) ran again, no artifacts! Clocked it back up, ran again and no artifacts... Re-fitted the sink, ran again at stock and overclocked and yup! no artifacts.

Looked to be good news, the card had "magically" fixed itself it seems. Stuck the XFX back in slot2, and the EVGA in slot1 enabled SLI and tested again...

Bad news! Artifacts in 06 again

Can anyone confirm/deny if this IS an sign of the XFX's upcoming death? The artifacts do indicate that it might well be. HOWEVER the fact that putting the card in slot1 produces none suggests this might be related to something else?

Tried different drivers (91.xx) and the latest officials on the Nvidia webby, switched Tagan from combined to single rail mode. Theres a 120mm fan blowing over both cards. The sinks attached to the regs DO get hot so contact seems decent....

It's been suggested that backing off the CPU overclock might help things a tad, didnt get these issues initially with the XFX though (although Canyon 06 did produce random black screens, this was resolved with the reg sinks though)

Cheers guys!

 

[K404]

Just to confirm..either card, in slot 1 by itself causes no artifacts? Could you say if this is an SLI issue or slot 2 issue?

Have you had any other SLI running in the board? Can you try the cards in another SLI board?

K

 

[Highland3r]

Not tested the EVGA card in slot2 yet, that was on the cards last night but needed sleep and didnt get chance before work thismorning.

Need to switch the cards over and test at somepoint too, the mouting hardware on the maze4 pokes out a bit far on the EVGA atm which needs trimming to get it to fit...

Might whip the sink off the XFX again and see whether that solves the problem.....

No other SLI run on the board, and dont have another board to test SLI (until the yonah arrives anywho)

It;s sounding more and more like the XFX is broken tbh, want to confirm 100% before it gets RMA'd though, as this Slot1/2 issue is a little confuzzling at the mo!

 

[maverik-sg1]

A Few things to try:

1) BIOS flashing to a common BIOS (even tho drivers allows for diff cards, it's possible there is a comms issue)

2) different driver set

3) complete - strip down of chipset and GPU drivers, driver clean then re-install?

If you have already tried single card action and the same occurs then i guess the only other thing I could think of would be power, but if in single mode one works and one don't......

I feel your pain my friend - hope its not an RMA job

 

[Highland3r]

1's the next option! Flashed the EVGA to match the XFX's clocks, but not bios.

Tried 84.xx and the new 91's, been recommened some 83's so thats something else to try.

Tried 3 already, next option is a complete re-install!

Pretty sure its not a power issue, grabbed the powerstream from the Dothan rig and its doing the same!

Switched the cards over in the PCI-E slots now too, and thats not helped!

Its a real PITA! Funny thing is, if it artifacts in last test, kicking off 06 again causes artifacts. A quick driver reload and its all fine again!

 

[boardy]

I get the same thing on my XFX GTX's mate. I dunno what is causing it but we'll soon find out if it's heat or if the cards are dying.

Also for some reason '03 refuses to give me a score at the end, just hangs on the test complete screen with a big black block across where the score would normally show. Bloody annoying if you ask me cos I know it's big just looking at my fps.

Hope you sort it out matey!!

Boardy

 

[maverik-sg1]

Have either of you considered lowering the CPU clock? I find that I can't run 03/06 at the same clocks as 01/05/aqm3.

Mav

 

[Highland3r]

Have either of you considered lowering the CPU clock? I find that I can't run 03/06 at the same clocks as 01/05/aqm3.

Mav

Yup, 2.9ghz -> 2.0ghz made no difference.

 

[maverik-sg1]

Yup, 2.9ghz -> 2.0ghz made no difference.

I think your is more BIOs/Driver related mate - hopefully not RMA.

Boardy - I truly believe that yours is either power or cpu clock related.

Cheers

Mav

 

[boardy]

I've tried 2.16ghz, 2.6ghz, 3ghz, 3.2ghz and all pass up to the point of the last 2 scenes of Deep Freeze. I did lower the RAM speed on the card and that gave me some improvement and got me my 11k score but I forgot to get the save file off before reformatting the drive Stupid bloody thing!

I'm convinced its the MOSFETs overheating cos they are bloody hot when you touch em at that point. I may try putting some water blocks on them.

Boardy

 

[boardy]

Mine can't be power mate, there is no possible way.

I have a 600W PSU dedicated to the GFX cards alone, that is definately not the issue.

Boardy

 

[maverik-sg1]

Boardy tis a weird one my friend.

I got me a 120mm fan blowing on both mosfet coolers - you are right they do tend to get very hot to the touch without any active cooling and the last two tests of 06 are the most power hungry. The cooling block on my innovatek block is a thick piece of alluminium spurring off the main waterblock - I fitted OCZ RAMsinks to the alluminium block also to draw the heat away.

I recall scorchio using in individual PSU just to power the cards - I think it worked better than other ancillaries were powered by the 2nd psu and the mainboard and GPU's were powered by the same PSU - not 100% sure though mate.

Mav

Of course it could be driver related for you also, or even the two XFX card may have different BIOS installed, worth a try?

 

[Highland3r]

Given up now!

New drivers, flashed both cards to the same bios same problem! Getting lockups/hangs in 03 as well now

RMA time, gonna try and get an EVGA card back so fingers crossed things will work a bit better this time

 

[boardy]

It's a very strange one on mine indeed. Both cards are running perfectly fine I can get 760 / 910 on them both which is giving me about 43.5k in '01 on single card!! Put her in SLI and it's a different story, my stock SLI score dropped by 10k from having these lines going up an down the screen.

I don't think my cards are coldbugged as I was running last night on single card @ -17c so I was barely even pushing the core to what it's true potential can be and that was @ 1.65v.

I can see RMA coming on for both my cards aswell now which isn't as easy as it sounds.

Boardy

 

[maverik-sg1]

It's a very strange one on mine indeed. Both cards are running perfectly fine I can get 760 / 910 on them both which is giving me about 43.5k in '01 on single card!! Put her in SLI and it's a different story, my stock SLI score dropped by 10k from having these lines going up an down the screen.

I don't think my cards are coldbugged as I was running last night on single card @ -17c so I was barely even pushing the core to what it's true potential can be and that was @ 1.65v.

I can see RMA coming on for both my cards aswell now which isn't as easy as it sounds.

Boardy

With respect - if both cards work individually, what are you going to RMA??

Sounds like a mobo (or Intel SLI driver) issue to me.

Mav