7900GT + Zalman VF900-CU = Problems!

[htmlmasterdave]

I just swapped out the standard cooler on my XFX 7900GT for a Zalman VF900-CU (with AS5) and ramsinks and I'm having a lot of stability issues suddenly. Anyone know how tight the cooler should be installed? It seems like the board is slightly bending because of the pressure. (could this be causing issues?) I also heard that if the board is dirty somehow, that it can then not perform the same. Getting a lot of these sudden black screens and corruption. Temperate at idle is around high 30's C and mid 40's under load.

Any suggestions welcome! (I haven't done any volt modding yet, but planning on the 1.4V one)

Thanks

 

[JPH1121]

What are your frequencies?

 

[htmlmasterdave]

I've even been having some issues at stock.... stock is 520/1500 and I was running at 560/1800 before. I'm starting to wonder a little bit if it's software related, like something got corrupted or recked on my windows install. It seems to generally work fine until I open up the properties window for the display. I remember even having problems selecting overclocking, but using stock speeds last night, yet if I played ut 2k4 just after booting... it seemed fine.... it's weird.

Do you think that the bending of the card a bit might be causing problems? Or that as the card heats up it bends more and maybe loses contact with the PCI-E bus? Thanks

 

[EffeX]

Its not supposed to bend man. Those thumbscrews are to hold the cooler on not strip the card. Definately loosen it.

 

[nullpointerus]

Originally posted by: htmlmasterdave
Temperate at idle is around high 30's C and mid 40's under load.

Your temperatures are not the source of your problems. From what I can tell, overheating should only be happening at temperatures greater than 70C.

 

[imported_gbenson]

I experienced exactly the same issue (minus the warped card, don't overtighten and everything should stay straight). With stock cooler I was running 48 - 55C, Zalman dropped temps to 41 - 42C but stability went horribly bad. Can only run at stock settings now. With stock cooler I had comfortably OC'd +38 Core + 60 Memory now it crashes out unless I revert to exactly stock settings. Seems stable at stock settings, but even +1 on the core causes it to go nuts if I actually exercise the GPU.

In looking at the stock cooler it is a very large flat copper unit, wondering if it makes contact with another chip somewhere that the Zalman misses, or fan connector sees no fan attached and feaks out into some 'default mode' or voltages get out of sorts. I used Artic Silver between the GPU and Zalman and was careful not to use too muchand have it spill out onto any leads, but wondering if perhaps I used too little and parts of the GPU are not making contact, although I would think I'd see an overall temp rise in this situation. Only other thing I can think of is the memory chips don't like the heatsinks.

With stock coller I just used coolbits and ran the 'detect' to get my first OC point. Running detect with the Zalman installed results in the graphics first going weird (random pixels changing to weird colors) then image disappears entirely, then computer rebooting itself.

I have not changed drivers or anything else, just the cooler on the card. WIll tinker this weekend with the amount of Artic Silver and perhaps reinstall the stock cooler and see if it resolves the issue.

 

[nullpointerus]

There are 7900GT's with manufacturing defects that show up typically in the first week or so of use. Is it possible that you are suffering from these issues? If so, you might want to try putting the stock cooler back on and RMA-ing the card.

 

[Pens1566]

Originally posted by: nullpointerus
There are 7900GT's with manufacturing defects that show up typically in the first week or so of use. Is it possible that you are suffering from these issues? If so, you might want to try putting the stock cooler back on and RMA-ing the card.

 

[lektrix]

Originally posted by: nullpointerus
There are 7900GT's with manufacturing defects that show up typically in the first week or so of use. Is it possible that you are suffering from these issues? If so, you might want to try putting the stock cooler back on and RMA-ing the card.

What the hell?

Do you have any more info on this?

I just bought a used 7900GT (used for 1.5months) and I dont want no BS problems!

 

[d3lt4]

"Check the SHIM vs CORE height. My XT has the core just above SHIM height, but my brothers XT has the CORE just even.

If any of you have a core BELOW the SHIM height, it will cause issues with the Zalman or Accelero, as they are FLAT and dont have the nump-out that the stock cooler has over the core. So bad heatsink/core contact can result.

Always check with a straight edge to make sure, if the core is too low, you will have to pop the shim off to let the sink work 100%."

This is something I found at XS, and it seems to be true from the stories I have heard, so i would try taking off the shim, and your problems will most likelly be fixed.

 

[Pens1566]

Originally posted by: lektrix

Originally posted by: nullpointerus
There are 7900GT's with manufacturing defects that show up typically in the first week or so of use. Is it possible that you are suffering from these issues? If so, you might want to try putting the stock cooler back on and RMA-ing the card.

What the hell?

Do you have any more info on this?

I just bought a used 7900GT (used for 1.5months) and I dont want no BS problems!

Linky to just one thread over @ guru3d. Also many on EVGA forums, xfx even shut down their support forums due to the massive response on this issue.

edit:

If you've already gone a month and a half w/o problems, you're card is probably ok. They tend to die after a few days to a week of usage. Depending on how hard you push them.