Display driver crash when O/C HD 7970

As far as I can see the card is week. I've had this on some cards were they don't wanna overclock, while others broke records...like my 4850 on air http://hwbot.org/submission/986040_videl_3dmark06_radeo...

...and that was with a phenom II 550 unlocked into a quad b50. If I had my 9080x I would have had higher points and the record would be higher up on that list lol...

I'm guessing the chip here...or on one of the chips is a bit low binned and therefore the transistors have some voltage bleeding issues, although the temps (59C) don't really say so...

Usually when you have voltage bleeding your card needs more voltage to run at certain clocks...but at the same time the temps go up which introduces more voltage bleeding...

And that means the card will have a certain limit to overclocking that cards that have no issues with voltage bleeding wouldn't have.

Basically I think this is one of the unfortunate cards and the gpus here overclock differently and need different voltage settings to run at the same speeds so when you push them to 1250 one of them starts dropping out of sync or has to many calculation errors and so the chips don't work well together and you get a crash of the driver rather than artifacting.

Basically I think this particular card has issues where others might be luckier and built better...and have no issues going over 1250...If you could trade it or something that could be an easy solution for you.

You could also try pushing the voltage higher (more than 20%) if possible on that card...

Im not sure what you mean by the GPU's are dropping out of sync? Its a single GPU card.

I have been reading of people flashing their refrence 7970 to the Ghz Edition and seeing lower power draw for the same performance. Tempted to try that and see if that lets the stock voltages have enough headroom to overclock better.

I know in MSI afterburner you can manually set th voltage, but I cringe at doing that without having a good bead on the limitations of the card, last thing I want to do is fry a $450 card. Besides the maual setting on afterburner I dont know how I would go about increasing the voltage past %20.

The ASIC quality rating on the card (from GPU-Z) is 67%. So not a very high rating... Really is kinda crap if your OC potential is a total crapshoot.

The same thing happens with my 6870 when I was playing with over clocking it. I wouldn't see any ill effects when approaching the cards limit, It would just lock up, blank screen, then everything comes back to normal and the system would say the display driver stopped responding and recovered or something along those lines. I wasn't to worried about it cause over locking the 6870 wouldnt result in a worthwhile performance boost. I ended up running 950 core 1150 mem with the card undervolted @1.1V. I figured I would rather keep the temps and power consumption a little lower rather than trying to squeeze out a measly %5 performance increase.

So does this mean I just have a crappy overclocker? Or is this a typical result of overclocking?

This is extremely characteristic of the 6000 and 7000 series cards. They don't artefact much, they go straight to crashes. 1200 is extremely good for that card, many won't go above 1100

So am I likly at the limit of my card? Or would edging up the voltage a bit perhaps get me a bit more oomph? I have more room on temps, so I would like to try and push it a little more if I can without the risk or frying the card.

Best answer selected by kendrose.

Perhaps I shall just be happy with where I am at. Wait to see the new Ghz editions bios spread among the other hardcore overclockers see if it does indeed offer up a bit more headroom before trying to push the card any higher.

Update, small tweek to the voltage with trixx, and im stable and under 60C at 1200/1700. Going to leave it there for now, though I did stay stable at 1250 and was still in the 60C range, but the VRM would spike to 105C under burn in tests (reguar benchmarks and games it stayed only a few degrees higher then my GPU). Any thoughts on how much punishment the VRMs can take?

VRMs can get hot but the lifespan decreases dramatically when the temperatures are too high, about a 10 fold decrease in lifespan over the functional spread. The exact threshold depends on the quality of the components in the VRM though.

So should I use the same restraints in regards to the VRM temps as I do with the GPU temps? Right now I consider 80C my ceiling, but that doesn't seem to be an issue since when gaming I don't see the temps break 62.

They're designed to take abuse. I wouldn't let them get too hot though, but 90+ should be fine for most of the time

Thanks. Sometimes its hard to find any info on these sorts of things.