HD4850 lock-ups causing me nightmares.

[thestjohn]

I was wondering if any of you guys could help me with a problem which is driving me a bit insane. I bought a Asus HD4850 512Mb with the Glaciator cooler from Scan during June for my brother as he was complaining of low frame rates in WOW. After a month or so he decided he was quitting the game and gave the card to me. "By the way," he said "it's been doing weird things like locking the computer up during heavy raiding." While it would have been nice to know earlier, I shrugged and installed it in my computer to test it.

The first thing that made me think "hmmm" was that the default memory clock on the card as reported by both CCC and GPU-z was slightly higher than advertised on Scan, but only by 10Mhz. I didn't think that was too much of a problem, temps were all within range, and the few games I briefly tried didn't seem to have any issue with it. I just left the card and clocks where it was and carried on.

Flash forward a week or two and I had the time to play some games properly, and I started getting some serious errors. For example, complete lock-ups during the same point in Crysis (fully patched) involving coloured vertical lines whilst the sound looped in the background which required a reboot, or a frequency out-of-range message on the monitor which allowed me to restart windows, albeit without seeing what I was doing.

I enabled the temp logging feature of GPU-z to see if there was a temperature problem, and with the fan on the card set to auto the highest core temp I saw before an error was 71C, and the highest VRM temp was 85C, which I thought was within range. However, I set fan speed to 100%, took the side of the case off and plonked a desk fan pointing into the case to see if it was an overheating problem. The error occurred in exactly the same place, yet with lower temps than before (63C gpu core, 75C VRMs). I can also replicate the same sort of errors in Empire:Total War (again, fully patched) and again, they always occur in the same place in the game.

I've uninstalled and reinstalled Windows XP, uninstalled and reinstalled Catalyst (although I've only used 9.7), even tried Windows 7, yet the only difference I've noticed is that Win 7 occasionally allows me to see that the VPU has done a recovery. I've reset my CPU to stock clocks, which made no difference. I tried clocking the card's memory at reference clocks, but almost immediately after applying the clock setting I experience a hard lock with coloured vertical lines again if I have my second monitor plugged in. With only one monitor I am able to set reference clock timings on the memory, but I still get the same errors.

I really want to send this card in on a RMA, but I want to eliminate the possibility that it's my configuration that's at fault. The only thing I can think of is that I haven't tried is a beefier power supply, but as I understand it mine should be sufficient (Seasonic SII-12 430w). I can get my hands on a Hiper Type-R 580w if needs be for testing purposes. I also haven't tried any previous Catalyst releases.

Apologies for the long post, but can anybody recommend what my next step would be, as I'm feeling slightly stupid right now!

[watercooled]

That PSU should be fine and since the problem has happened on two computers I'd say it's a faulty card. You could do more to test it but I doubt anything different would happen so you'd just spend hours messing about with installing it/drivers etc. Personally I'd just RMA it, assuming it's under warranty that is.

[thestjohn]

Oh it should still be under warranty. Haven't mucked about with the cooler at all, and it's only a month and a half old. Thanks for pointing out that more testing is probably futile, I think I needed someone else to tell me that though! Oh well, RMA it is.

[Jonessie]

I would RMA this as well, I did read an article somwere... not sure which site this was on where people were getting issues with ASUS GFX cards, not that I'm knocking them, I use ASUS motherboard and have done for 10 years now. but there was an issue which you explained and you could not put your finger on it what was causing the problem. Right just search on Google with Problems with ASUS HD 4850 and the list is endless.

[sammorris]

Errors like that imply faulty card, have it returned.

[shaithis]

Does't sound too much like the card to me TBH.

It may be some really weird problem with the hardware but lock-ups at the exact same points each time tend to be more driver/software related then hardware....especially when you have tried the old "take the side off and slap a huge fan on it"

[sammorris]

Possibly, but it could be a matter of a faulty memory block. I've yet to see artifacting as a result of driver issues, whereas I have seen plenty of hardware defects a fan has absolutely no effect on.