WARNING ATI users: Catalyst 3.8 can kill your Card / Monitor

I dont know if this has been posted yet, but this is pretty important to ATI owner .
Original Post http://short-media.com/forum/showthread.php?s=&threadid=4820

WARNING: ATI R300/350/360 USERS SHOULD READ

I've got some distrubing news that, if left unchecked on your system, will severely damage your video card and monitor systems to the point of RMA.

Cards Affected:
All Radeon 9800 XT's
All Radeon 9800 Pro's
All Radeon 9800 Non-Pro's
All Radeon 9700 Pro's soft-modded to 9800 series.
All Radeon 9700 Non-Pro's soft-modded to 9800 series.
All Radeon 9500's soft modded to 9800 series.

Note: Radeon 9600, 9200, 9100, 9000, 8500 or previous generations are NOT affected by this serious issue.

Software Causing Hardware Failure:
ATI's Catalyst 3.8 Driver Series
OMEGA Catalyst 3.8 Driver Series

1) While we're not quite sure what part of the Catalyst 3.8 drivers are causing this, some part of the driver (whether it's the new VPU Recovery feature, HyperZ Enhancements or the ATI Overclocking software) is causing the core on BBA and OEM Radeon cards that are identified as "RADEON 9800 XT", "RADEON 9800 PRO" and "RADEON 9800" in Windows to have their GPU's overheat by as much as 15*C over normal during 3D graphics sessions. This has caused a number of fans on the Radeon 9800 XT series to fail, allowing the heat to completely destroy the card.

This news is especially important for individuals who performed the 9800 to 9800 Pro BIOS flash and did not include additional cooling on your card, as the GPU is already overclocked and producing more heat than normal.

As a safety precaution, I am strongly encouraging that anyone with any of the affected cards above roll their drivers back to the Omega Catalyst 3.7 drivers or the ATI Catalyst 3.7 drivers, as they are verified not to cause this overheating issue.

I did some testing with my Radeon 9800 NP BIOS-flashed to a 9800 Pro. The results are not encouraging.

Running looping 3DMark2003 for 30 minutes and utilizing a temperature probe on my GPU (regular cooling), the following temperatures were recorded:

Catalyst 3.8: 135 Fahrenheit
Catalyst 3.7: 117 Fahrenheit

The heatsink on my Sapphire Atlantis 9800 was so hot during the Catalyst 3.8 test that I burnt my hand when I accidently touched it.

2) The Catalyst 3.8's seem to have a bug in some systems that allows the driver to process the SECONDARY adapter settings when a game is started and tries to switch to the resolution and refresh rate specified in the program.

Normally, the card is limited by the INF file settings for the PRIMARY adapter, so that no resolution or refresh rate is attempted that is beyond the capability of the monitor hardware. But in the 3.8's, it seems to be processing the SECONDARY adapter settings first. In a system without a monitor plugged into the second adapter, there is no INF file limiting refresh rates and resolutions, so, the system is trying to force refresh rates and resolutions beyond what the monitor on the PRIMARY adapter is capable of.

So instead of having the monitor "flicker" once when it switches to the desired resolution/refresh, monitors are "flickering" between 4 and 8 times, with resolutions and refresh rates that are beyond the capability of the actual monitor.

As a result, some monitors have been damaged, while others have blinked out and recovered. Some users monitors are under warranty and can be RMA'd, but for those who don't have a monitor under warranty, their monitor is damaged and they must buy a new one.

Again, as a safety precaution, I am strongly encouraging that anyone with any of the affected cards above roll their drivers back to the Omega Catalyst 3.7 drivers or the ATI Catalyst 3.7 drivers, as they are verified not to cause this monitor refresh-related problem.

My god, you've just scared the s*** out of me, ive a freakin P4 cooler on my 9800pro, overclocked to the max with 3.8 installed.
I've noticed recently that when windows starts up, instead of it running at my set 1280x1024 @85hz, the monitor is running at something behond 1600x1200 @60hz. I normally fixed it by just switching to 85hz manually. The 3.8 must be the problem. The monitor also flickers sometimes for a split second.
I know what im doing the second i burst in the front door this evening...

I've had the same problem with my monitor flickering off for a second and then back on randomly during 3d apps. I'm rolling back to the 3.7's first thing. I hope the 3.9's fix this...

Can anyone explain how a driver can cause the gpu temps to increase substantially? I thought the only variables which had any real effect on temps are the clock speed and voltage.

hmm nothing official from ATI and there are no links from that site to where the original alert came from.

I read a theory that its the shader effects. something about even when you have it set to noone, the gpu part that does the shader effects is still active, adding heat. This did not come from ati though, but it is the only new thing besides vpu recovery that effects 9700 cards and above

I just installed the Omega 3.8s on my radeon 8500 and I got this in counterstrike

If you look there are a couple of green checker boards on the wall. They only seemed to show up when you shot the wall. they seem to have gone now after I turned my clock speed down but that speed was fine with omega 3.7s.

hmm i havent noticed any problems with my stuff with the 3.8's.

although i do have w/c for my gpu

I just got an AIW 9800 Pro refurb from newegg a few days ago. I installed the cat 3.8's and it ran good for a few days. My monitor did flicker a few times though. The big blue "Moitor is Shutting Off" would pop up, like it wasnt getting a video signal, then it would come back on. Well I got home from work the otherday and monitor was on, everything was huge and blocky, there was a Hardware Failure msg box, and the ATI recovery box. The card wouldn't work for crap after that. I removed the drivers and pulled it to have a closer look at the card. A few of the traces on the part that goes into the AGP slot appeared scorched. I put my 9500np back in, installed the 3.8's and went to newegg for a RMA#. It's going out in the mail today. I thought I probably just got a card that somebody had maybe tried to overclock or something, but thought it was strange that it worked perfect for a few days then went crazy. After reading this post, it's possible that the drivers fried a perfectly good card. I'm gonna keep a close eye on this thread and hopefully things will be resolved by the time my replacement card gets here. If not I'm gonna go for the older drivers just in case. BTW, my 9500np seems to be fine with new drivers, no monitor flickering or anything.

peace.
unloaded

... Good thing I haven't installed them yet My GPU gets quite hot with it's OC, so I REALLY don't want to try it out now!

JigPu

yea im glad i waited to see how they were...has ati said anything yet

I think something as severe as this would have been noticed in testing. They test them a lot longer then they have been out...so its kinda odd that it would show up so soon when they are put out, but not when they were being tested.

I think the info is correct! I just had installed my Radeon 9800 Se@Pro AIW and used the Catalyst 3.8 driver. With the Cat 3.8 my CPU temperature under load while running Prime95, was 47°C.
Then I deinstalled the 3.8 Catalyst drivers and installed the Catalyst 3.7 drivers. My load temp went down to 44°C. It was the same room temperature and nothing was changed except he drivers.
That means, that the graphics card was emitting a lot of heat, although no 3D application was running, when the Catalyst 3.8 drivers were installed. This heat caused the CPU temp to rise. Now with the Cat 3.7, the heat problem seems to be gone!

wow............ and i was about to install 3.8...

sticking with 3.7 now!

This is wicked strange. I can't imagine a driver release that can burn up a card and fry a monitor. If this is true ATI could have some major problems on their hands very very soon.

i don't mind guini pigging a bit, and i have a temp probe laying here. I have been using the cat 3.8's since their release (ati's the day they came out, and omega's version of the leaked 3.8's) since then. I will spend the next while doing some testing with 3dmark and the temp probe, then revert to the 3.7's and do some more. i'll post back with the results.


J.

My brand new 9700pro is using 3.8s

I guess its time to roll em back.

Also glad i didnt switch, i really didnt have any problems with the 3.7's to change to another driver anyway.

Hmm. I don't know what to think. running 3dmark03 3 loops i saw the temps stabalize about 56 with the 3.8's and 52 with the 3.7's. there is a difference there but it hardly seems enough to kill a card to me. I have not experienced any of the monitor flickering, when it changes res for a game or if i change it manually it's one click.

I am not convinced, has anyone seen mention of this elsewhere, or anything documented? I will leave the 3.7's for a while, sort of being safe, but more because i'm to lazy to reinstall the 3.8's tonight. In the course of using them, i liked them very much. IQ in all my games was greatly improved, as were FPS when running with AA and AF cranked up.

Well, time will tell. Perhaps it's specific to certain setups or something, because the 4 degrees i saw in difference is a far cry from 20.


J.

i remember the drivers on the 5800 gpu that would adjust the fan speed as necesary, it had a problem that if you ran a 3d screen saver it would get hot enough to kill the card, dont remember the exact resons behind it, i just remember that it wouldnt adjust the fan speed when the screen saver was running.

I haven't noticed any of those symptoms on my 9800np (pro bios flash), but I also have a cooper cpu heatsink and fan mounted on the gpu. I can't monitor the card temps, but my case temp hasn't risen.

**edit** ran 3dmark03 and now I'm getting artifacts with my normal OC which worked flawlessly with the cat3.7. Definitely something to this.