Here is my problem. I have a Sappire Ultimate X1950 Pro, which has a Zalman VF900CU cooler. It seems to be overheating, as I am getting 39-40 C temps idle and around 60 at load, with my room temp being 23. After some time on intensive games, the screen might become all screwed up and then I have to restart my pc because this stays on my desktop. Overclocking makes things worse obviously.
Sounds like the cooler is not seated on the card properly.
Take it off, clean everything with denatured alcohol, put a tiny dab of thermal compound on the GPU and smear it around with a piece of plastic over your finger. Then remount the cooler making sure that it sits flat on the GPU. Slowly tighten each of the screws a little bit until the cooler is snug to the GPU. It does not have to be mounted so tightly that it warps the card.
60 C. is not hot, but if you get artifacts then it's overheating. Also check the airflow in your case.
60 is far from overheating. At stock clocks that thing could reach up to around 100C and still be stable. I have a 1950XT which is basically the same arch and at stock clocks it can go up to 112C before signs of artifacts and shutsdown at 118C. Your card is defective if it is artifacting or you do not have ram heatsinks installed and your ram is overheating.
Note: My card does not reach those temps it maxes at 85C overclocked by 75Mhz and 100Mhz on ram (200mhz effective). I put a piece of tape on the vent to test it's limits.
GPUs can take much more heat than CPUs and 60 deg load is not bad. I have the same cooler on an x1800xl, and at load it is at 50 deg, with the ambient temp being adround 17deg (thermostat on AC is SCREWED). That being said, if the GPU is running at 60deg you will notice the Heatsink being rather warm to the touch. Where are you reading the temps from?
Another thing to think about is does the ram have the heatsinks attached? When i purchased the cooler, it came with 8 ram heatsinks. Are all of them installed? One might have fallen off, or did it even come with these heatsinks installed? Ram could be overheating.
Hope this helps
EDIT: Just looked at a picture on newegg and it does have the ramsinks installed. Make sure they are all still there.
My X1900XT at stock speed and stock fan idles at 52C (126F) and under load it goes up to around 80C (176F). The temperature in the room is 28C (82F).
My conclusion is your Zalman is seated properly and your card is not overheating at all.
I recently had a very similar problem you are describing. I was afraid the VRAM was going bad. But before I started to really panic, I simply took the card out and unplugged the power cable and wiped off the minimal dust on the card. Then I reinstalled the X1900XT and all was fine.
1. my cooler is definitely well seeted, I've tried detaching it a few times an trie d screwing the thumpscrews tight, medium, or lightly, and it didnt change anything. I still have all ramsinks on.
So I seem to not be having too bad temps, and my card should'nt be crashing at around 60 is basicly what I've gathered.
My card is still under warranty so thats an option.
2 things to consider if your' all gonna help me trough this
1. My games dont "artifact" per say, not little white or yellow dots appear all over the place. Never has once, even while overclocking. It only just crashs, with the entire screen getting filled with permenant and static dots and lines and boxes and deformations of the image and at that point, either I can esc. the game rapidly, and I have some residue of dots on my desktop. If I dont escape rapidly, my screen usually turns black or the game freezes with alt+tab not working.
2. Even when my temps are 60, my heatsink isint hot. Is this something normal with this zalman heatsink or not?
Is my "static screen messup" instead of the typical artifacting normal for a x1950 pro?
If you guys can answer all or some of these questions, that would be awesome
Thanks again, great help
PS for Jaguar, I've done the unplug replug ( with different pci-e power connector too) thing about ten times, worked the first one, prob came back shortly after, hasent worked since. weird[/img]
It's sort of like that there is always about 10 times more artifacts and they're usually bigger. I've been using ati tool for a long time and never noticed voltage changes, would you mind telling me where I can change that?
I currently have version 0.27b1
Thats about as severe I could get it because anymore I wasn't able to take a ss. I use version .26. You go to settings then click on the bar up top and go to voltage control. Im not sure all cards are supported for changing the voltages. If not it needs to be RMA'd.
I noticed that the card in the pic reached over 80C before the "slight" artifacting, while mine does heavy artifacting when it gets over 60C. Does anyone at all think that silver thermal compound (once again because my heatsink is never hot) would make a difference or everyone definitely agrees on RMA?
Are you referencing the pic I posted? Because it's only at 60 but what I did is turned the memory voltages way down then ran a artifact scan. The screen went like that in about 2 seconds then I got a ss of it. At normal voltages it doesn't artifact at those clocks till about 104C. You need to RMA it. Arctic silver will not make a difference. And your core is not hot at all so it doesn't need to be cooler.
Seems like the damage is already done. If you say that your heatsink is cool to the touch, then it only meens one thing. Now you know why I do not overclock, its to risky and for what 10-15 more frames. Your Vram seems to be fried. Just cross your fingers that you can pass an rma.
This happend to a 7600GS of mine and in the end I found out it was because the ram on the card took on too much static. At least that's what I could figure. Whenever I did some 2D stuff it was fine and even in some 3D stuff but then once I got to more demanding 3D stuff it went all nutty on me. I'd restart and it'd be fine 'till about 3 minutes into a demanding game. My temps were super low for a GPU too, 33 LOAD. Open windows in winter helps.
I own two identical eVGA 7600GT CO's with identical bios and I find that, by switching them in their slots, one will read ~46C idle, but the other one will read 50C idle at the same time of year.
It really makes you wonder about the manufacture process and variation in components as they roll off the production line...
Hey, one has purple capacitors and the other has silver...