Seemingly unsolvable 8800GT problem...

Hey everyone. I'm hoping someone can either help me with this problem, or at least point me in the right direction to get help. I would really appreciate any advice as I'm about to kill myself and post it on youtube.

Here's the problem: I recently upgraded from an x800 Pro to a PNY 8800GT. It worked well for about a week. Obviously it made an amazing difference that I didn't think was even possible. Then, I upgraded the drivers on the card to the latest ones in a greedy attempt to squeeze a few more FPS out of a specific game. I uninstalled the old drivers, and used DCP. Immediately after doing this, the card started flipping out. Most often when launching or playing games or other apps, my system would either:

A) Completely freeze, requiring a reboot
B) BSOD with a message about a fault with nv4_disp.dll
C) Turn a small notch of the top left corner of my screen into what looked like a tiny 4 bit rectangle of Oregon Trail 2-type colors. This would be accompanied by my system grinding to a near hault, responded and refreshing once every 5-7 seconds or so before completely freezing. Then it may or may not turn into the BSOD mentioned above.

I've since learned that the nv4 error message is generic and in many cases is merely a symptom rather than the cause.

So here's what I've done so far: Obviously, I immediately uninstalled the newest drivers, and reverted to older ones. (again, I used DCP) The problem persisted. I then tried a third set of official drivers, followed by two sets of beta drivers from this site, and even the drivers off the CD that came with the card which originally worked fine. I uninstalled and cleaned all old drivers before each new install. Nothing worked. I flashed my bios to the most current version offered by ASRock.

The problem would seem to fluctuate between the three listed problems above, but by far the worst and most prevelant is problem C). I updated my bios, and spent a lot of time troubleshooting the problem by turning every relevant bios setting on and off to see if it still happened. Nothing I did made a difference. So, next I put my old x800 back in and that ran fine for about 10-12 hours. Next, I exchanged the card for a different one hoping it was just a bad card...no luck there either.

As i'm writing this, I just reformatted my drive, and am now in the process of reinstalling everything (always a fun night) and, yep...just got problem C) again.

My system:

AMD Athlon 64 X2 4800+ Toledo
ASRock 939Dual-SATA2 939 ULi M1695
Antec TRUEPOWERII TPII-550 550W ATX12V Power Supply (I believe this power supply has at least 38 amps between the 12 volt rails)
Western Digital Raptor WD740ADFD 74GB 10000 RPM SATA 1.5Gb/s Hard Drive
CORSAIR XMS 2GB (2 x 1GB) 184-Pin DDR SDRAM DDR 400 (PC 3200) Dual Channel
Windows XP Home SP2
PNY 8800GT 512MB
Creative Audigy

Before my reformat I downloaded all the common diagnostic tools to monitor the wattage, and heat of my system. While my CPU runs a little warm, around 57 degrees idle, the card rarely if EVER gets above 60. Speaking of CPU, I tried adjusting the voltage and frequency of the chip down to produce a temp of around 51 degrees and that had no effect.

The problem is hard to reproduce at will, but it will almost assuredly happen once an hour. It is much much more likely to happen while I'm doing things, especially launching games and apps, and seems much less likely to happen once something's already up and running.

At this point, the only thing I can think of is that this card is just completely incompatible with the rest of my system, but I find that pretty hard to believe as I haven't found anyone having my exact problem with my exact system (or even relatively close to the same).

I'm really all out of options here, and having already returned this card once, I'm not sure if I'd get away with returning it again. If that was an option, I'd probably just get another ATI card. I'm really loving this 8800GT though (when it works) and i'd like to get it working. More than that though, i'd love to get my system back. If anyone can offer any help with this, I'd really appreciate it. Thanks for your time.

edit: since I've reformatted, the several times the problem has come up, it's been a little different. Now I get a progressive audio loop along with a flicker to my screen which blinks once every 5 seconds. I can move my mouse around and it will update to a new location after every 5 seconds, so it's not exactly a freeze. Also, the audio, while looping, will also jump ahead every 5 seconds to a new looping segment. This problem sometimes has the tiny 4 bit rectangle from problem C, but doesn't always. There is always some smearing on the screen which is static.

 

Is that 57° Celcius for your idle CPU?
If it is, you should make sure it isnt really just the CPU that causes your problems.

Do a small torture test with Prime95 and see what that gives.
Maybe the extra heat from your new videocard is just enough to make your CPU fail.

If that doesnt result in a crash, do some memtest32.


I hope you find the source of your headaches...

 

I'd try another PSU if I was you. Even the finest quality hardware can fail sometimes if you're simply out of luck. I'm not familiar with ULi chipsets but do they get hot like nforce4 SLI? Like Belz said overheating cpu/chipset can cause nasty freeze and shutdowns. Just make sure they get enough cooling.

 

I couldn't easily get my hands on another PSU at the moment. I agree though that I'm really just down to that or the motherboard being incompatible.

The temps according to speedfan in my system are all within acceptable limits with the exception of my CPU which runs between 54-57 depending on whether it's hot in my room or regular room temp.

As Belz suggested, I ran Prime95 for about 10 minutes without getting an error. I realize that I should probably run the test longer, but being as my system usually freezes after about 20 minutes or so anyway, I don't think there's any point to continue the test. My memory checks out too.

Also, my specific problem has seemed to more or less bottomed out into a single type of occurance. The past 10-15 crashes have been just traditional lock up/freezes. No response to any input and a frozen screen. I haven't seen the flicker/loop crash in quite some time.

I guess all signs are pointing to the PSU. I just don't buy it though. It FEELS more like a motherboard or bios thing. I'm really just trying to track the thing down for sure because I'm on a tight budget right now, and I can't afford to replace everything.

Anyways, thanks for your responses.

edit: btw, the main reason I don't believe it's the PSU yet, is because my x800 STILL works in my system.

 

Last-ditch-effort bump.

 

What are your CPU temps while Primming (under load)? If it is hitting 57*c while idle, it could easily be hitting 70*c + under load. As mentioned above the 8800gt could be putting out just enough heat to raise it above that of the stable x800.

Let Prime run as long as you can while keeping an eye on your temp. Run the Western Digital Diagnostics tool on your HD and test the RAM with Memtest. Some how some way find another PSU to test and another system to try your card in, clear your CMOS back to default. If after all that it still takes a dump on you; return the card again.

 

Thanks for the reply. I'll try all of that when I get home.

 

Well I would lean mostly towards a faulty card. Do you get any errors with ATItools test. I am leaning toward some sort of issue that just manifested itself at the time you changed drivers. Personally I would RMA the card or try it in a different machine.

I was leaning toward a driver issue until you said you formatted and reinstalled. One question what driver did you use when you reinstalled?

BTW yes your temps are on the high side but they could be just fine. IMO with a stock CPU cooler and that chip that is pretty normal. Also your sensor could be off. I am not familar with ASROCK MB's but PNY cards are pretty top noch, I get very few returns on them.
The 550w PSU you have is enough for your system (your around 350-400w peak I would think). It is an Antec so good quality PSU there.
If you have another free PCI slot move your Audigy and see if that makes a difference, I have had that issue before.
Another thing is turn your head slightly left and place your left hand on your right ear and click install on the driver install, hell anything is possible

Sorry for the long winded and no help post


Edit: Oh also try reseating the card remove it and reinstall it.

 

Haha, thanks for the reply. I appreciate the thought you put into it. As for the card issue, I kinda already eliminated that since I've already returned the card to the store and gotten an identical (stupid, I know, but it was the only 8800gt they had) PNY card. They both give me NEARLY identical problems. Only difference seems to be in the area of severity or length of time of the crashbug.

I'm using the 175.16 drivers right now. Those are the ones that I installed when everything went to hell. But when I reformatted I used the ones on the disk, which were 169.xx I believe.

I'll definitely move my Audigy when I get home. Like you, I believe in the PSU and still don't think it's the culprit, but I'm going to borrow a multimeter tomorrow and test it myself since I can't get another PSU to put in there without buying one.

Thanks again for replying. I really do appreciate any help I get.

 

I was also thinking about an IRQ conflict with your Audigy, but forgot to mention it.

But dont just move it, just remove it completely.
Also, remove 1 of your 2 sticks of RAM.
I always try to remove most components when I'm looking for a weird issue like this.

Use ATItools to heat up your videocard, and if its not causing a crash, throw Prime95 in the mix.

For me its one of these 4 problems:
1) cpu too hot
2) unstable memory
3) irq conflict with Audigy
4) ASROCK mobo really doesnt like your card and needs a BIOS update


EDIT:
One thing for sure, you're not the only one with problems with this mobo/videocard combination: http://forums.pcper.com/showthread.php?t=452845
Try enabling the "AGP P2P Deep Fifo" setting in your BIOS as suggested in that forum.

 

I have read about driver cleaner getting a little too cleaning active on the nvidia chipset drivers when cleaning with the nvidia filter
maybe reinstall the chipset driver package

 

I would like to see some CPU temps while idle and while under load? 57C sounds kinda high for idle. This could tell a lot.

 

I have yet to mess with the Audigy yet, that's my next step. What I've done so far, per suggestions from you folks, is run a Prime95 torture test while monitoring the temp. I used speedfan for that. Apparently my CPU idles around 57C idle during the day when it's around 90F+ here. At night, it's around 53-54C. I ran the Prime95 just now (night) for over 20 minutes and encountered 0 errors/failures. Temp peaked at 78C at full load and didn't rise above that.

Without using something like nTune or Rivatuner, my cards temp is usally around 55-58C idle and no higher than 69C under the most load that I can throw at it not being able to install any serious tests. (I've been installing older and smaller games to try and test things since my comp invariably crashes in the middle of the long installs of bigger, modern, stressful games)

With a third party app boosting my gpu fan speed to even 50% from the default 29% the gpu idles at around 47C at night.

But the card seems stable up to 69C as far as I can tell. Since I don't have much stuff reinstalled yet, 90% of my freezing is on the desktop when neither my cpu nor gpu are being particularly taxed.

Next I checked my HD with the WD diag tool. Everything fine there as well. I actually came across that same forum you linked, and thanks for looking. It makes me think more and more that it's a mobo problem.

I'll definitely troubleshoot the ram and the Audigy next. Thanks again for taking the time to help. I appreciate it.

 

You know I was thinking the same thing the whole time leading up to my reformat. Not necessarily about the chipset, but I was thinking more of a system file getting nuked. Once I reformatted I removed that as a possibility. That's a good idea though, I wouldn't have even thought it was possible for DCP to harm a critical chipset driver. I guess you never know, right? I just downloading the ULi drivers from nVidia's site. I'll put that on the list of stuff to check out next. Thanks a lot!

 

Yeah, as I alluded to in my other post, I may have overstated the idle temp a little. Or at least misled you all about it 57C was the highest I saw it at while idle through speedfan. Granted, it was probably at least 80F in my room at the time. After rebooting and going straight to my bios the temp had already dropped to 55C so I couldn't really swear that the speedfan temp was accurate.

That being said...53-54C seems to be the idle temp at night like now, when it's about 67F in my room. Prime95 got it up to 78C where it stayed at full load for more than 20 minutes without incident.

I'm using a Thermaltake 80mm fan with Arctic thermal paste. That fan and thermal compound were the first I ever installed, so while I believe I did it right, I'd be willing to accept that I might have screwed it up somehow, leading to improper cooling.

Either way though, if I can't reproduct the freezing I'm getting at 57C with a temp of 78C for 20 minutes, it seems to me that the CPU overheating probably isn't the issue. I would keep voltage irregularity a possibility though.

I dunno, someone please correct me if I'm wrong to make that assumption, I'm still learning this stuff. Thanks for your reply, Belz.

 

It's probably not causing your problem but I'm pretty sure those temps are very high. I don't think you should ever get above 70, if someone could verify this either way that'd be great.

 

Id at least take the CPU heatsink off and clean it out, im sure its probably covered in dust. Also re apply some thermal paste on the CPU which should bring the temps down a little. As the others have suggested have you tried flashing the motherboard bios? It could be that the board has a problem with that card that needs a newer bios for it to be seen correctly. Just ideas.

 

Well according to this site, it should be at 65c max.

http://www.pantherproducts.co.uk/Articles/CPU/CPU Temperatures.shtml

 

Yeah. I've always thought it was a little on the hot side, but since I never encountered problems I kinda just forgot about it. You're right though, It won't hurt to take a look at that. I probably just botched the paste job since it was my first time.

I'll do it tonight when I test my PSU with a borrowed multimeter. Oh and btw, yeah I upgraded my bios to the most recent version offered by ASRock. I downloaded the most recent beta drivers from OCWB too, but I'm considering that as a last resort before I destroy my already delicate short-term financial future by going out and getting a new mobo/chip.

Thanks for the replies.

 

One thing we must always keep in mind when measuring temps on the CPU. It could be off, even the bios could be reading the temp wrong or the sensor is just not correct. Personally I would try a temp probe before trusting a temp sensor. I have seen temps on my old MSI Neo Plat Nforce 2 board that where 10-15c too high (and it was not a set number the BIOS was screwy) after a bios update it was always 1-3c of actual measured temp on the Heatsink.