Graphics GTX 295 - BSOD at Startup - It's back

Just a quick question/rant for you today everyone. Remember that old problem there used to be with Nvidia drivers? You'd install your shiny new graphics card, install the drivers for it and when the computer asked to restart Windows, at the next startup (duting the "Windows welcome" screen) the machine would BSOD with a screen of text that disappears too fast for you to see what it says? That's just happened me all of a sudden, on one of my AM2 test beds whilst running an old GTX 295 through it's paces. Here are the specs:

MSI K9A2 Platinum
AMD Athlon 64 X2 6400
2GB OCZ Reaper HPC 1066
Antec TruePower Quattro 1000W
Windows XP Professional SP3

And the card I was testing, an XFX GTX 295 dual PCB model with the latest 260.99 Nvidia drivers.

The problem started when I realized that the graphics driver had crashed on that machine - the two folding clients that I had been running on it to long-term stability test the card were down and I saw that the video card driver had crashed, so I hard reset the machine and that was when the first BSOD occurred. The next thing I tried was starting in safe mode (without video card drivers) and that worked fine. So I put a different card in there, uninstalled the GTX 295 drivers and re-installed the 295 plus drivers. Restart required - upon restarting, the machin BSODs again.

That's as far as I've gotten with it today. This card processed and submitted over 100 WUs before this problem started, so I know that it's not faulty. It's a driver issue because the second that it's about to load the sky blue XP desktop "Welcome" screen it BSODS. With no driver installed (i.e. running on generic WinXP display drivers) the card is fine.

Any suggestions on how to get around this would be great. I've also tried it in a Windows 7 machine this morning with more modern motherboard, but unfortunately the PSU in that machine doesn't have enough juice to power the card, so it installed all right, but was unstable under load.

Thanks in advance. I thought Nvidia had gotten rid of this problem (which was actually more common with Windows 7 machines than Windows XP) long ago I'm off to try a complete uninstall -> driver cleaner -> reinstall again.

 

Driver sweeper and reinstall didn't work.

Any other ideas?

 

Does Physx install as part of the VGA driver on the new builds?

If so may have seperate driver to be cleansed?

Alex.

 

you can see what's causing it in the event viewer

http://www.windows-help-central.com/windows-xp-event-log.html

probably not related to your video driver if you sweeped it (in safe mode)

 

Just Googled. Cant believe how many people have or have had this issue, no one seems to post a frizzin answer, but some suggestions have been PCIE Clock Spectrum to Disable in BIOS and amending the registry settings for the driver, which as thehippoz said Sweeper should have removed clean.

Will keep looking on tinterweb while im not doing anything constructive.

Alex.

 

Also possible cause is Realtek network drivers aparantly... random.

Alex.

 

But but but.... it was working fine for about 3 or 4 days?! Gaaah. Driver/software conflicts annoy me so much Thanks for all the comments and suggestions so far guys, I'll have a look at a couple of those options tomorrow and see if I can get to the bottom of it. Bottom line, I want the answer to be "it's not working on this machine because -insert stupid software related conflict here-, but there's nothing wrong with the card itself" so I'll probably have to plug it into another test bed tomorrow and make sure that is the case.

I will post an update tomorrow.

 

When does the BSOD occur? If it's happening before Windows actually starts then it wouldn't be graphics drivers - they load with Windows.

 

XP Vista and 7 only load Windows, they don't have a DOS shell, just a console. The drivers load up immediately, nvdisp.vxd or something like that. Start in safemode and you can see them all.

 

Do you have a spare HD to try a real clean install of windows on? Try it from a bare metal restore.

IE Nuke the site from orbit, it's the only way to be sure.

 

Awesome.

Alex.

 

Thanks again for the continued suggestions and discussion - unfortunately because I was busy with the ice hockey and we a few problems at the arena, I had to leave early on Saturday and didn't have time to look at this. Hopefully I'll get a look at it tomorrow, probably in both a new rig with a newer board and also on a fresh install of Windows.

In other news, my gaming rig has started switching off randomly or not so randomly as it's happening when the graphics cards (2x Asus GTX 260 TOP Editions) get anything more intensive than displaying the desktop to do for a few minutes

Not a crash, not a restart, just switching off, then starting up again. Almost as if something is overheating and it's shutting power off to protect itself, but the temps are fine under load... It looks as if it's going to be ok then suddenly off she goes. Loud sigh

 

Further testing & outcome

I thought I would update this thread with my findings so far regarding the error I was getting on this card:

I received this card second hand, and there was always the possibility that it had been dismantled and possibly liquid cooled before, so I dismantled it completely and performed a thorough inspection. I found what appeared to be insufficient contact between the small copper cooling plate on the heatsink and the NF200 SLI chip, caused by use of the wrong thickness of thermal material. Again, this was only a preliminary asessment so I completed my disassembly and inspection of the card - finding no other major shortcomings other than the use of non-standard GTX 295 screws and slightly inferior thermal paste - then cleaned the whole card using isopropyl alcohol and reassembled it using AS5 for contact between the two G200 chips and the heatsink, also replacing the thin thermal material on the NF200 chip with the proper thickness thermal pad.

This time, just for good measure, I started the system in safe mode and uninstalled and cleaned all the display drivers. I then booted into Windows XP normally, downloaded a fresh copy of the (larger than normal GTX 2xx series cards) driver install package and installed it. Upon the required restart, exactly the same thing happened again as before - the machine throws a BSOD whilst on the Windows XP loading screen. Being utterly disappointed at it continuing to happen and at this stage thinking the card was actually faulty, I booted XP back into safe mode and in device manager, disabled the second of the two "Nvidia Geforce GTX295 Video Card" entires on the hardware list. This time, with only one of the boards active, the machine booted into Windows first time. I ran some benchmarks on the single core and it behaved exactly like it should - producing roughly half the points in every benchmark compared to a fully functional GTX 295.

I tried re-enabling the second chip on the card from Windows in the device manager but it produced the "This device cannot start" error a couple of times and even crashed (reboot, no BSOD) when I tried it once.

So my assumption with this card is now that it's faulty and I will be contacting XFX this week to discuss a possible RMA/replacement. I just wanted to ask one more time if, given this new information, anyone had any different ideas or can see something that I have overlooked before I try to send it back. I got some good suggestions before and am hoping that some more will be made because at this point I have exhausted all my own techniques and ideas to try and bring this card back to life.

Thanks for reading.

 

Welcome to the "I have a faulty GTX295 adorning my shelf and I don't have the heart to throw it away" club

 

Aww, I thought when I saw a reply to this thread that you were going to give me some suggestions on how to fix it!

So, this seems to leave only only question to be answered - What is XFX's warranty on 295's?

 

the only time my 295's bsod is when on fallout New Vegas or FO3 ..
I put that down to Bethesda being lazy shits and not writing the software properly before selling it (mini rant over)
I used to get it loads though with SLi'd 8800GTX's ..hope you get it resolved

 

I think it's 2 years, but please check that. Mine has expired anyway... both card, and warranty.

 

Thanks Pookey - I've emailed XFX to ask for two things - information on their warranty periods, and a full set of screws for another XFX 295 that I have which is out of warranty and still working, but on a liquid cooling block at the moment and was supplied to me with no screws to put the air block back on.

What happened to your 295? Did it just simply die?