Installing *any* drivers for GTX 280 causes hard freeze on boot

[InfidelPimp]

My rig's a custom built machine with an eVGA nForce 780i SLI mobo and a single GTX 280 (1 GB DDR video RAM), running Vista Ultimate x64.

A couple days ago, I started to get hard freeze/crashes while playing Dragon Age - Pink Screen of Death, is the term, I believe. (Image from back in February when I last had to RMA my GTX 280.) I *was* able to usually recover from those after a few reboots, until today, when booting up Windows normally results in a hard freeze, with my monitor displaying a single, solid color (sometimes it's blue, like a BSOD, sometimes a pale green, and I even had a orange/yellow in there once... recently it's gone to a greyish-blue).

Anyway, the BIOS screen is fine, and I can boot into safe mode, uninstall the drivers, and then boot up normally. At which point Vista decides to automagically install drivers (stupid *#$%ing P.O.S. OS), and if I reboot, I get a solid color hard freeze. To rectify the situation, I've tried going into safe mode, completely uninstalling my display drivers and everything associated with them (nVidia control panel, PhysX, 3D stereoscopic viewer), rebooting into safe mode again, and installing every single driver revision since 180.48, published a year ago. I've even done this manually once or twice, with 3rd party driver file removal tools and registry diving. And every time I boot into Windows normally with any nVIDIA drivers installed, hard solid color freeze.

Now, I've called up eVGA's support, and got them to agree to a 2nd RMA on this card, but was told that it's probably not a hardware fault, since I can boot into safe mode, and the display's not garbled at the BIOS level. And I don't necessarily disagree with them, since I have a very good cooling solution in place, and the card's never gotten over 85 degrees C (104 degrees C being the max temperature this card can handle, and 85 being well within operating specs, at load). Of course, they also told me the card I originally RMAed passed their hardware tests, despite the fact that it Pink Screen of Deathed on me any time I put any load on the card (gaming, folding@home GPU, anything), so I'm not sure I trust their hardware tests to be worth a damn, to be honest.

Has anyone had any similar experiences with this card? Or have any suggestions for getting it working without exchanging it for a 3rd one which will probably also crap out in another 3 months? Or can someone confirm that this particular card is just utter garbage and prone to flaking out for no reason? (I've tried searching for reviews on the card, but noticed that it's not really sold anywhere anymore, so maybe they did discontinue the product for being a P.O.S.) I'm an IT professional, so this &#^% has me pulling my hair out - I'm not used to technology problems I can't solve, and this one is giving me a splitting migraine. Anyone with a similar experience to share would put my mind at ease, and at this point, there are no bad suggestions, IMO, since I'll have to be without a video card for a couple weeks over the holidays if I go through with the RMA... and I really wanted to catch up on my gaming over the holidays

 

[Mousemonkey]

You say you had to RMA a card before, was it for the same problem? and if so did the new card work OK? if yes to both then it would suggest a problem with the EVGA variant rather than the the entire line which has been EOL'd to clear inventory for the next gen, something about a global recession causing a cutting back of stock levels.

 

[Viperrox]

I got exactly the same error with a xfx GTX 280.... just tryed at Win XP 32, Vista 64 and Win7 64.... all them results in the same graphic error and them shutdown as soon as i stard a 3D application

 

[mincka]

I have the exact same problem with my ASUS engtx280 word for word. My card idled ~45C and under heavy load maxed out at 80(using evga precision), although only usually around 70-75 for my gaming needs. Normally a system restore would fix the problem but as of today this is no longer the case. I get the rainbow screen after login (like your screenshot). Safe mode works, a fresh install works until drivers are installed.

Also, I opened the card to replace the heatsink goo and that has unfortunately voided the warranty from ASUS. So really I(we?) are left with expensive doorstops.

nforce4max 11-22-2009 at 03:35:20 AM


Nvidia cards tend to fail after a year or more when operating at 70c + since when it cools it can cause micro-fractures. The heat can also damage power stage since that is the hottest part of the card during normal use (same applies on low end cards as well). Remove any dust that may remain on the backside of the pcb and you have said that you have already cleaned the cooler. Manually set the fan speed your self while keeping temps in the 60s.

Here's a quote that may shed some light on the situation.

 

[InfidelPimp]

Thanks, guys. I figured it was probably micro fractures... and am kind of glad to hear other people are having the same problem. Hearing from eVGA that they found no defects in my first RMA just made me want to check to make sure. At least it's not just me.

Also, I opened the card to replace the heatsink goo and that has unfortunately voided the warranty from ASUS. So really I(we?) are left with expensive doorstops.

Since yours is wrecked anyway, and not under warranty, you might consider baking it in your oven. (No, really.) It repairs the micro-fractures by melting all the solder connections evenly. Obviously, you'll want to remove all the plastic bits from your video card first, which will void the warranty. I figure it's a trick I'll have to use soon enough now that my card is no longer under warranty. Worst case scenario is that it doesn't work, and your expensive doorstop is, well... an expensive, slightly baked doorstop.

Of course, since the manufacturers obviously know about this issue, I'm having a hard time figuring out why they haven't done something about it (like put extra cooling around the power leads which are usually the hottest part of the card), or even change the recommended operating temperature ranges, but... <sigh> oh well.

 

[Mousemonkey]

This topic has been closed by Mousemonkey