Malfunctioning GTX295

Aloha!

Having a problem with the GTX295 card, to be exact, there are actually two of the same cards. Paid a great quid to enjoy the quad, in the end it went pretty terrific. Unfortunately, I never managed to build a SLI system on them, and they had been stored for ages unpacked. Until the last year, when I slotted one onto this spare PC.

The symptoms seem to be classic, as I've found out many users have been complaining so far. On heavy load, particulary during gaming, not exactly bloody Crysis or something, but FX intensive games, the screen turns black with No Signal. Anything on background, shall it be Skype or in-game music, keeps going, however at times it happens to completely cause the PC to hang on unresponsive (notably, with sound buzzing). The first card lasted half a year, tested it, came to a conclusion of it going FUBAR, packed back up and slotted another... just to know it to do the same in 6 months again.

It doesn't display any kind of artifacts that may point it to be a GPU/soldering failure - no tearing, lines, cubes of some kind, or anything used to determinate the obvious condition. The cards rarely crashed in 8bit>driver BSOD with games being minimized while simultineously watching Youtube or loading browser with crap. Mostly it was just the No Signal one. So far the only suspicious crash was a few days ago, it messed up 2D textures on a HUD, had to reboot.

The system is not glorious:
Q6700 3.6ghz 1.44v
P5KR 450mhz
Asus GTX295 TOP 604/1053/1368
6GB mixed up
CoolerMaster GX550W Bronze rev

The hardware is totally cold and there is enough power output. All lanes tested with a voltmeter, they are fine. The PSU is hardly warm on the heaviest possible load, Furmark/Linpack (and yes, it DOES NOT crash the GPU, the games do). And yes, people have had far better PSU and yet to produce the same crashing, certainly not the power.
One constant issue that could be the culprit is memory. It is failing with huge numbers of errors during memtestG80/VMT (over/underclocker, ref stock), however no other tests seem to care.

Should you check out the web for related problems with the GTX295, you will definitely find every second issue to be exactly the No Signal failure. I don't have any high-end replacement, so probably will go for a repair.

Did anyone run into the same problem operating a 295? What are the possible solutions?

Garbage and get something remotely modern!

If you are having memory issues, correct that first, but a black screen with functionality of backround processes isn't a memory issue. Have you tried different drivers?

That PSU though... frighteningly mediocre at best... http://www.jonnyguru.com/forums/showthread.php?t=6424. ANd if it's a few years old, and gets warm (though you say it doesn't), it may not be able to handle the card and CPU. I know you said it wasn't the issue, but that is brave running on that PSU with that setup... this coming from someone who runs a 295x2 (AMD dual 290x card) and an overclocked Hex Intel on a 750W PSU...

does the other card do this?

your PSU is too small for two of the cards. The single GTX295 can pull around 300W by itself.

http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-295/specifications

http://www.guru3d.com/articles-pages/asus-geforce-gtx-295-review-(engtx295),9.html

I think he is only using one?

Until the last year, when I slotted one onto this spare PC.

Click to expand...

Ahh missed that part, just saw that the OP had two of them.

Nope, the PSU is a year old and it never had crashed the system or ran out of power. The fact is the cards were working for quite a long time before they went broken. Running everything on stock proves it, but I have doubts. Noticed many did run their 295 on luxary PSU, still got the crashes. May also be not enough amperage/wattage, as it is PCI-E 1 and short on 5-10 amps from the PSU. Yet is doesn't explain why it worked just well out of the box.

Don't have intentions to recycle them for now, short on cash. And honestly, want them to run for a little more.

P.S. Running one at a time, obviously
P.P.S. The cards are single PCB in the ref cooling.

May also be not enough amperage, as it is PCI-E 1 and short on 5-10 amps from the PSU. Yet is doesn't explain why it worked just well out of the box.

Click to expand...

I believe that PCIe1.1 still provided 75W at the slot. You may want to confirm by looking up that specification.

As far as it working out of the box, well, new PSU has fresh caps and everything so it should be able to output its rated power. As time goes on, things aren't 'perfect' as they are brand spanking new. That PSU has cheap internals and is only rated for 25C operation. As was noted in the link I provided, the 'experts' stated that there is a really steep curve above 25C as far as how much power it can put out.

Have you looked at temperatures of the card?
Again, did you try reinstalling your driver, new one? Older one?

Quade said:

short on 5-10 amps from the PSU.

Click to expand...

There's your problem

I dont understand how he says its enough in one breath(post) then says he is 5-10A (SIGNIFICANTLY) short in another..............

That could definitely be the PSU. Just here, trying to figure it out. However, that is quite odd the cards were performing well freshly unpacked.

The temps are just fine, never above 65c in games and Furmark hits 80s, barely. Nothing that can beat this 8800GTX's 65 at idle which I plugged in for a test today The 12v rail was stable 12v back when I checked it with a device.

So that PSU is a no-go? Should grab them and test with a better one, perhaps, maybe the cards have not been permanently affected.

......

Again, did you try reinstalling your driver, new one? Older one?

Click to expand...

Yes. 290 (I believe), 296, 310, and the latest 2. The older are causing LESS crashes. I did not try any earlier as the 296 was quite a punch of the performance to the games I usually play.
Complete wipe with DDU.

Hmm... well, at this point I would try two things...

1. Try it with a bigger, high quality PSU
2. While core temps may be in line, the VRM temps may not be. Is the back of the card hot? Can you try blowing some air across the back of the card and see if that helps?

Otherwise, i would sell them and get something different. Not worth the time or hassle.

Quade said:

That could definitely be the PSU. Just here, trying to figure it out. However, that is quite odd the cards were performing well freshly unpacked.

The 12v rail was stable 12v back when I checked it with a device.

So that PSU is a no-go? Should grab them and test with a better one, perhaps, maybe the cards have not been permanently affected.

Click to expand...

Aging PSU caps can cause issues to start happening when you're borderline on power.

Just because the 12V rail was stable doesn't mean it can give enough power.
And if you checked with a multimeter you can't see ripple anyway.

Yes, test with a better one.

Definitely give the dust bunnies a clean out, if appropriate. I cleaned mine out 2 days ago, and am noticing a pretty big drop in temps (~10C+ [50F+)) afterwards. Have you tried checking the PSU? Might need a clean out!

with my evga gtx 295 i had same issue only fix for me was to change clock speeds ... from memory on the evga gtx 295 FTW you need 42A on the 12v rail

Hey, thanks for your input!

Nope, the system is crystal clean. I've recently had a clean up, besides the whole PC is open stand, so the temps are not an issue whatsoever.
Just did an express test out of curiosity, Linpack 1k spam and Furmark burn-in (SLI mode). It lasted for 30, then 60 minutes, then I actually had to shut it down due to boredom This should have had basically loaded it up pretty bad, enough to instantly crash it. Didn't happen. Started Hitman Absolution, crash to No Signal in the menu.

The temps were fine: CPU 60c, GPUs 80c. Everything else is stone cold.

This just doesn't make any sense. It crashed even on 50% clocks downgrade.

P.S. The PSU holds 44A on the rail, the official specs stated you actually need 46A. Theoretically, it should do it.

I can get access to the same evga card to show you ... mine only crashed in game tried every fix I read only thing that got in working in game for me was lowering clocks and didnt have to lower far either

DopeLovinghippy said:

I can get access to the same evga card to show you ... mine only crashed in game tried every fix I read only thing that got in working in game for me was lowering clocks and didnt have to lower far either

Click to expand...

Yeah, it doesn't seem to help in my case... It was crashing bad even with the 2D clocks.
Did you try any drivers prior to 200?

Yeah I reckon the driver I had best luck with was 197.45