GTX 260 crashing

I've been having an issue with my EVGA GTX 260 SSC. It worked fine from when I brought it (Feb) till around August, when it started crashing randomly with a full colour on the screen (when first loading a game, or even hours into a game, although the crashes became more frequent until it would crash all the time within a few minutes, it even crashed once on watching an episode of South Park). I RMA'd the card but EVGA didn't find anything. In the meantime I was using my 8800GTS with no probs for months, and now I've clean installed Windows 7 and am still getting crashes with any driver.

There are 2 things that I think could be causing it; temps and power, and these are the tests I did for each:

Temps:

I've run the card at stock SSC clocks with 100% fan and the case side off, temps never go above high 50's - low 60's, and it crashes.

I've run the card underclocked with auto fan speed and case side on, temps hit low 70's but fan stays at 40%ish, no crashes, so it's not heat related.

Power:

I used to run the card undervolted at 963mV instead of the normal 1063mV, which worked fine for months until the crashes started happening. However all tests now are at stock voltage settings, so it's not an undervoltage issue.

I've tried the stock SSC settings with my CPU also set to stock settings, and I get crashes, so it's not the system's overclock (100% stable with my 8800 anyways).

So I'm left to believe that it's either the card itself (but EVGA found nothing wrong with it), or the PSU is dying and cannot provide enough current on the 12V rails (must provide enough for the 8800 but not enough for the 260), although I'm pretty sure this isn't the case as it still crashes when my system is not overclocked.

Anything else I can try?

pluck that card out and see if yer PC crashes again if not its the PSU.

it might be that something on the card is getting worn out. If it doesnt produce any crashes at normal 260 speeds, but does at the stock ssc speeds, I'd say the video card is the culprit (or possibly drivers).

I was getting strange/similar issues with my old 260 a while ago when I overclocked it a lot. I upgraded the drivers and blew it out with a can of air and it was fine again. Not sure what the actual problem was, but If I had to guess I'd say that either something inside the GPU was getting warmer than it was reporting, due to the dust, or just simply a buggy driver. I thought it might of been the PSU, but my new card draws a lot more power than the 260 and haven't had a single issue

I'd try cleaning all the dust out of it and use another driver version and try it again.

@dinlee: Been swapping my 8800 GTS and GTX260 over the last few months, and the 8800 has always been rock solid (had it in my system for the last month and a half).

@dr.stevil: Actually I think it started around the time of a new driver release during late July/early August, but since then I've tried multiple drivers (older and newer), on my old Vista install and my new Win7 install, all with the same issues. Also tried cleaning the card/fan with pressurised air, but I'll give it another shot.

I should note, that I'm currently using the latest 195.55 Beta drivers with the 260, although I've tried others previously.

does it artifact at all?

If you can't figure it out, I'd deff. RMA the card and ask for a new one. Something is obvously wrong with it IMO if new drivers didn't help ya. Id give some really old drivers a shot too... just to test it.

ironically enough, thats around the time my card started giving me problems, so I'd say that the issues (while not widespread) are releated. Just the video driver crashes right? Then windows recovers? Or does it physically lock the machine up? (or BSOD?)

There aren't any artifacts, but it does lock-up. It's not a BSOD, but the screen completely changes to a random colour (usually red, yellow or grey, depending on the card's mood on the day), as when a graphics card is dying. I'll give some really old drivers a shot and see how it goes; played a few campaigns of L4D2 at underclocked settings and it seems stable, let's see how the old drivers work at the stock SSC settings (underclocked is even more underclocked than stock GTX 260 settings; 450MHz core and 950MHz memory just to make sure). Thanks for the help

I had similar problems with one of my XFX 260 Black Editions. Games would lock up and the screen would either turn a random solid color like you described, or sometimes it would display a kaleidoscope effect with a static pattern of pretty pinks, greens, etc. Either type of lockup would require me to turn the PC off by holding the power button. It started happening more frequently; sometimes I wouldn't even be able to game for 10 minutes before the freeze occurred. And it was always on more graphically demanding games. I never did have any artifacting, though.

I did some research and found some other people experiencing similar problems with their 260s. Common suggestions were either an inadequate PSU or a bad capacitor on the video card. I had tried the card on both a 650W Silverstone and 1000W Corsair PSU with lockups on both, so I ruled the power supply out. I contacted XFX to ask if I should RMA the card and they wanted me to go through all kinds of troubleshooting, including trying different drivers (I had tried several sets already), lowering the card's clocks, and increasing fan speed. I didn't think I should have to lower the clocks on my factory overclocked video card, but I tried it and it seemed to help stability. These cards run HOT. The Black Edition cards were overclocked from the factory to perform on par with a GTX 280. When I added a second card for SLI, I really had to crank up the fan speeds to keep the cards stable...even then they were burning hot to the touch when I'd shut my rig off. In the end I think my problem stemmed from the aggressive factory overclocks on the cards combined with a lack of sufficient cooling. Granted, my case could probably be cooler as I'm just running two 120mm fans in a Rocketfish (one intake, one exhaust). The Antec 1200 that I sold would probably keep the cards a lot cooler.

If i were u try borrowing a friends PSU or sumthing a 500 watt PSU will do but make sure its a good brand like Antec, seasonic, Corsair, etc... and if it doesnt have problems with that then there you go problem solved

That's EXACTLY what I'm experiencing. I sometimes used to get the kaliedoscope effect also, or a weird effect where 3D models were shown in wireframe mode and the rest of the screen was pink/green e.t.c (but usually it's a fullscreen single-colour crash). You're right about the heat, my card after a few minutes is too hot to touch (with the case open and fan speed at 100%).

Good idea dinlee, I've asked a friend to let me try it out on his rig: he has a 5870 (lucky guy!) so his PSU should handle my card with no problem. That will be this Saturday, so for the next few days I'm going to give running it at stock SSC clocks a shot, with the card's fan at 100% but also with a huge desktop fan on it's highest setting to make sure there isn't a deadspot on the card or something.

I'll report back, thanks for the help guys

I had an eVGA GTX 260 SSC as well that started to crap out on me.

Does it look like this?


Machine would hard lock and look like this,


and after a few seconds it would look like this,


I'd have to hit the reset button the pc to restart.

My card was only 2 months old when it started to do this and it only did it in Counterstrike :Source while playing Gungame on 1 particular map, Lego Arena.

a youtube video
http://www.youtube.com/watch?v=cGtsCSXW4QI

Instead of RMA'ing the card I opted to step up the card to the 285 and decided to pay the $20 extra for the 2 day Advance RMA feature,
this way if the card has issues they'll send me a card first, then swap, and send the old one back, all prepaid by eVGA.

Yep exactly like that; either a single colour or like the images that you showed (sometimes with the wireframe of the 3D models present too), except mine only ever returned to desktop once I think and now it's always a single solid colour. Was yours the 65nm version also? Here's some pics I took a few months ago (10 points for figuring out which L4D1 level the second shot is from):





I wish I could have traded up, but it happened 6 months after I brought the card, and at the time I got the card for cheap so it would be a more expensive step up. On another note, I got through a whole campaign of L4D2 now with the fan at 100% and the desktop fan pointed on it, if this continues then there must be a deadspot on the card which heats up more than anywhere else, which the desktop fan is cooling (as it crashes at 100% without the desktop fan), rather than the PSU? I'll continue for the next few days and report anything that happens!

I'm still putting my money on the video card. Maybe a dead spot or possibly a manufacturing defect thats finally showing up from the higher clocks... try just dropping your vram frequency down and leave the rest where it is from the factory. If that still locks the machine up, lower your core/shader and leave the memory where it normally is.


Should make it easier to narrow the problem area down

I've seen an FX5200 do the same thing out of the box... just not nearly as bad as the shots you guys posted (displayed those 'artifacts/blobs', but you could still use the computer)

I've had my GTX 260 crash just once like that. It was a hard lock in about 10mins of CSS. I had an unstable CPU overclock at the time. Once I got a stable OC, it never occured again.

Mine was the 65nm version I believe, 192 version that showed up as 216.

that L4D map is where you start in a loft and either jump down or go through the door down at the end of a small staircase.

My pretty screen:



Any game newer than TF2 does this to my computer in less than 5 minutes. XFX GTX260 Black Edition.

edit: BTW this last time I had a GPU-Z sensor log going and my gpu temp was at 61C when it crashed. =/ What the deuce?

Ok so after messing with this all day, I've found that if I underclock my Black Edition card to reference clocks and crank the fan up to 100%, I can get through a 15 minute run of FurMark. The reason I'm posting is I took a screenshot of this last run and am wondering if you guys see anything that I'm not that could be causing such bad behavior from my card. Peace!

So I did a few more tests this weekend. I ran the card on a friend's rig and it didn't crash at the time, but the test was only a brief 20 minute OCCT run. It got to 84 degrees and didn't crash, however later on I did the same test on my computer and the card lasted just as long, but crashed when I opened IE (this happened a few times). I think it was a fluke that it didn't crash at my friend's house, ambient temperatures are pretty much the same (it's cold here in the UK, and we have our heating off). I started to test with Furmark instead, and crashed within a few seconds of getting into the 70's. I also took home his spare (fully working) power supply and tested with everything in my system running at stock, and I experienced the same crashes on my computer, so it's not my PSU.

That IE crash made me consider my motherboard/memory, so I increased the voltage on memory and all mobo components (inc PCI-E slot) and got the same crashes. I then tested my card and PSU on my dad's computer and got the same crashes (it kept crashing within a few seconds of getting into the low 70's in Furmark) so I'm sure there's a heat buildup somewhere on the card. I took the heatsink off and applied some AS5, it runs slightly cooler but crashes the same. I'm just wondering how EVGA didn't pick it up when they were testing it?

@jermany: That's exactly what I'm getting too, underclock to stock 260 clocks and/or fan at 100% will be stable, normal SSC clocks and it crashes. Sounds like you're getting a heat buildup somewhere on the card aswell? Have you tried it in another system?

Here's a pic of a resistor I thought looked a bit dodgy, but at the time I didn't notice the chip with the pink circle. Does it look like it's burned?

Zepher said:

that L4D map is where you start in a loft and either jump down or go through the door down at the end of a small staircase.

Click to expand...

Zepher += 10 points;

Rajveer said:

@jermany: That's exactly what I'm getting too, underclock to stock 260 clocks and/or fan at 100% will be stable, normal SSC clocks and it crashes. Sounds like you're getting a heat buildup somewhere on the card aswell? Have you tried it in another system?

Click to expand...

I haven't. This is the only rig I have. Fortunately XFX is being pretty responsive with the support ticket. I'll let you know how it turns out. It does sound like you're right though, and some part of the card is getting fried.

http://hardforum.com/showthread.php?t=1462586


might this be the cause? i also had some unstable issues and moving to another driver fixed mine for the moment. im worried something on the card has burned up since its still acting a little odd compared to how it did before this driver update

I've been keeping up on that thread as well, and I think dbphelps might be on the right track. In that the previous drivers haven't been utilizing our hardware fully, and the new drivers are taxing the cards to a point that our factory overclocks are no longer stable. Is anyone having these problems on a vanilla card?

The only thing that makes me doubt this idea is that my card still crashes at reference GTX260 speeds unless I crank the fan speed. This could be because my card is already damaged though. Hard to say...

Ugh Ive been having this same issue with my EVGA GTX 260 Superclocked edition, Ive had the card since November of last year, It worked great until the last couple months it started Green screen crashing when playing L4D1 and now L4D2 (during this green screen the sound is stuck in a loop and pressing Reset doesnt work, have to power down)

Its totally random when it happens but Its starting to occur more frequently, I recently upgraded my motherboard / cpu / memory but the crashing still occurs, Im running WinXP 32bit and Win7 64bit (dual boot), The crashing occurs in both OS's which are on different hard drives altogether, so the only real hardware left to blame is either my PSU or Videocard.

PSU was bought with the GTX 260 so its only a year old as well (BFG 650W) The voltages all check out fine, even under load.

It's worth noting when the crashing first occured I started monitoring GPU temps and they were close to 90C during L4D2 sessions, I removed the card and blew it out with air, this brought load temps down to 75C but it didnt help with the crashing.

Ill have to try underclocking to stock GTX 260 speeds and see how long it lasts before crashing. (its currently overclocked from factory and ive never OC'd it any higher)

Current PC: (previous build in signature)
i7 920 @ stock speed
6GB Corsair DDR3 1600
Gigabyte EX58-UD3R (rev 1.6)
Realtek HD sound (onboard audio) I initialy had my X-Fi card installed but when I got my first Green screen crash on new system I took the card out to troubleshoot, crashing still occurs.
WinXP SP3
Win7 Ultimate 64bit