RyuHimora

Honorable
Jul 24, 2012
10
0
10,510
Hello,
As I write this my GTX 260 is idling at a temperature of 85 C. Within the last few weeks, this card has suddenly developed crashing issues during games: the video freezes but the sound still works, albeit jerky, or the entire game crashes but I can still three finger salute out of it. I just put new silvered thermal paste on it last week and the problem has gotten worse with newer drivers. My power supply is a 585-watt model with a total of 38 amps across the 12v rails. Any suggestions?
 
38amps isn't a lot. Give us the full specs of you machine and tell what which gtx260 that you got either wise you are only going to bet noobs replying to your thread that aren't helpful.

The gtx260 really comes in two versions hardware wise. The old 65nm model that has the backplate and the other is the slightly newer 55nm version that is common. The problem is that the 55nm version was built very cheaply while the older version has a lot of problems with inductors going bad.

When it comes to cleaning these cards you have to blast them very good for a long time to remove all the filth inside the cooler or take it apart to be cleaned by hand.
 

Dangi

Honorable
Mar 30, 2012
192
0
10,690



I'm pretty sure your GPU is full of dust.

I also have one GTX260(216 cores) and is running 50Cº on idle and 74Cº on load
 

RyuHimora

Honorable
Jul 24, 2012
10
0
10,510
My hardware is thus:

AMD Phenom II X4 965 Black 3.4 Ghz
Asus M5A88-M
Corsair DDR3 2000 4GB
BFG Nvidia GTX 260
X-Power 585 PSU
Seagate ST3750

Since replacing the thermal paste involves taking the heatsync off completely I cleaned that out already.

My gaming problems have only started in the last month; before then everything ran fine including the idle temp.

How do I check for the version of my card?
 

Dangi

Honorable
Mar 30, 2012
192
0
10,690
Everest should tell you all about your rig and HDWmonitor everything about your temps.


Have you ever clean the fan before ?? Not the heatsink just the fan, and also try to blow air inside the card to undust your GPU without the need of dissambling it
 

RyuHimora

Honorable
Jul 24, 2012
10
0
10,510
I have determined over experiments in the last several hours that something has gone wrong with the heat transfer between the GPU and the heatsync. Normally when I play a game I can't put my hand anywhere near the exhaust stream, but now when I do it's barely warm. Any chance I might have damaged the heatpipe when I took it apart?
 
You can't damage the heat pipes by accident as you have to cut into them. Inside the pipes it is a vacuum like an old style light bulb so that the water boils at a lower temp to make that work. Take a look at the heatspreader as some have managed to pull it off when removing the cooler unintentionally.

Have any of the screws broken off or remain lose when they should be tight?


IS it the older 65nm version that has the backplate?
 

RyuHimora

Honorable
Jul 24, 2012
10
0
10,510
All the screws are accounted for and screw in tight, at least as far as I can feel.

There's a backplate that I have to put in 10 screws, yeah.

What's the heatspreader look like? If it's a gray thingy on top of the actual GPU die with the Nvidia logo on it it's still firmly attached.
 
OK its the older one then and is like my two gtx 280 which is basically the same. Is the heatspreader even all the way around? Just look along the edges where it meets the base where the gpu die is on and if it is even it should be the same all the way around. I found on one of my cards that it isn't even and temps are higher than I would like. When you apply thermal compound on these cards it has to be the entire heatspreader with a thin and even layer.

Also check the thermal pads at the end of the cooler for the power vrm phases that all of them are there or the mosfets will get very hot. The box like inductors are ferrite shielded and these are what kills these cards when they go bad. The easy way to tell if they are bad is to touch them and see if they wiggle at all or rattle when you shake the card a little.

At the end of the card where the dvi ports are there are two screws that have to be in place for the cooler to mount properly just in case you forgot. If the screws are lost you can improvise provided you manage to find other screws with the same thread.
 

plantlet

Honorable
Jul 18, 2012
84
0
10,630
Try applying thermal paste or using compressed air to clean some dust off. Maybe the cards are just failing. My dad has been running a GTX 260 SLI for 3 years and when I play Team Fortress 2, the cards only get the 58C. We clean the fans quite often in the case, though my dad has been meaning to apply thermal paste for forever now but never got to it.
 


The mosfets on these cards (65nm models) are fully digital and don't run as hot as many modern cards do. Also they got thermal sensors like some amd/ati cards do.


Compound alone doesn't work, it needs thermal pads or metal shims to allow better flow of heat to the main cooler or it will heat up to quickly for the compound to pass it on but you can check the vram temps in gpu-z towards the bottom. Even logs the amp loads :sol:


EDIT: I don't know why I forgot to say this earlier but when you first change the compound and it is arctic silver 5 or MX-4 you will need to bake it in for a while for temps to slightly improve. I always kept my fans at 80% even when not gaming and it kept the cards at decent temps despite my hot room.