GTS250 in a water bath.
-
Hello everyone, I'll tell you what happened to me on the beach PC. It's a Q8200 (stock speeds) mounted on an Asus P5Q3 with 4 gigs of ram at 1333, all put in an Asus Vento A8 case and a 500W Tacens power supply.
Although I don't use the PC to play much, I had that graphics card (GTS250) in case I felt like playing some games. When I started the system (a Windows 7 64 bits) I detected that one of the hard drives (WD 3Tb) was starting a scandisk, I left it and since it's 3Tb and full at 2Tb, it took about 20 minutes, just at the end when it gives you the scandisk report, it froze and a few moments later it turned off. I tried to turn it on and it turned off instantly. When I approached it, since I have it on the table, I noticed the heat coming out of two 80mm vents on the back, which I cut to put some fans, but right now there's nothing. I quickly disassembled it and you couldn't even touch the graphics card, let alone hold it. The fan had stopped because the bearing wasn't fine.
I proceeded to change the fan and what I had on hand was a 90mm one, which fits almost perfectly. I disassembled the heatsink to change the Titan nano grease thermal paste, and to be able to attach the fan to the heatsink with some brackets.
I tested the temperatures and they were crazy between 60-67ºC at Idle and at full load, tested with Rage around 106ºC. If I put the furmark we're not even talking, they exceed 110ºC in a minute. Apparently everything points to a bad placement of the heatsink, but I've placed it around 10 times and every time I remove it I observe the mark on the heatsink of the paste and it's perfectly seated. When the heatsink is placed incorrectly the temperature skyrockets in seconds, this is not the case, the temperature rises progressively. Another important detail is that the heatsink is boiling, so there must be good contact between the core and it. To do tests I placed a Titan CPU heatsink, which practically doesn't fit in the case and obviously the temperatures are lower but it starts to heat up and doesn't stop and in the end we're the same as with the stock heatsink.
The graphics card is now like this:

The case although well ventilated, I've also tried leaving it open, with identical results.
The room is at 27ºC but I've put the air conditioning on and it's still the same.
I saw on PCComponentes a graphics card heatsink the Accelero L2 Plus with a 90cm fan too, but it gives me the feeling that the graphics card might have been damaged.
When I play in 3D I don't notice any artifacts, or anything weird, nor does it slow down or anything, it just works at 105ºC.
I've changed the drivers and get the same temperatures.
The heat it emits inside the tower is unbearable.
What do you think?
Could the graphics card have been damaged, even if it's still working correctly?
Regards and thanks in advance.
P.D. sorry for the long post. -
The graphics card must have died. It's like some preshots that got messed up and worked at 80º all the time. After a while they died.
But what I find strange is that after changing the heatsink, you still get those temperatures. Try taking that heatsink and putting it on with screws instead of with clamps, so that when it heats up it doesn't warp and not make good contact and fool you when you take it out cold.
And for what the heatsink costs you, for a little more you'll find a better second-hand graphics card, I think.Regards.
-
I can't find another explanation, but as an electronics expert, it's hard for me to accept that something that works "correctly" only has the problem of getting hot.
When I get to Seville, I'll put an RL (I have it unused) on the graphics card, we'll see how it behaves.
Thanks.
Salu2. -
Man, there may have been electromigration, and it may have messed up what regulates the temperature of the graphics card, some sensor or something else. I didn't believe the "presshot" thing either, they put it on the forum until I got two in the same week. You don't know what it was like to start the PC and have the temperature go up to 80º in no time. And it stayed that way no matter what you did.
I thought it was the motherboard, I changed the motherboard and it was still the same. Changed the micro and it was perfect.
You got hit. -
Well, I thought that since the heatsink is pressed against the GPU core with 4 springs, I could add 4 washers to apply more pressure.
Conclusion, where it was before at 105°C now it didn't go over 74°C. A smile came to my face.
My joy didn't last long, after 1 minute this came out.

It hasn't started up again.
I don't know what happened to it. -
I think that some washer may have made contact with you, or that you tightened them too much and damaged the core, or some of the memories.
I know that there is a pressure at which the heatsink makes perfect contact to improve temperatures. Because in many old ones (now I don't know about the issue of hooking up with levers) you had to apply a lot of pressure to hook it up. In fact, more than one socket tab broke, and you looked for heatsinks with hooks for all the tabs, which applied even more pressure and distributed it better. The less thermal paste left between the core and processor the better. And that's how it's achieved.
But I think you went too far and made contact or broke it. It used to happen with the k6-2 that we opened up, that from changing and testing so much we were breaking pieces of the core and they ended up dying. Well, and for going too far with the oc.Greetings and may you find another cheap graphics card.
-
Yes, it sounds completely like cracked silicon… This reminds me a lot of an Athlon XP that had the pads in bad condition, in the end the heatsink broke the corners of the silicon core and that little chip turned into a crack from tip to tip. End of its life.
If you are not planning to invest, look for a cheap second-hand graphics card, there are good options and even better than that GTS250.
Regards
-
Well, I've disassembled the heatsink and it seems to be perfect. I'll post a photo later so you can see it.
The solution was simple, Gigabyte GT630 with 2 gb ddr3 128bit. 62€.
This one in particular: GIGABYTE - VGA - GeForce 600 Series - Chipset GeForce GT 630 - GV-N630-2GI - Specifications
But I think what was happening is that from the first overheating, a solder joint weakened (it wouldn't make good contact), when the pressure increased it made contact and worked better, but in the end it ended up breaking.
When I'm in Seville I'll give it a little heat in the oven, to see how it goes.
Regards and thanks to everyone.
So that's it, I'll upload the photo of the core later.
http://img41.imageshack.us/img41/2979/suqu.jpgBy the way, the temperatures of the new one are nothing like the old one.
Idle –> 32ºC
Full Furmak --> 52ºC
The power consumption is also nothing like the old one, 150w from the GTS250 against 65W from the GT630.
I've continued playing Rage, which I had forgotten about on the beach PC, and as for gameplay, I haven't noticed any difference in performance. -
Well, I've disassembled the heatsink and it seems to be perfect. I'll upload a photo later so you can see it.
The solution was simple, Gigabyte GT630 with 2 gb ddr3 128bit. 62€.
But I think what was happening is that the first time it heated up, a solder joint weakened (it wasn't making good contact), when the pressure increased it made contact and worked better, but in the end it ended up breaking.
When I'm in Seville, I'll have to put it in the oven for a bit, to see how it goes.
Regards and thanks to everyone.
Anyway, I'll upload a photo of the core later.well if that graphics card is going to the trash, let me know, I'm looking for things like that to keep myself entertained. I get very bored
-
bueno si esa grafica va para la basura, avisa, que ando buscando cosas asi para entretenerme. que me aburro mucho
Ok en cuanto llegue a sevilla la hornee y no rule si quieres te la mando.
Salu2. -
I have to tell you that the guy at the store told me that if I didn't like that gt630 (it was the only one they had in stock) I could exchange it, and after trying it out a bit I decided to exchange it for an Asus HD6670, by putting in 6 € more.
Salu2. -
it's not a bad choice, I use yours as my main card and an nvidia as a secondary one. -
By the way, how do you see the gpu?

Thanks. -
Do you have two resistors bridged?
-
It seems it's just thermal paste. It doesn't look bad, it looks good at first sight.
-
I would say they are capacitors, but they are not bridged, it's the thermal paste, which is supposed to be non-conductive. At least that's what I think.
Anyway, I'll remove the paste to be sure.
As for the gpu, I was referring to the fact that I had been told that the image of the failure was normal when the silicon broke.
Apparently nothing is visible from the outside.
Regards. -
I would say they are capacitors, but they are not bridged, it's the thermal paste, which is supposed to be non-conductive. At least that's what I think.
Anyway I'll remove the paste to be sure.
As for the gpu, I was referring to the fact that I had been told that the image of the failure was normal when the silicon broke.
Apparently nothing is visible from the outside.
Regards.The SMD whether it's a resistor or capacitor, it doesn't matter, the thermal paste is never 100% insulating and more so if it's metal-based; clean it well if you don't want problems. It also looks like a separation in the silicone that surrounds the core, right? And I didn't understand that about the silicon breaking, what do you mean, do you see the surface of the core broken or loose? Usually they used to break visibly, it's possible that some internal solder was already bad as you say.
-
It's the effect of the light, it came in from that side and was reflected giving that impression.
Regarding the broken silicon, I meant that the colleagues had said that the screen capture I had put up before with the artifacts could be a symptom of the core being broken due to the increased pressure from the heatsink springs.
"In my defense" I have to say that the washers are tenths of a millimeter and even when I pressed them with my fingers, they still gave way and sank more. So they didn't have to apply much pressure to bend those washers.
What I do notice is that the 4 corners of the core are scratched and there's nothing in the center, could that be a symptom of a bad fit between the core and the heatsink?
Salu2. -
It doesn't look broken, what we were talking about was something more striking, like silicon shavings scattered around and the corners "pierced".
I don't know about a bad fit, but what it is a symptom of is that the heatsink has moved quite a bit over the core, manually I suppose, since those scratches aren't made with the standard placement (or at least I'm not aware of it).
Regards
-
The truth is that it is not so scratched at least with the naked eye, what happens is that with the macro zoom and 18.8mpx you can see even the hairs like sticks.
I probably made a mistake when I said to remove it and I put it more than 10 times.
I will continue to investigate to see what it could be.
Tomorrow I will remove all traces of thermal paste and I will tell you.
Regards.