Is this GPU unfixable? (675m)

[Jord1990]

First off I'm on mobile so I can't add dump files sorry, but I've had multiple people going over the dump files already and they asured me that there was no problem there.

It was a very progressive problem one even while playing skyrim I suddenly got my first blue screen ever(note that nothing was overclocked nor did I switch drivers recently) the message was 116 nvkflt.sys. so I thought well that is strange gave my pc a rest and the next day booted it again.

Played another full day again and it crashed in the evening. And the next day it crashed once in the afternoon and once in the evening. And so on to the point where I couldn't even launch a game to the main menu.

When I checked in gpu z I noticed something that struck me as strange the gpu voltage would hang around 0.800 but as soon as I launched something using the dedicated gpu it would spike up till 0.975 and a 100% load. Sometimes it would go down to 0.870 or 0.850 but most of the time it would be stuck on 0.975.

I have removed the gpu and am now running on horrible fps with intergrated but no more bsods. Software wise I have tried everything and the card is out of warranty.

Thanks in advance

[FireFox]

The voltage doesn't seem to be out of the ordinary, my AMD Radeon HD 6470M does the same but if the GPU is causing BSOD then it probably will be beyond repair, just stick with the i-GPU or look for a replacement GPU though I have no idea on costs.

[lani100]

Try increasing/decreasing the voltage if possible or the clocks.

[Jord1990]

Oh something I forgot to mention that before it got this problem sometimes when it had to switch to or from a full screen game the screen could go black and stay black until I put it in sleep mode and back out of it.

[Jord1990]

Try increasing/decreasing the voltage if possible or the clocks.

I have tried that however it just seemed to ignore that I've also tried different biosses on lower voltage.

[Jord1990]

What graphics card and PSU are you using?

Please be specific with the model.

Gpu is a 675m and for the psu I play on a laptop so no idea.

[Jord1990]

So its a laptop...

If it is indeed a hardware problem and you can't get the seller/manufacturer to fix it, you're out of luck.

Like I said I got the gpu out of the laptop so how would you fix a desktop gpu that has this problem?

[GuruMeditationError]

You could try this

 

If choose you do that, don't blame me if the card dies completely.

 

The graphics card isn't just one component. Its hundreds of components on a circuit board. If one of them fails, the whole thing can become unusable. The failed component may not be visually identifiable, let alone replaceable.

 

If you do try that, be sure to leave it plenty of time for it to cool down. If you pull it out of the oven hot/warm, the solder will still be liquid/soft and it'll ruin the board. You have to give it adequate time to cool and re-solidify.

 

As far as I know it's not really a permanent solution but it might give the card a longer life.

 

If it works but then fails again after time, you could see about getting the chip re-balled but if it holds out long enough it might not be cost-effective by the time it fails again (not ideal but at least you got more value out of the card).

[Jord1990]

Thank I'll try this in 3 days, because I'm away from home atm, if it works for even a week i'll be happy.

Will post an update

[Guest Generallee]

If it's an MXM gpu, ultimately you could just replace it

[Cwazywazy]

Open HWMonitor or something similar and play a game for awhile and then use it to check your max temps. I'm thinking it could be overheating. (90c+)

 

If it is, try replacing the stock thermal paste with something good. (IC Diamond or Noctua NT H1 or something similar.) If not, try taking it out and then baking it for awhile. (Look up reflowing GPU in an oven.)

 

Or send it to me. :3

[Jord1990]

Open HWMonitor or something similar and play a game for awhile and then use it to check your max temps. I'm thinking it could be overheating. (90c+)

If it is, try replacing the stock thermal paste with something good. (IC Diamond or Noctua NT H1 or something similar.) If not, try taking it out and then baking it for awhile. (Look up reflowing GPU in an oven.)

Or send it to me. :3

It isn't overheating doesn't get above 76 degrees Celsius before it crashes.

[Jord1990]

If it's an MXM gpu, ultimately you could just replace it

I'm gonna get a new desktop pc but I would love to salvage this for now so that I can see what the amd will bring to the table with the 300 series.

[Cwazywazy]

It isn't overheating doesn't get above 76 degrees Celsius before it crashes.

Okay, so if the chip is removable you can try taking it out and oven reflowing it. If it's integrated into the motherboard I'd recommend using a small butane torch over the chip to reflow it. (I've done both methods and both have worked.)

[Jord1990]

Okay, so if the chip is removable you can try taking it out and oven reflowing it. If it's integrated into the motherboard I'd recommend using a small butane torch over the chip to reflow it. (I've done both methods and both have worked.)

How bad will it smell because I might have a hard time explaining to my parents that I put a gpu in the oven.

[Cwazywazy]

How bad will it smell because I might have a hard time explaining to my parents that I put a gpu in the oven.

As long as you take off any plastic stickers or anything like that and the thermal paste it shouldn't really smell at all. I tried to oven reflow a laptop board once and forgot a couple plastic stickers and it smelled awful. I also did a couple desktop GPUs and those didn't smell much because I took off all the stickers.

[Jord1990]

You could try this

 

If choose you do that, don't blame me if the card dies completely.

 

The graphics card isn't just one component. Its hundreds of components on a circuit board. If one of them fails, the whole thing can become unusable. The failed component may not be visually identifiable, let alone replaceable.

 

Open HWMonitor or something similar and play a game for awhile and then use it to check your max temps. I'm thinking it could be overheating. (90c+)

 

If it is, try replacing the stock thermal paste with something good. (IC Diamond or Noctua NT H1 or something similar.) If not, try taking it out and then baking it for awhile. (Look up reflowing GPU in an oven.)

 

Or send it to me. :3

 

So with a bit of skeptesism I did this and i'm posting this after a 2.5 hour session of arma 3 without any blue screen. (knocks on wood)

Thanks for taking the time to post something that might extent the life of my gpu even a day.