Clevo P751DM2-G GPU upgrade feasible?

There must be some horrible ripple going on for a ceramic cap melt like that, or the cap itself was a Y5U crap, given the current shortage of passives, wouldn't surprise me that its second or third grade ceramics, and it was too much ripple for it, being in the middle of the MOSFET's and thus hotter only accelerated the problem, it might be repairable, at least I would de-solder the cap, clean the whole pcb with IPA, solder a new 10uF 50v X7R 0805 cap and see how it goes.

If you dont want the card I will pay shipping for her (only half kidding).

Also, did you solder the extra MOSFET?

EDIT:
The gate trace comes from the bottom mosfet through a via, the top mosfet in the photo isn't doing anything, the gate trace is interrupted with the pads of a resistor, so, that mosfet was flapping in the breeze, if you soldered the mosfet yourself, lack of gate resistor might be the culprit, even more so if you did the same mods to both cards and you didn't notice that.

At least it looks like so, there is a via near those pads, but I can't be sure if its connected or not.

 

He didn't do any Mods to the 1070 MXM, that's how the cards look when they are shipped from Canada.

His P5000 is also vBIOS locked and can not even be overclocked as it is.
It was all stock.

The above driver doesn't change voltage but suddenly drops clocks to almost nothing and right backup to full, causing many games to stutter and as we can see even hardware to fail...

 

Constantly jamming the clocks from min to max is extremely stressful for the VRMs.

 

So is it possible the issues I was having with crashing and artefacts was as a result of the driver?

 

https://forums.geforce.com/default/...y-driver-feedback-thread-released-5-9-18-/21/

 

Interesting, I rolled back to 391.24 and I no longer get the artifacts, instead it just crashes after a similar amount of time (maybe slightly longer than before). Each time it managed to recover though but the the error "Display driver nvlddmkm stopped responding and has successfully recovered." is still logged in the event viewer. I might try rolling back further and see what happens when I get a chance but unforunately not looking too promising.

 

Sounds like the drivers might just be extra stressing a failing part.

 

A little late, but I just wanna say, 4.7GHz 7700Ks bottleneck a single 1080 in heaven in some points, so any heaven run (which is single-core limited) is not properly indicative of a full-power 1080.

And this is what a stock 1080N should be getting in heaven (at least in DirectX 11, I don't know about that OpenGL run). You can see my second card is completely idle.

 

Heaven is getting on and 1080p is getting on the low side for the 1080 to be driving.

 

Great information as always!