Would GPU still function with damaged mosfets?

Hi,

I have a Clevo P157SM running a GTX1070 in Optimus mode with integrated graphics. I was running the Forza Horizon Benchmark, worked my way up slowly eventually to 4k Ultra settings and suddenly in the middle of the 4k Ultra run, the demo simply stopped and exited to Windows desktop, no notifications. It also did this a few times when playing Dota at 4k, until I switched to 1080p after which it was fine (although Dota even at 4k isn't meant to be as intensive as Forza).

I've been wondering what could cause this behavior, and I realized the thermal pads I put on my heatsink didn't actually contact some of the mosfets, and I'm worried some of them overheated. During the Forza benchmark my GPU was often at 100% load and GPU temp at 91C (max allowed by BIOS), so I'm worried the mosfets were at a ridiculous temperature.

Does anyone know if the game abruptly exiting is what would happen if this was the case? I'm thinking the GPU either is preemptively shutting off to protect the mosfet, or a mosfet is already damaged and when I push the card it triggers a failure. Just to check, I can still run non-graphics-intensive games fine, but I'm afraid to push the card to it's limits again. HWMonitor doesn't show VRM temps, I think this card doesn't have those sensors.

So, does anyone know if my card would be working at all if I have a damaged mosfet? Thanks!

EDIT: Possible important is that during the crashes I was running a +200Mhz Core and +400Mhz memory overclock, but no change to the stock voltage.

 

Gutsy, I'm surprised that you're trying to push the 1070 that far with the given heatsink. I was told to not try any overclocking with mine, the customized heatsink apparently isn't good enough to be able to handle any higher temps.

 

You're right about the heat, on these benchmarks the GPU shoots up to the max allowed temp (91C) very quickly. I don't plan on actually using the GPU at these settings in regular use, I was just flexing it out. Danishblunt on these forums said I could probably improve the temps by soldering on a copper shim onto the given heatsink, I may try that in the next few weeks. I also still haven't tried undervolting the card which might help with temps a bit as well.

 

If the card still functions then the FETs are undamaged. It is certainly important to have them all covered with a heatsink. The crash may have saved you from blowing FETs.

As for the crash, it sounds to me like you were just over-overclocked. +200MHz is a lot on the core.