Low overlay temps with overheating GPU ?

[Bretchen]

Conundrum: I have 2 identical brand-new GTX1070 graphics cards. Both work well during normal gameplay, but one of them crashes after just 10-20 minutes of hi-intense 1080P gaming. The screen suddenly goes black and says "No Signal", while the GPU fans surge to 100% and beyond. The game and OS audio continues for a short while until the system hangs and I have to reboot. To me, it seemed a lot like the one GPU was overheating while the other was happy gaming forever. The weird thing is they both report low/cool temps on the overlay.

Changing PSU, RAM, SSD, and eve re-installing windows 10 had no effect.

I took the bad GPU apart and, while I'm no expert, found the heatsink to come apart far too easily (literally just fell off) and it broke one of the fan blades (DOH! my first klumsy-oopsie ever I swear) and the thermal paste feeling rather dry. If this was a case of bad factory thermal paste, why did the overlay report cool temps right up until the GPU crashed ? Could there be another issue I haven't considered ?

[emosun]

3 minutes ago, Bretchen said:

The weird thing is they both report low/cool temps on the overlay

Then it wasn't overheating and you essentially broke the gpu for no reason. Could have very easily been returned prior to destroying it

[Bretchen]

Yeah I know it was dumb of me but the RMA would have taken up to 6 months with shipping billed to me, so I got impatient and creative.

Anyway, I have a replacement fan blade enroute for just $20 and I'll put good thermal paste on when I put it back together. I was hoping that someone here could give me some tips on what else could be wrong with this particular card ?

[KarathKasun]

Defective GPU, defective memory, defective part in VRM, or any combo of the above.

 

Post application long shelf life TIM is dry-ish/wax-like.

[emosun]

13 minutes ago, Bretchen said:

Changing PSU, RAM, SSD, and eve re-installing windows 10 had no effect.

We would have told you to rma it at this point.

if you ever get the gpu back together and replace the paste and that doesn't fix it then you may be out of options.

[Jurrunio]

if the cooler juse fall off, it could be the screws either breaking (maybe low quality causing them to shatter?) or not even there.

 

As for the reason of shut down, it could be other components being overloaded and/or overheated. You cant monitor those in software, but some cards do have protections built into them for these scenarios.

 

For the paste, that's normal. These paste only care about lasting long and being cheap, one way to achieve this is to use solvent that evapourates more slowly. That usually means the paste is less fluid and more solid.

[KarathKasun]

1 minute ago, Jurrunio said:

if the cooler juse fall off, it could be the screws either breaking (maybe low quality causing them to shatter?) or not even there.

 

As for the reason of shut down, it could be other components being overloaded and/or overheated. You cant monitor those in software, but some cards do have protections built into them for these scenarios.

 

For the paste, that's normal. These paste only care about lasting long and being cheap, one way to achieve this is to use solvent that evapourates more slowly. That usually means the paste is less fluid and more solid.

I think OP was expecting the heatsink to be somewhat stuck to the card after removing the screws.

[RobbinM]

Do the fans spin up before or after it hangs?