Downfall in GPU Performance

Hello it's Shivam and in the last few weeks I've noticed there's a major fall in performance of my system..now it can't even hold Stable 60fps in Tekken 7 and in GTA V where i earlier used to enable Long Shadows and other stuff along with Nvidia TXAA and still get almost stable 58-65fps but now even in the Daytime with everthing turned off i barely hit stable 60fps.. also performance degrade in Fortnite as well as in CS:Go where i get unstable fps of around 140fps and it sometimes falls down to even 55-80fps when playing with Bots..I've posted the Valley Benchmark results..I'm using a Clevo P750DM2-G an using the latest Nvidia Driver 411.63.. Also Afterburner has been set to default..

 

Overheating?

 

It's usually at 90C the same as before however one guy did helped me in undervolting the GPU..That improved the situation marginally..But Just now turned the Nvidia Share recording setting to low.. It seemed to have helped a bit

 

Except 90c is way to high.

Go repaste, delidd, undervolt etc. to fix your temps, otherwise you will one day have a black screen greeting you.

 

Looks like 91C is the GPU's thermal throttle point

It'll drop clocks hard to stay at 90C. Undervolting would have led to slightly less worse clocks, but until you fix the cooling (or turn up the fans to maximum if you haven't?) so its under 90C your performance will be poor.

 

What happens if you force max fan? What about cleaning out the heatsink and doing a re-paste?

 

Buy some contact pressure paper (it's worth it).
https://www.amazon.com/Innovation-C...573550&sr=8-5&keywords=IC+innovations+thermal

This will tell you if you have good contact on the GPU with the thermal pads or if you need to use thinner pads. You also should find out which pads are used by default.
Then repaste the GPU. You should try Coolermaster Gel Maker Nano, which is a very durable paste. Yes that will cost you money but it's worth it. One sheet should do four to six GPU pressure tests.

There have been many cases on some of the badly engineered older Clevo heatsinks where warping or too thick thermal pads prevented full contact with the CPU or GPU from the heatsink. The easiest way to find out what is going on is to test with pressure paper.

Might also want to grab a sheet of 1.0mm and 0.5mm Arctic Thermal Pads, of size 145mm * 145mm. They are cheap and it's always good to have pads sitting around. You never know when you may need them. Some things are good to have extras available.

 

As your (too high) temperatures are "the same as before", we should also rule out any software issue: have you updated windows recently? You say you're using the latest nvidia driver. Have the problems began after installing the new driver? Have you checked the nvidia control panel if settings have remained the same?

Also, have you checked your CPU temperatures? Perhaps it's throttling as well.

 

Software should not be able to drive the temps like that unless something malicious is driving load to high levels in the background.

 

GPU might underclock to maintain temperatures under 90*C.

What thermal paste do you use?

 

You are right, but he stated that the was already hitting 90°C for some time before the performance drop, if I interpret correctly. Either the gpu is degrading because of long-term (?) overheating (don't think so, throttling should keep it safe), or there is another problem on top of the obvious fact that his chips runs too hot.

 

Uggh I recognize this and going up to 91°C as well on my laptop. Recently I tried to install Overwatch on Linux to see if i could fix the gpu crashes on Windows, but that didn't do it. The weird part is that in linux i get all the stats about the card, nothing seems to be broken, yet the system crashes as soon as one game goes on (or even before).

 

You really shouldnt go above 85c on any kind of notebook.

 

Chips don't degrade in performance, they are digital in nature, it would simply crash if the transistors had experienced too much electron migration and raised the required voltage.