[SOLVED] Short Frame Rate Drops

Oct 11, 2019
3
0
10
Hi all,

I am having an issue with strange frame rate drops. System specs first:

AMD 2700x Ryzen (not overclocked)
Be Quiet Silent Loop
MSI x470 Gaming Plus
32 GB Crucial Ballistix (XMPed to 3200MHz)
EVGA 1080Ti Hybrid SC2
Be Quiet DarkForce 850W
Samsung 870 Evo M.2
Windows 10 Professional


About 4 months ago, I moved the system to a different case, that is when the troubles started. The only thing that changed was, that I reinstalled windows, because a year earlier, my Mainboard shorted out and I switched from Gigabyte to MSI and I had all the GB junk still on it.
While gaming, I get a second or two in which the frame rate just tanks (60 down to 20), after that short span it goes back up again.

Things I tried:
  • turned off all Windows 10 Spyware
  • turned off Icue, EVGA Precision, Steam overlay etc.
  • reinstalled drivers

Now for the weird part, which blows most common forum answers out of the water:

- it is not thermal throttling: GPU, VRAM etc never goes past 62 Cs
- CPU is rock solid at 55 Cs

GPU Z shows a significant drop in Clock Speed (down to the hundreds) when it happens

And here is the kicker:
It happens across all games. Yesterday, I have tried the following:
Run a game till it happens, exit, let heat bleed out of the system (to make sure it is not thermals). Run different game. Same drops. Right away.

It almost feels like a memory leak, because only a reboot can fix it.

Appreciate your comments!

Marko
 
Solution
Ok well to see if it is a memory leak you can open task manager during gameplay and see if the value just increases to infinity however the behavior described does not match a memory leak which would be indicated by a gradual decrease in performance followed by a crash.

If the GPU is dropping MHz that dramatically then there are 2 possibilities.

1: it is thermal throttling but not reporting temps correctly

2: There is a power delivery issue to the card (you have a good psu so not confident on that)

Large performance dips during gameplay as you describe are often caused by loading data from disk so i would run a disk repair utility and see if anything comes of it.
Ok well to see if it is a memory leak you can open task manager during gameplay and see if the value just increases to infinity however the behavior described does not match a memory leak which would be indicated by a gradual decrease in performance followed by a crash.

If the GPU is dropping MHz that dramatically then there are 2 possibilities.

1: it is thermal throttling but not reporting temps correctly

2: There is a power delivery issue to the card (you have a good psu so not confident on that)

Large performance dips during gameplay as you describe are often caused by loading data from disk so i would run a disk repair utility and see if anything comes of it.
 
Solution
Thanks LORD,
I stuck two extra thermal sensors to the back pf the graphics card. Those temps are consistent with what I see in Precision.

I can no longer reproduce the Mhz drops in GPUZ, however when the error happens, power consumption seems to drop from 100% to about 70%

Is the PSU failing?