CPU bottleneck with NVENC

[MrFurball]

I streamed Spider-Man and Horizon Zero Dawn this week and even though I'm using the Nvenc encoder and OBS was utilizing less than 10% of the CPU, I was getting way lower FPS in these games than I should. I noticed that the GPU was being missvely under utilized.

It concerns me a lot since it's a big drop too. In livestream I was getting 45 fps in Spider-Man and when I tested only playing the game, it was around 100fps.

Does anyone know of any settings that could be causing such a big hit on the CPU?

My Setup:

Ryzen 7 5700x
GTX 1060
16GB RAM
Windows 10

Settings:
NVENC encoder
Preset Quality
Profile High
6000 CBR

In my scene, I only had Streamelements overlay, game capture window and facecam. I have some more layers but they were hidden.

 

[PaiSand]

Please restart OBS and perform a test stream/recording for more than 30 seconds doing as you normally do, including playing games, full camera or whatever you normally do.
Now look inside Help menu. Upload the current log file and paste the url to the log in here. Click on the Analize button to start troubleshooting common issues.

 

[MrFurball]

Please restart OBS and perform a test stream/recording for more than 30 seconds doing as you normally do, including playing games, full camera or whatever you normally do.
Now look inside Help menu. Upload the current log file and paste the url to the log in here. Click on the Analize button to start troubleshooting common issues.

https://obsproject.com/logs/acGK8W6bM-QwLIEU

 

[PaiSand]

You're using SE.live. Please contact them for support.
Alternatively, uninstall it. Then uninstall and reinstall OBS in order to fix the changes made by SE.live. Now please do a test stream/recording of more than 30 seconds and post the log file here.

 

[MrFurball]

You're using SE.live. Please contact them for support.
Alternatively, uninstall it. Then uninstall and reinstall OBS in order to fix the changes made by SE.live. Now please do a test stream/recording of more than 30 seconds and post the log file here.

Oh, sorry. Didn't know SE could change so much that it could be a problem in on it self, but I analyzed the logs like you suggested and the critical issue was the Hardware-accelerated GPU Scheduler and disabling it solved the problem.