Analytics say high rendering and encoding lag

[shadis777]

OBS Analytics and log analyzation state I have GPU encoding overload and high rendering lag but the performance monitor during streaming does not show that level of utilization. At no time do I see the CPU or GPU go over 60%. I have gone through the threads on the forum and taken many of the actions suggested. So what do I do to resolve this issue?
Please ignore the multiple capture sources, I have gone through and removed the duplicate game capture sources that I had on some scenes.

https://obsproject.com/logs/aPqGmhBUaHytyBkg

 

[shadis777]

Updated log file, I tried changing the settings down from what they were in the previous and still getting the same rendering and encoding lag without real explanation since the performance monitor does not show the same utilization that the log claims.

https://obsproject.com/logs/yZGZZolZ3zQDmPBG

 

[Suslik V]

From the log:

...
09:30:04.138: ┣tick_sources: min=0 ms, median=0.043 ms, max=2146.94 ms, 99th percentile=192.321 ms

means that there is very slow OBS source in your setup.

You can make copy of your current scene collections and start to remove OBS sources from the scenes one-by-one until this "tick_sources: ... 99th percentile=" will return to sane borders (far less than 16ms for 60fps).

 

[shadis777]

Newest log file. Removed a lot of unnecessary sources and I think I got rid of the the tick_sources problem. Still having some problems but things seem smoother. Any guidance is welcome.

https://obsproject.com/logs/6zk5LsST8JdXlWrW

 

[Suslik V]

I don't know. Rendering lag means that video are busy with drawing of something. Encoding lag means that encoder preset (resolution/framerate etc) is too high for this particular task and encoder cannot complete frame encoding in time.

Attach here full log from this recording/streaming session (find it by time/date in the OBS folder that easily accessible from the main menu Help > Log Files > Show Log Files) maybe someone will find bottle neck in your setup.