Cpu/Gpu Usage Difference between Obs and Intel's built-in recording software

Bobdan092

New Member
I have had this issue for very long and I am finally decide to get some sort of explanation on this topic. When I record with OBS, using h264_qsv (quicksync h.264) at 1080p60 with no downscale @37200kbps bitrate, my cpu usage as well as gpu 3d usage shoots through the roof. I have noticed that even though I am using quicksync h.264 (which uses built in hardware for video processing), the 3d section of the integrated graphics sits above 60% usage while the video decode and video processing sections are barely working (around 20% and 0% respectively.) Meanwhile when I record using Intel's built in recording software, the 3d utilization drops as well as the cpu usage, and the video processing and video decode sections kick in. I don't know if there is a difference in the encoder, but I believe that Intel would use their quicksync encoder to record since it is their creation. Any explanation would be helpful, and any solution will be greatly appreciated. I will attach video clips recorded from Obs and Intel to show you the difference.

P.s. I am also not very educated in this topic so please correct me if there is anything wrong with what I have said above.

Obs Recording Link: https://streamable.com/3nfyy4
Intel Recording Link: https://streamable.com/q0yvrp
 

koala

Active Member
OBS does more than just recording. The Intel recording function just grabs the screen frame buffer and sends it to the encoder.
OBS does compositing in addition. This is the scaling, filtering, merging and overlaying of multiple sources to one video. Even if you have only one source, this compositing is always an additional step, which takes place on the GPU with a hidden frame buffer. Since the computing power of the iGPU is not very high, this produces a non-negligible amount of additional GPU usage if compared with the load of a software that doesn't do that.
 

HeitorMR

New Member
OBS does more than just recording. The Intel recording function just grabs the screen frame buffer and sends it to the encoder.
OBS does compositing in addition. This is the scaling, filtering, merging and overlaying of multiple sources to one video. Even if you have only one source, this compositing is always an additional step, which takes place on the GPU with a hidden frame buffer. Since the computing power of the iGPU is not very high, this produces a non-negligible amount of additional GPU usage if compared with the load of a software that doesn't do that.
So that means OBS Studio will never use video processing? Why it doesn't use video processing for rendering scenes?
 
Top