Render speed improvement, using all CPU cores
I think following steps would improve render speed:
- Read how many CPU’s the PC has (physic and logical CPU’s)
- Split the render length into amount of CPU’s with virtual zones
- Create a render script per virtual zone
- Render each script in parallel on one CPU and create temporary rendered files
- Merge the temporary files with
ffmpeg -f concat
(https://trac.ffmpeg.org/wiki/Concatenate#samecodec)
Pros
- We don’t care about non threat save effects
- We use all CPU cores available
- We could remove
Parallel Processing
Cons
- Below a certain render length splitting has no benefit
To be checked
- How
Guide Multi-Export
should be rendered in this scenario? - Check issue #349