Set hwaccel to cuda#5
Closed
utkustnr wants to merge 1 commit intoEposVox:masterfrom
utkustnr:utkuustuner-patch-1
Closed
Set hwaccel to cuda#5utkustnr wants to merge 1 commit intoEposVox:masterfrom utkustnr:utkuustuner-patch-1
utkustnr wants to merge 1 commit intoEposVox:masterfrom
utkustnr:utkuustuner-patch-1
Conversation
You can just change auto to cuda and "-pix_fmt" will still be usable. However it won't give the same effect. "-hwaccel_output_format cuda" is required from what I read and that can't be used with "-pix_fmt". Maybe they can be used together but I couldn't figure it out. As for the "-bf 0", I was getting "No decoder surfaces left" error when encoding starts. Some reddit thread said setting b-frames to 0 instead of default -1 (auto) OR using "-extra_hw_frames 2" (keep increasing 2 until error goes away) fixes it so I chose b-frame one. After doing these changes my speed went from 1.5-2x to 7.5-8x and gpu usage increased dramatically compared to original batch. Also used #4 to fix naming.
Author
|
I didn't mean to close this pr. Apparently I don't know how to use github. Anyway, if someone wants to speed up their process feel free to do these changes. Update: Just found out that increase in speed comes with increased file size. I still think it's an improvement nonetheless. I processed the same folder with and without cuda and they had 9 gb difference (both was around 1/3-1/4 of original size just to be clear). That being said that 9 gb costed just as much hours to process. Anyone who wants to use this version should keep this in mind. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
You can just change
autotocudaand-pix_fmtwill still be usable. However it won't give the same effect.-hwaccel_output_format cudais required from what I read and that can't be used with-pix_fmt. Maybe they can be used together but I couldn't figure it out. As for the-bf 0, I was getting "No decoder surfaces left" error when encoding starts. Some reddit thread said setting b-frames to 0 instead of default -1 (auto) OR using-extra_hw_frames 2(keep increasing 2 until error goes away) fixes it so I chose b-frame one. After doing these changes my speed went from 1.5-2x to 7.5-8x and gpu usage increased dramatically compared to original batch. Also used #4 to fix naming.