This appeared among the daily updates of Arch Linux: aom-v1.0.0-errata1. More news is in: https://aomedia.googlesource.com/aom/+/refs/tags/v1.0.0-errata1 Looks like they've disabled multithreading. Is it important for CinGG?
Andrea, thanks for the notify. Looks like they've disabled multithreading. Is it important for CinGG?
AOM is very slow, because I think that the version GG added was not
multi-threaded. But maybe since then multi-threading was added to a newer version and then removed? Anyway, ffmpeg is supposedly adding Dav1d and it is faster according to Andrew earlier email. But it is not yet in a released ffmpeg (I think it will be in the upcoming 4.2 version). This reply is mostly speculative and may not even be correct! Phyllis
In the changelog of ffmpeg 4.2 it was announced that Dav1d will be integrated. As far as I know it is faster than AOM. Sam On 26.07.19 14:58, Phyllis Smith wrote:
Andrea, thanks for the notify.
Looks like they've disabled multithreading. Is it important for CinGG?
AOM is very slow, because I think that the version GG added was not multi-threaded. But maybe since then multi-threading was added to a newer version and then removed? Anyway, ffmpeg is supposedly adding Dav1d and it is faster according to Andrew earlier email. But it is not yet in a released ffmpeg (I think it will be in the upcoming 4.2 version). This reply is mostly speculative and may not even be correct! Phyllis
In the changelog of ffmpeg 4.2 it was announced that Dav1d will be integrated. As far as I know it is faster than AOM.
I agree that Dav1d is a good thing to implement but, as has been said in other messages, it's just a decoder. So aom will continue to be necessary for the encoding. I propose to implement it with active multithreading and not as it is by default, i.e. single thread. Unless the reason they're back in the single thread is really important (I don't know it).
participants (3)
-
Andrea paz -
Phyllis Smith -
Sam