-
Notifications
You must be signed in to change notification settings - Fork 246
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Eliminate buffer and copy for sifting codable coefficients
For large transforms we reorder coefficients from our forward transform so that the codable coefficients are first. Before this patch, we would do this by buffering the output forward transform then writing it out in the correct order. I'm a little suprised by the lack of impact in the microbenchmark results. I expected the copy to be inefficient enough to provide a performance boost when removed. It seems that a copy in the L1 cache just doesn't have a noticable impact. In the real world, performance does seem to improve on awcy by about half a percent. https://beta.arewecompressedyet.com/?job=master-redo-70dc401df%402020-01-19T14%3A02%3A47.242Z&job=embed_fwd_reorder%402020-01-19T14%3A25%3A07.898Z
- Loading branch information
1 parent
9110e82
commit 37509b2
Showing
2 changed files
with
63 additions
and
70 deletions.
There are no files selected for viewing
This file contains 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
This file contains 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