chacha20: Parallelize AVX2 backend #87
Merged
Conversation
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
The AVX2 backend was previously computing two ChaCha blocks in parallel, then throwing one away. This updates the implementation to always compute two blocks in parallel when the AVX2 backend is enabled, resulting in a ~2X speedup. Unfortunately for `cipher.rs`, originally adapted from the `ctr` crate, I deleted the original parallel computation code, and in lieu of that the implementation diverges from what was originally in `ctr`. See here for a reference: https://github.com/RustCrypto/stream-ciphers/blob/907e94b/ctr/src/lib.rs#L73 Ideally we can come up with some generic counter management and buffering abstraction in the `ctr` crate which works in all cases.
This was referenced Jan 16, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
The AVX2 backend was previously computing two ChaCha blocks in parallel, then throwing one away.
This updates the implementation to always compute two blocks in parallel when the AVX2 backend is enabled, resulting in a ~2X speedup.
Unfortunately for
cipher.rs, originally adapted from thectrcrate, I deleted the original parallel computation code, and in lieu of that the implementation diverges from what was originally inctr. See here for a reference:https://github.com/RustCrypto/stream-ciphers/blob/907e94b/ctr/src/lib.rs#L73
Ideally we can come up with some generic counter management and buffering abstraction in the
ctrcrate which works in all cases.Benchmark results: