Skip to content

polyval: implement Karatsuba multiplication for arm64#181

Merged
tarcieri merged 2 commits intoRustCrypto:masterfrom
ericlagergren:karatsuba
Jun 23, 2023
Merged

polyval: implement Karatsuba multiplication for arm64#181
tarcieri merged 2 commits intoRustCrypto:masterfrom
ericlagergren:karatsuba

Conversation

@ericlagergren
Copy link
Contributor

@ericlagergren ericlagergren commented Jun 22, 2023

Improves performance by ~200 MB/s on a 2020 M1.

@ericlagergren
polyval: implement Karatsuba multiplication for arm64
704f074 
Improves performance by ~200 MB/s on a 2020 M1.

Signed-off-by: Eric Lagergren <eric@ericlagergren.com>
@ericlagergren
Copy link
Contributor Author

ericlagergren commented Jun 22, 2023

The code is taken from https://github.com/ericlagergren/polyval-rs/tree/dev, which also has "wide" implementations (8 blocks at a time), which has significantly better performance (~0.17 cycles per byte instead of ~2).

@ericlagergren
Copy link
Contributor Author

ericlagergren commented Jun 22, 2023

I also have an x86 version I can submit as well if you'd like.

tarcieri
tarcieri reviewed Jun 23, 2023
View reviewed changes
@tarcieri
Copy link
Member

tarcieri commented Jun 23, 2023

Parallel and x86 versions would be appreciated, although perhaps as separate PRs to ease reviewability

@ericlagergren
use #[inline] and #[target_feature]
9a56be1 
Signed-off-by: Eric Lagergren <eric@ericlagergren.com>
tarcieri
tarcieri approved these changes Jun 23, 2023
View reviewed changes
Copy link
Member

@tarcieri tarcieri left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Tested locally on an M2 Max, where I observed the reported speedups.

Percentage-wise it's about a 17% speedup.

@tarcieri tarcieri merged commit 973fe29 into RustCrypto:master Jun 23, 2023
@ericlagergren
Copy link
Contributor Author

ericlagergren commented Jun 24, 2023

Parallel and x86 versions would be appreciated, although perhaps as separate PRs to ease reviewability

Actually, your x86 implementation only uses 3 clmul instructions, so I don't think the serial version can be improved much.

I'll look at adding parallel implementations. Off hand, do you know if the current API supports it? The input probably needs to be in one contiguous buffer. (Maybe not?) But that's the common case, at least for stuff like non-interleaved AES-GCM-SIV or HCTR2.

@tarcieri
Copy link
Member

tarcieri commented Jun 24, 2023 •
edited
Loading

Take a look at poly1305 for an example of a parallel multi-block backend (AVX2): https://github.com/RustCrypto/universal-hashes/blob/0054b30/poly1305/src/backend/avx2.rs#L188-L198

baloo added a commit to baloo/universal-hashes that referenced this pull request Mar 3, 2024
@baloo
polyval v0.6.2
0b1de2e 
Added
- add `new_with_init_block` (RustCrypto#195)

Changed
- implement Karatsuba multiplication for arm64 (RustCrypto#181)
@baloo baloo mentioned this pull request Mar 3, 2024
Merged
tarcieri pushed a commit that referenced this pull request Mar 3, 2024
@baloo
polyval v0.6.2 (#197)
4e7d04f 
Added
- add `new_with_init_block` (#195)

Changed
- implement Karatsuba multiplication for arm64 (#181)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@tarcieri tarcieri tarcieri approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@ericlagergren @tarcieri