Skip to content

polyval: implement Karatsuba multiplication for arm64 - #181

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

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

Conversation

@ericlagergren

@ericlagergrenericlagergren commented Jun 22, 2023

Copy link
Copy Markdown
Contributor

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

ericlagergren commented Jun 22, 2023

Copy link
Copy Markdown
ContributorAuthor

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

ericlagergren commented Jun 22, 2023

Copy link
Copy Markdown
ContributorAuthor

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

tarcieri
tarcieri reviewed Jun 23, 2023
View reviewed changes
Comment threadpolyval/src/backend/pmull.rs Outdated
@tarcieri

tarcieri commented Jun 23, 2023

Copy link
Copy Markdown
Member

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

@tarcieritarcieri left a comment

Copy link
Copy Markdown
Member

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:masterJun 23, 2023
@ericlagergren

ericlagergren commented Jun 24, 2023

Copy link
Copy Markdown
ContributorAuthor

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

tarcieri commented Jun 24, 2023•
edited
Loading

Copy link
Copy Markdown
Member

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)
@baloobaloo 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 freeto join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@tarcieritarcieritarcieri 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