mirror of
https://github.com/saymrwulf/risc0-curve25519-dalek-source.git
synced 2026-09-04 20:03:40 +00:00
Fix AVX2 docs formatting, remove obsolete AVX512 note
This commit is contained in:
parent
d67e895619
commit
844da9712b
1 changed files with 2 additions and 8 deletions
|
|
@ -463,18 +463,12 @@
|
|||
//! `ymm16..ymm31` registers from AVX512VL), and of approximately 1.0x
|
||||
//! for Ryzen (which implements AVX2 at half rate).
|
||||
//!
|
||||
//! When used for variable-time double-base scalar multiplication \\( aA
|
||||
//! + bB \\) for fixed \\(B\\) (as in, e.g., signature verification),
|
||||
//! When used for variable-time double-base scalar multiplication
|
||||
//! \\( aA + bB \\) for fixed \\(B\\) (as in, e.g., signature verification),
|
||||
//! this strategy provides a 1.4x speedup on Skylake-X over the same
|
||||
//! operation as implemented in `ed25519-donna`, the fastest
|
||||
//! production-quality Ed25519 implementation.
|
||||
//!
|
||||
//! (Note: since testing this, the experimental `llvm50` Rust branch
|
||||
//! used to compile the experimental `stdsimd` intrinsics have fallen
|
||||
//! out of sync and it is no longer possible to compile for
|
||||
//! `skylake-avx512`. This is why all of this branch is part of the
|
||||
//! `yolocrypto` feature, pending upstream work.)
|
||||
//!
|
||||
//! [sandy2x]: https://eprint.iacr.org/2015/943.pdf
|
||||
//! [avx2trac]: https://trac.torproject.org/projects/tor/ticket/8897#comment:28
|
||||
//! [hwcd08]: https://www.iacr.org/archive/asiacrypt2008/53500329/53500329.pdf
|
||||
|
|
|
|||
Loading…
Reference in a new issue