pinkforest(she/her)
pinkforest(she/her)
PR's up: - #467
Ref issues I raised earlier: - https://github.com/dalek-cryptography/curve25519-dalek/issues/635 I've pointed the AVX512 probably could be default-removed because it cannot be tested and maintained reliably. IMO it probably should be made opt-in...
How can we CI it? The original idea was that github used to have AVX512 capable runners at reasonable ratio and the nightly job tested it out. But now that...
fyi - we're continuing all development via in `main` - Could this be re-based to main which is basically from release/4.0 - Thanks! Also would we have any benchmarks between...
> Of course - Just to make sure: is that going to remain? I did the rebase four days ago on release/4.0. Yeah this is going to stay in `main`...
Instead of flooding some changes here is some .patches: src/backend/vector/neon/field.rs ``` diff --git a/src/backend/vector/neon/field.rs b/src/backend/vector/neon/field.rs index 5cf57ee..85630cd 100644 --- a/src/backend/vector/neon/field.rs +++ b/src/backend/vector/neon/field.rs @@ -21,7 +21,7 @@ //! arm instructions. use...
started [a benchmark repo](https://github.com/RustCrypto/benchmarks) here - I ran some benchmarks earlier (will re-run now) on [Mac M1 Max Aarch64](https://github.com/RustCrypto/benchmarks/tree/main/curve25519/nightly-aarch64-apple-darwin) comparing between backends FWIW I'm going to re-organise it to cover...
Awesome - For CI there is ARM already for cross - [.github/workflows/cross.yml#L19](https://github.com/dalek-cryptography/curve25519-dalek/blob/main/.github/workflows/cross.yml#L19) We also have `simd/x86_64-AVX` related tests in [.github/worksflows/curve25519-dalek.yml#L118](https://github.com/dalek-cryptography/curve25519-dalek/blob/main/.github/workflows/curve25519-dalek.yml#L118) But I think we could split backend tests into a...
I am seeing (or I think I am) spam triggered as the ones I tested with (see your internal system) generally I get auto-reply if I do them manually but...