Skip to content

Implement SIMD where applicableΒ #49

Open
@olback

Description

@olback

What got me looking into this was the somewhat slow key generation and the related issue #29. A solution or rather, improvement would be to implement SIMD, Single Instruction Multiple Data. I found a paper by freescale semiconductor on this topic here: http://application-notes.digchip.com/314/314-66328.pdf.

Since all processors do not support the AVX/SSE/SIMD family of instructions, this would have to be implemented under a feature flag or as in the case of aes_gcm, the feature is enabled when the compiler is passed these flags:

RUSTFLAGS="-Ctarget-cpu=sandybridge -Ctarget-feature=+aes,+sse2,+sse4.1,+ssse3"

Update: What takes time during key generation is finding big primes, and that is done here: https://github.com/dignifiedquire/num-bigint/blob/master/src/bigrand.rs#L324-L371

I might take a deeper look at this when my exams are over.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions