Challenges Arising from Lack of Stable SIMD Support
Problem
Rust's stable release lacks native SIMD support, forcing developers to rely on nightly builds or multi-version strategies, increasing compile complexity, limiting performance, and requiring manual handling of platform‑specific instruction differences. Fearless SIMD aims to address these issues, yet gaps remain in functionality and cross‑platform instruction customization.
Pain points
- Lack of a stable SIMD API, requiring reliance on nightly Rust
- Need to customize instruction sequences for different target platforms to achieve optimal performance
- Existing SIMD crates have limited functionality, resulting in sub‑optimal performance
Questions to verify
- Have you already used Fearless SIMD to replace std::simd in a real project and observed improvements in build stability and performance?
- Across different CPU architectures (e.g., x86 and ARM64), do you need to customize instructions for specific data types such as NaN or zero?
- What specific missing features (e.g., support for certain vector widths or special functions) do you need in Fearless SIMD?
Evidence
quote · @匿名
This is quite amazing. Fearless SIMD is what finally allowed my FFT crate, PhastFT, to run on stable Rust [0]. PhastFT previously depended on Rust’s std::simd (as well as multiversion), which still requires nightly [1].
fact · @匿名
PhastFT previously depended on Rust’s std::simd (as well as multiversion), which still requires nightly
quote · @匿名
Have been really happy with Fearless simd, I used it to write [memchr-n][1], which searches for instances of an arbitrary set of bytes, with simd, which actually tends to [outperform][2] the rust memchr crate, even for sets of 1-3 bytes (what's supported by memchr)[1]: https://github.com/Dr-Emann/memchr_n#performance [1]: https://github.com/Dr-Emann/memchr_n#performance [2]: https://github.com/BurntSushi/memchr/pull/241
Disclaimer: this card is distilled by AI from community discussions as “pain point evidence”. It is a starting point for need analysis and validation, not a “validated market need”. AI has analytical limits; please open the original discussion above to manually verify before key decisions.