CFAVML-core v0.3.0

What's Changed

• CFAVML-NONE: Add simsimd back to distance benchmarks by @ChillFish8 in #7
• CFAVML-2: Add cmp ops (eq/neq/lt/lte/gt/gte) by @ChillFish8 in #6

Breaking changes

• Horizontal ops naming schemes changed from op_horizontal to just op to make it consistent with the existing aggregate ops like sum
• AVX512 required CPU features now include avx512bw which means first gen avx512 servers like Skylake will fallback to AVX2 on the safe apis, and require updating target features if using the unsafe APIs.

Full Changelog: cfavml-v0.2.0...cfavml-v0.3.0

CFAVML-core v0.2.0

What's Changed

• CFAVML-1: Add min max by value ops by @ChillFish8 in #3
• Fix NEON instruction routines
• Refactor safe API and exported functions to be generic rather than exporting hundreds of functions.
  • Some pre-configured generic functions still available for various archs i.e. avx2, avx512, etc...

New Contributors

Full Changelog: cfavml-v0.1.0...cfavml-v0.2.0

CFAVML-core v0.1.0

CFAVML v0.1.0

The initial release of CFAVML's core SIMD library. Full arithmetic support is implemented for all rust integer types, along with min/max and sum aggregations as well as some core distance functions like dot product, cosine and Euclidean.