OpenNMT / CTranslate2

Fast inference engine for Transformer models

google / highway

Performance-portable, length-agnostic SIMD with runtime dispatch

simd-everywhere / simde

Implementations of SIMD instruction sets for systems which don't natively support them.

DLTcollab / sse2neon

A translator from Intel SSE intrinsics to Arm/Aarch64 NEON implementation

xtensor-stack / xsimd

C++ wrappers for SIMD intrinsics and parallelized, optimized mathematical functions (SSE, AVX, AVX512, NEON, SVE, WebAssembly, VSX, RISC-V))