Description
Dot products and similarity metrics can be computed quickly with SIMD-backed routines. Data and ML developers use numkong for vector search, ranking, embeddings, and numerical workloads. Numerical precision and CPU feature support should be tested on target machines.