High-performance audio and imaging, AI/ML with auto-vectorization

HiFi 5s greatly improves TTM by allowing the compiler to automatically vectorize data processing, eliminating the need for time-consuming hand-optimization, opening up the DSP to every embedded programmer. It brings double-precision MACs and ISA to accelerate modern audio algorithms requiring high range and precision. Lightweight imaging and vision algorithms are accelerated with special ISA.

Auto-vectorization support unburdens the programmer from expert-level hand-optimization, opening up the HiFi 5s DSP to a vast pool of non-DSP programmers, lowering development cost, and accelerating TTM. The expanded ISA and MACs allow modern audio algorithms to run more efficiently OOB by utilizing the added double-precision features and run lightweight vision cases along with audio.