Tensor processing primitives: a programming abstraction for efficiency and portability in deep learning workloads.
Evangelos GeorganasDhiraj D. KalamkarSasikanth AvanchaMenachem AdelmanCristina AndersonAlexander BreuerJeremy BruestleNarendra ChaudharyAbhisek KunduDenise KutnickFrank LaubVasimuddin MdSanchit MisraRamanarayan MohantyHans PabstBarukh ZivAlexander HeineckePublished in: SC (2021)