TensorFlow Runtime, TFRT, is a new runtime for TensorFlow. TFRT has some plausible designs we can spend some time on. I’d like to cover its lock-free executor, device runtime framework, and how it uses MLIR.
TensorFlow Runtime, TFRT, is a new under-development runtime of TensorFlow. It comes with many interesting designs and useful features. I’ll provide an overview of current TFRT we can see on GitHub, its lock-free graph executor, how it decouples host runtime and device runtimes, and how it use MLIR (Multi-Level Intermediate Representation) which is one of the most exciting new compiler infrastructures.
About Koan-Sin Tan
Koan-Sin Tan is an old programmer, learned to use “open source” stuff on VAX-11/780 running 4.3BSD before the term “open source” was coined. He is interested in running neural networks on edge devices the past four years. When learning TensorFlow, he sent some PRs.