Skip to content

Latest commit

 

History

History
14 lines (8 loc) · 520 Bytes

README.md

File metadata and controls

14 lines (8 loc) · 520 Bytes

Auto-Bench-for-TensorComputation-Research

[PRs Welcome]

Benchmark scripts for Tensor Compilers and Popular Frameworks

Background

Currently, there are lots of deep learning frameworks and compilers for the performance improvement.

When I want to know the specific optimization difference on a backend, it is difficult to find a tool which can do the automatic generation of tensor programs and testing to make benchmarking easier.

Our Solution