Skip to content

Latest commit

 

History

History
6 lines (4 loc) · 400 Bytes

README.md

File metadata and controls

6 lines (4 loc) · 400 Bytes

About MLAS

MLAS is a compute library containing processor optimized GEMM kernels and platform specific threading code.

Unit tests for MLAS

Unit tests for the SGEMM kernels are available under onnxruntime\test\mlas. These tests run over a range of inputs that then execute the various special cases for aligned and unaligned outputs. The tests have failed if any "mismatch" strings are printed.