onnxruntime/onnxruntime/core/mlas/README.md
2018-11-19 16:48:22 -08:00

400 B

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.