onnxruntime
7196d420 - Adding Transpose3d and Transpose4d special case kernels for Rocm (#5837)

Commit
5 years ago
Adding Transpose3d and Transpose4d special case kernels for Rocm (#5837) * add transpose3d; seeing memory fault on rocm3.7 * cleaned up code; commit to switch machines * tested working on gcr-openpai-35; 168 ex/sec * remove debug HCC_ENABLE_PRINTF
Author
Suffian Khan
Parents
Loading