Files

Mateusz Tabaka 063ff5a5c5 [GPU] Insert transpose for gemm input if its trans attr is set (#8460 )

If gemm input dimensions are not multiple of 16 and any of
transpose_a/transpose_b attribute is set - cldnn picks
'gemm_ref' kernel in favor of faster 'gemm_tiled_opt'.
By emplacing explicit permute operation on the gemm input that it requires,
we make cldnn to pick 'gemm_tiled_opt', which in result improves
performance.
For some input shapes, transpose(s) + gemm_tiled_opt can be slower than
just gemm_ref. Based on benchmarks - the cutoff point was set for inputs shapes > (64, 64).

Ticket: 67271

2022-11-02 10:42:52 +01:00

auto

Bell/fix priority (#13667 )

2022-10-28 04:58:25 +03:00

auto_batch

support cache dir in multi/auto/batch (#13507 )

2022-10-19 10:15:03 +08:00

hetero

Warning as error for Windows (#13291 )

2022-10-06 13:44:21 +04:00

intel_cpu

Enable call set_device_name (#13340 )