diff options
| author | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-06-27 17:00:52 +0100 |
|---|---|---|
| committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-07-26 11:55:15 +0000 |
| commit | cfa2bba98169cb5ab1945462514be1b6badf7d98 (patch) | |
| tree | 1635e6e9463e9798c7195f0aa71b5df3f2650df1 /src/runtime/NEON/functions/NEGEMM.cpp | |
| parent | f59b16f42ef68bde877b70816ffb953d64c8baa3 (diff) | |
| download | ComputeLibrary-cfa2bba98169cb5ab1945462514be1b6badf7d98.tar.gz | |
COMPMID-2178: Update GEMM assembly code.
Perform offset reduction and requantization within the assembly wrapper.
Change-Id: I5d5b3e1f6f9ef4c71805362c57f88ff199c027a3
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/1541
Comments-Addressed: Pablo Marquez <pablo.tello@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/runtime/NEON/functions/NEGEMM.cpp')
| -rw-r--r-- | src/runtime/NEON/functions/NEGEMM.cpp | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/src/runtime/NEON/functions/NEGEMM.cpp b/src/runtime/NEON/functions/NEGEMM.cpp index 2f36397c8e..37d0e09fc9 100644 --- a/src/runtime/NEON/functions/NEGEMM.cpp +++ b/src/runtime/NEON/functions/NEGEMM.cpp @@ -58,7 +58,7 @@ void NEGEMM::configure(const ITensor *a, const ITensor *b, const ITensor *c, ITe _run_vector_matrix_multiplication = a->info()->dimension(1) < 2; _original_b = b; - bool run_optimised = c == nullptr && bool(NEGEMMAssemblyDispatch::validate(a->info(), b->info(), d->info(), alpha, beta, gemm_info)); + bool run_optimised = c == nullptr && bool(NEGEMMAssemblyDispatch::validate(a->info(), b->info(), c != nullptr ? c->info() : nullptr, d->info(), alpha, beta, gemm_info)); if(run_optimised) { @@ -66,11 +66,11 @@ void NEGEMM::configure(const ITensor *a, const ITensor *b, const ITensor *c, ITe { GEMMInfo gemm_info_ntb = gemm_info; gemm_info_ntb.set_pretranpose_B(false); - _asm_glue.configure(a, b, d, alpha, beta, gemm_info_ntb); + _asm_glue.configure(a, b, c, d, alpha, beta, gemm_info_ntb); } else { - _asm_glue.configure(a, b, d, alpha, beta, gemm_info); + _asm_glue.configure(a, b, c, d, alpha, beta, gemm_info); } ARM_COMPUTE_ERROR_ON(!_asm_glue.is_configured()); } @@ -178,7 +178,7 @@ Status NEGEMM::validate(const ITensorInfo *a, const ITensorInfo *b, const ITenso } // Check if we need to run the optimized assembly kernel - const bool run_optimised = c == nullptr && bool(NEGEMMAssemblyDispatch::validate(a, b, output, alpha, beta, gemm_info)); + const bool run_optimised = c == nullptr && bool(NEGEMMAssemblyDispatch::validate(a, b, c, output, alpha, beta, gemm_info)); if(!run_optimised) { |