aboutsummaryrefslogtreecommitdiff
path: root/src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.h
diff options
context:
space:
mode:
authorGeorgios Pinitas <georgios.pinitas@arm.com>2021-07-16 16:16:43 +0100
committerGeorgios Pinitas <georgios.pinitas@arm.com>2021-07-22 02:25:50 +0000
commit4ee8b1599dbaf7634d25607fa5ac96ba3dc6b0f2 (patch)
tree2f8362d33cdad4212f4b96995681c68184c759e1 /src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.h
parent59fd7a722e5bc7e85309d6200bc37a772721a719 (diff)
downloadComputeLibrary-4ee8b1599dbaf7634d25607fa5ac96ba3dc6b0f2.tar.gz
Update GEMM assembly kernels
- Introduce Fp32 kernels with internal calculations in Bfloat16 when fast_mode is enabled - Improve kernel selection heuristics Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I68a9e7e862b6fd2721b46e0d7cc791091c4ab279 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5965 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.h')
-rw-r--r--src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.h1
1 files changed, 1 insertions, 0 deletions
diff --git a/src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.h b/src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.h
index 355273adeb..88cfed002a 100644
--- a/src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.h
+++ b/src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.h
@@ -51,6 +51,7 @@ struct AsmGemmInfo
int64_t padding_top{ 0 };
int64_t padding_left{ 0 };
float padding_value{ 0.f };
+ bool fast_mode{ false };
};
/** Assembly kernel glue */