diff options
author | Aleksandr Nikolaev <aleksandr.nikolaev@arm.com> | 2020-06-25 12:25:52 +0100 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2020-07-16 16:02:50 +0000 |
commit | a084b46835d20fdfe6e590b91b7ca64fba3542df (patch) | |
tree | e4abe49cc10f3406dcaad9d3dfbbc22cc54476a6 /src/core/NEON/kernels/assembly | |
parent | f738fe6b6e059916294c48b942952c261569df18 (diff) | |
download | ComputeLibrary-a084b46835d20fdfe6e590b91b7ca64fba3542df.tar.gz |
[ONCPUML-97]: Implement "int8" support for 2D decomposition at high core counts
Interleaved2d functionality was extended to uint8 and int8 kernels.
Change-Id: If78facbce56e9ec7b2f4c23436af0bd5db7f7b69
Signed-off-by: Aleksandr Nikolaev <aleksandr.nikolaev@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3467
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/assembly')
-rw-r--r-- | src/core/NEON/kernels/assembly/arm_gemm.hpp | 1 | ||||
-rw-r--r-- | src/core/NEON/kernels/assembly/gemm_common.hpp | 2 |
2 files changed, 2 insertions, 1 deletions
diff --git a/src/core/NEON/kernels/assembly/arm_gemm.hpp b/src/core/NEON/kernels/assembly/arm_gemm.hpp index 1a613e2160..e3e15478fb 100644 --- a/src/core/NEON/kernels/assembly/arm_gemm.hpp +++ b/src/core/NEON/kernels/assembly/arm_gemm.hpp @@ -42,6 +42,7 @@ enum class GemmMethod GEMM_INTERLEAVED, GEMM_INTERLEAVED_2D, QUANTIZE_WRAPPER, + QUANTIZE_WRAPPER_2D, GEMM_HYBRID_QUANTIZED }; diff --git a/src/core/NEON/kernels/assembly/gemm_common.hpp b/src/core/NEON/kernels/assembly/gemm_common.hpp index 8feecf440e..e9e56842c7 100644 --- a/src/core/NEON/kernels/assembly/gemm_common.hpp +++ b/src/core/NEON/kernels/assembly/gemm_common.hpp @@ -77,7 +77,7 @@ public: return false; } - /** Main execute member fucntion + /** Main execute member function * @param [in] work_range specifies the range of work we want to be computed, total range defined by get_window_size() * @param [in] thread_locator where are we inside of the thread space * @naram [in] threadid a unique threadid |