diff options
author | Sang-Hoon Park <sang-hoon.park@arm.com> | 2021-05-18 10:46:00 +0100 |
---|---|---|
committer | Pablo Marquez Tello <pablo.tello@arm.com> | 2021-05-27 16:33:44 +0000 |
commit | b3be45759bdd0749ae3a16fe470820f0d9830ea9 (patch) | |
tree | 10bb8c1c0a049a23c00781c64e993f1b197c0d05 /arm_compute/runtime/NEON/functions | |
parent | bc91297c865808ed2c321febc405179f63195ff8 (diff) | |
download | ComputeLibrary-b3be45759bdd0749ae3a16fe470820f0d9830ea9.tar.gz |
Implement memory injection in CpuDirectGemmConv2d
The following operators are now stateless by implementing
memory injection.
- CpuDirectGemmConv2d
- CpuGemmAssemblyDispatch
A test case is added to test if CpuDirectGemmConv2d can
run on different group of tensors with a single configure.
Resolves: COMPMID-4506
Change-Id: I48f44ed41236ca7e18da2de07bdbacc9007a3c5e
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5718
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com>
Diffstat (limited to 'arm_compute/runtime/NEON/functions')
-rw-r--r-- | arm_compute/runtime/NEON/functions/NEGEMM.h | 3 | ||||
-rw-r--r-- | arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h | 3 |
2 files changed, 4 insertions, 2 deletions
diff --git a/arm_compute/runtime/NEON/functions/NEGEMM.h b/arm_compute/runtime/NEON/functions/NEGEMM.h index 6fa30bd545..a5d6bb6534 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMM.h +++ b/arm_compute/runtime/NEON/functions/NEGEMM.h @@ -146,7 +146,8 @@ private: bool _reshape_b_only_on_first_run; bool _is_prepared; - ITensorPack _asm_glue_tensors{}; + struct AsmGlueTensors; + std::unique_ptr<AsmGlueTensors> _asm_glue_tensors; }; } // namespace arm_compute #endif /*ARM_COMPUTE_NEGEMM_H */ diff --git a/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h b/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h index dc9783f9eb..ff50d6dbf7 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h @@ -171,7 +171,8 @@ private: bool _run_activation; bool _flip_signedness; - ITensorPack _asm_glue_tensors{}; + struct AsmGlueTensors; + std::unique_ptr<AsmGlueTensors> _asm_glue_tensors; }; } // namespace arm_compute #endif /*ARM_COMPUTE_NEGEMMLOWPMATRIXMULTIPLYCORE_H */ |