diff options
| author | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-12-13 12:33:09 +0000 |
|---|---|---|
| committer | Pablo Marquez <pablo.tello@arm.com> | 2019-12-19 14:18:23 +0000 |
| commit | 14d9d986c63eb8f07acec2f075383f9f4843ae1c (patch) | |
| tree | 543b9fc7781f7e2deba0287fbacd69d27fd16840 /arm_compute/core/NEON/kernels/NEPoolingLayerKernel.h | |
| parent | d7d7e9035ca28b1b5200b20a73825397d46830fc (diff) | |
| download | ComputeLibrary-14d9d986c63eb8f07acec2f075383f9f4843ae1c.tar.gz | |
COMPMID-2819: Retain configuration step data layout to avoid side-effects.
Configuring functions serially can lead to side-effects in tensor
attributes. One of them is the data layout changing in case functions
share same IO tensors. Retain DataLayout used during configuration.
Change-Id: I17538ce08b86df6986f0fcf21fa6544fbd5bd74b
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/2470
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Diffstat (limited to 'arm_compute/core/NEON/kernels/NEPoolingLayerKernel.h')
| -rw-r--r-- | arm_compute/core/NEON/kernels/NEPoolingLayerKernel.h | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/arm_compute/core/NEON/kernels/NEPoolingLayerKernel.h b/arm_compute/core/NEON/kernels/NEPoolingLayerKernel.h index 47421eae9a..b36e10cd40 100644 --- a/arm_compute/core/NEON/kernels/NEPoolingLayerKernel.h +++ b/arm_compute/core/NEON/kernels/NEPoolingLayerKernel.h @@ -194,6 +194,7 @@ private: const ITensor *_input; ITensor *_output; PoolingLayerInfo _pool_info; + DataLayout _data_layout; unsigned int _num_elems_processed_per_iteration; BorderSize _border_size; bool _is_square; |