diff options
| author | Sheri Zhang <sheri.zhang@arm.com> | 2021-04-22 14:41:12 +0100 |
|---|---|---|
| committer | Sheri Zhang <sheri.zhang@arm.com> | 2021-04-28 12:52:32 +0000 |
| commit | a47dcc229d912d4e4bb5afa37220d20451f243a7 (patch) | |
| tree | f8b296701fbdebfc7d29abc09144c49619bcca1c /arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h | |
| parent | 2b7fee089c76226bfafcae77ba49f1eddb1e01da (diff) | |
| download | ComputeLibrary-a47dcc229d912d4e4bb5afa37220d20451f243a7.tar.gz | |
Update operator list documentation
All the common information for the operators are stored in OperatorList.h.
All data type and data layout information for the operators are store in the function header files.
Partially resolve: COMPMID-4199
Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I272948cfb3f84e42232a82dd84c0158d84642099
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5511
Reviewed-by: SiCong Li <sicong.li@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h')
| -rw-r--r-- | arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h | 10 |
1 files changed, 10 insertions, 0 deletions
diff --git a/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h index fc4017e635..82cabed6c9 100644 --- a/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h @@ -58,6 +58,16 @@ public: ~NEDirectConvolutionLayer(); /** Set the input, weights, biases and output tensors. * + * Valid data layouts: + * - NHWC + * - NCHW + * + * Valid data type configurations: + * |src0 |src1 |src2 |dst | + * |:------|:------|:------|:------| + * |F16 |F16 |F16 |F16 | + * |F32 |F32 |F32 |F32 | + * * @note: DirectConvolution only works in the following configurations: * 1x1 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = F16/F32 * 3x3 convolution with stride_x = 1/2/3, stride_y = 1/2/3 data type = F16/F32 |