diff options
| author | Giorgio Arena <giorgio.arena@arm.com> | 2018-04-23 12:01:33 +0100 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:49:39 +0000 |
| commit | fda901f0485371e8b6a807c8dd9614560a924793 (patch) | |
| tree | d41f01633fb3d7ecb39472d1c897f9ff1bfb84bd /src/core/CL/cl_kernels/depthwise_convolution_quantized.cl | |
| parent | f485a100e3f11911d25a16b4ccc286c6c0816061 (diff) | |
| download | ComputeLibrary-fda901f0485371e8b6a807c8dd9614560a924793.tar.gz | |
COMPMID-1078 Fix CL DepthwiseConvolutionLayer QASYMM8 failing validation
Change-Id: Id540490e5faf11c466ff039a20880eeedd6e5ec7
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128612
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-by: Isabella Gottardi <isabella.gottardi@arm.com>
Diffstat (limited to 'src/core/CL/cl_kernels/depthwise_convolution_quantized.cl')
| -rw-r--r-- | src/core/CL/cl_kernels/depthwise_convolution_quantized.cl | 5 |
1 files changed, 2 insertions, 3 deletions
diff --git a/src/core/CL/cl_kernels/depthwise_convolution_quantized.cl b/src/core/CL/cl_kernels/depthwise_convolution_quantized.cl index 635bc9d50b..a0c0a8b1fb 100644 --- a/src/core/CL/cl_kernels/depthwise_convolution_quantized.cl +++ b/src/core/CL/cl_kernels/depthwise_convolution_quantized.cl @@ -333,7 +333,7 @@ __kernel void depthwise_convolution_3x3_quantized_nhwc_stride1( const int src_offs2 = src_offs1 + src_stride_z; const int cond_top = z - CONV_PAD_TOP < 0; - const int cond_bottom = z * (src_step_z / src_stride_z) + 2 >= SRC_DEPTH; + const int cond_bottom = z * (src_step_z / src_stride_z) + 2 > SRC_DEPTH; __global uchar *src_addr0 = first_elem + select(src_offs0, pad_offs, cond_top); __global uchar *src_addr1 = first_elem + src_offs1; @@ -580,8 +580,7 @@ __kernel void depthwise_convolution_3x3_quantized_nhwc_stride2( const int src_offs2 = src_offs1 + src_stride_z; const int cond_top = z - CONV_PAD_TOP < 0; - const int cond_bottom = z * (src_step_z / src_stride_z) + 2 >= SRC_DEPTH; - ; + const int cond_bottom = z * (src_step_z / src_stride_z) + 2 > SRC_DEPTH; __global uchar *src_addr0 = first_elem + select(src_offs0, pad_offs, cond_top); __global uchar *src_addr1 = first_elem + src_offs1; |