diff options
| author | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2022-12-16 15:34:27 +0000 |
|---|---|---|
| committer | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2022-12-21 12:41:01 +0000 |
| commit | 85260d8c21e7209d4777150f436b336f85812dce (patch) | |
| tree | 23f361e49f1f370cd15b89d7847d9cb021f7f5f9 /src/core/CL/cl_kernels/nhwc/direct_convolution.cl | |
| parent | 24c87f098c2ebb8f629a7069d1851f2546c28e42 (diff) | |
| download | ComputeLibrary-85260d8c21e7209d4777150f436b336f85812dce.tar.gz | |
Update direct conv2d kernel in dynamic fusion
Resolves COMPMID-5780
Change-Id: I34c764cd1df652f8a938772924dc49baf6ac16db
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8825
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/CL/cl_kernels/nhwc/direct_convolution.cl')
| -rw-r--r-- | src/core/CL/cl_kernels/nhwc/direct_convolution.cl | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/src/core/CL/cl_kernels/nhwc/direct_convolution.cl b/src/core/CL/cl_kernels/nhwc/direct_convolution.cl index 8be8e00f0a..4693a1fbcd 100644 --- a/src/core/CL/cl_kernels/nhwc/direct_convolution.cl +++ b/src/core/CL/cl_kernels/nhwc/direct_convolution.cl @@ -188,6 +188,11 @@ __kernel void direct_convolution_nhwc( a[i].v = ZERO_VALUE; }) + LOOP_UNROLLING(int, i, 0, 1, N0, + { + b[i].v = ZERO_VALUE; + }) + // Load tile from the src tensor T_LOAD2D_INDIRECT(SRC_DATA_TYPE, M0, K0, SRC_TENSOR_TYPE, src, ck, src_stride_y, my, a); |