From 8deee9bd9b9137c256c23b86be11dbf0466f3aa8 Mon Sep 17 00:00:00 2001 From: Michael Tyler Date: Fri, 30 Jun 2023 11:26:05 +0100 Subject: Depthwise channel pre-multiplication Resolves: COMPMID-6337 Change-Id: Ie9097b3f56e8071426c621386a5988bd7f7e8ef2 Signed-off-by: Michael Tyler Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/9852 Tested-by: Arm Jenkins Reviewed-by: Viet-Hoa Do Benchmark: Arm Jenkins --- .../NEON/kernels/arm_conv/depthwise/interleaves/a64_s8q_3x3_dot.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src/core/NEON/kernels/arm_conv/depthwise/interleaves/a64_s8q_3x3_dot.cpp') diff --git a/src/core/NEON/kernels/arm_conv/depthwise/interleaves/a64_s8q_3x3_dot.cpp b/src/core/NEON/kernels/arm_conv/depthwise/interleaves/a64_s8q_3x3_dot.cpp index 5e4bf99120..3de4bdc1fb 100644 --- a/src/core/NEON/kernels/arm_conv/depthwise/interleaves/a64_s8q_3x3_dot.cpp +++ b/src/core/NEON/kernels/arm_conv/depthwise/interleaves/a64_s8q_3x3_dot.cpp @@ -42,7 +42,7 @@ size_t interleave_a64_s8q_3x3_dot::get_packed_size(const DepthwiseArgs &args) { // We store 7 vectors for every of channels. const unsigned int n = arm_gemm::roundup( - arm_gemm::iceildiv((long unsigned int) args.input_channels, + arm_gemm::iceildiv((long unsigned int) args.input_channels * args.channel_multiplier, get_vector_length(arm_gemm::VLType::None)), 4lu ); return n * 7 * get_vector_length(arm_gemm::VLType::None); -- cgit v1.2.1