From ec0113dd7749991959ae351934eea0c0d8077dcb Mon Sep 17 00:00:00 2001 From: Gunes Bayir Date: Wed, 9 Nov 2022 09:26:27 +0000 Subject: Optimize Transposed Convolution for CL backend (FP32/16) This patch optimizes transposed convolution for CL backend by rewriting it in a single kernel instead of three (flip_kernel + upsample + conv). The new kernel skips the upsampling step which reduces the input space of convolution by stride_x * stride_y, resulting in significant performance improvement. It also skips the kernel flipping by traversing the weights accordingly, thus reduces the memory footprint. Resolves: COMPMID-5676 Signed-off-by: Gunes Bayir Change-Id: I8a333212dc7c5f7f0597aa58b0d56d44814baa14 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8588 Tested-by: Arm Jenkins Reviewed-by: Gian Marco Iodice Comments-Addressed: Arm Jenkins Benchmark: Arm Jenkins --- SConscript | 1 + 1 file changed, 1 insertion(+) (limited to 'SConscript') diff --git a/SConscript b/SConscript index c05c1fe484..42a03f0a04 100644 --- a/SConscript +++ b/SConscript @@ -460,6 +460,7 @@ if env['opencl'] and env['embed_kernels']: 'src/core/CL/cl_kernels/nhwc/scale.cl', 'src/core/CL/cl_kernels/nhwc/space_to_batch.cl', 'src/core/CL/cl_kernels/nhwc/space_to_depth.cl', + 'src/core/CL/cl_kernels/nhwc/transposed_convolution.cl', 'src/core/CL/cl_kernels/nhwc/upsample_layer.cl', 'src/core/CL/cl_kernels/nhwc/winograd_filter_transform.cl', 'src/core/CL/cl_kernels/nhwc/winograd_input_transform.cl', -- cgit v1.2.1