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 --- arm_compute/core/Types.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'arm_compute/core/Types.h') diff --git a/arm_compute/core/Types.h b/arm_compute/core/Types.h index b0a6475527..d5a4125c88 100644 --- a/arm_compute/core/Types.h +++ b/arm_compute/core/Types.h @@ -150,8 +150,9 @@ enum class DepthwiseConvolutionFunction /** Available DeconvolutionMethod*/ enum class DeconvolutionMethod { - GEMM, /**< Deconvolution using GEMM */ - DIRECT, /**< Direct deconvolution */ + GEMM, /**< Deconvolution using GEMM */ + DIRECT, /**< Direct deconvolution */ + UPSCALE_CONV2D /**< Deconvolution with Upscaling */ }; /** Available FuseBatchNormalizationType*/ -- cgit v1.2.1