diff options
author | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2018-06-28 16:29:29 +0100 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:54:54 +0000 |
commit | 215b4ea6c9dee480a22070d5873b0b8cb52531a0 (patch) | |
tree | 398e552c4d01c0b84d03a873098a9183ba8f82e4 /src/core/CL/CLKernelLibrary.cpp | |
parent | ad486e21e5870f41774f30825c270762e08ae71e (diff) | |
download | ComputeLibrary-215b4ea6c9dee480a22070d5873b0b8cb52531a0.tar.gz |
COMPMID-1277 - Optimizing CLIm2ColKernel for NHWC.
This patch includes:
- Im2Col optimizations for NHWC using a new data layout
- Refactoring of CLIm2ColKernel adding validation method and auto-init
- Removed im2col_reduced from CLIm2ColKernel and created a new kernel CLFlattenLayerKernel
Change-Id: I1620640b6796baa268324b33ae92cdd8de53e27c
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/141241
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Diffstat (limited to 'src/core/CL/CLKernelLibrary.cpp')
-rw-r--r-- | src/core/CL/CLKernelLibrary.cpp | 21 |
1 files changed, 12 insertions, 9 deletions
diff --git a/src/core/CL/CLKernelLibrary.cpp b/src/core/CL/CLKernelLibrary.cpp index 64519ff459..29b01e6cea 100644 --- a/src/core/CL/CLKernelLibrary.cpp +++ b/src/core/CL/CLKernelLibrary.cpp @@ -226,6 +226,7 @@ const std::map<std::string, std::string> CLKernelLibrary::_kernel_program_map = { "direct_convolution_1x1_3x3_5x5_quantized", "direct_convolution_1x1_3x3_5x5_quantized.cl" }, { "erode", "erode.cl" }, { "fast_corners", "fast_corners.cl" }, + { "flatten", "flatten.cl" }, { "fill_image_borders_constant", "fill_border.cl" }, { "fill_image_borders_replicate", "fill_border.cl" }, { "finalize", "optical_flow_pyramid_lk.cl" }, @@ -270,13 +271,12 @@ const std::map<std::string, std::string> CLKernelLibrary::_kernel_program_map = { "hog_detector", "hog.cl" }, { "hog_orientation_binning", "hog.cl" }, { "hysteresis", "canny.cl" }, - { "im2col1x1_stridex1_dchw", "im2col.cl" }, - { "im2col3x3_dchw", "im2col.cl" }, - { "im2col5x5_dchw", "im2col.cl" }, - { "im2col11x11_padx0_pady0_dchw", "im2col.cl" }, - { "im2col_generic_dchw", "im2col.cl" }, - { "im2col_generic_padx0_pady0_dchw", "im2col.cl" }, - { "im2col_reduced_dchw", "im2col.cl" }, + { "im2col1x1_stridex1_nchw", "im2col.cl" }, + { "im2col3x3_nchw", "im2col.cl" }, + { "im2col5x5_nchw", "im2col.cl" }, + { "im2col11x11_padx0_pady0_nchw", "im2col.cl" }, + { "im2col_generic_nchw", "im2col.cl" }, + { "im2col_generic_padx0_pady0_nchw", "im2col.cl" }, { "im2col3x3_nhwc", "im2col.cl" }, { "im2col_generic_nhwc", "im2col.cl" }, { "init_level", "optical_flow_pyramid_lk.cl" }, @@ -333,8 +333,7 @@ const std::map<std::string, std::string> CLKernelLibrary::_kernel_program_map = { "remap_nearest_neighbour", "remap.cl" }, { "remap_bilinear", "remap.cl" }, { "reshape_layer", "reshape_layer.cl" }, - { "reshape_to_columns_nchw", "convolution_layer.cl" }, - { "reshape_to_columns_nhwc", "convolution_layer.cl" }, + { "reshape_to_columns", "convolution_layer.cl" }, { "RGB888_to_IYUV_bt709", "color_convert.cl" }, { "RGB888_to_NV12_bt709", "color_convert.cl" }, { "RGB888_to_RGBA8888_bt709", "color_convert.cl" }, @@ -572,6 +571,10 @@ const std::map<std::string, std::string> CLKernelLibrary::_program_source_map = #include "./cl_kernels/fast_corners.clembed" }, { + "flatten.cl", +#include "./cl_kernels/flatten.clembed" + }, + { "fill_border.cl", #include "./cl_kernels/fill_border.clembed" }, |