Skip to content

Commit 8f587de

Browse files
author
Jenkins
committed
Compute Library v22.02
1 parent 91ee4d0 commit 8f587de

File tree

400 files changed

+14362
-11090
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

400 files changed

+14362
-11090
lines changed

‎Android.bp‎

Lines changed: 65 additions & 37 deletions
Original file line numberDiff line numberDiff line change
@@ -40,6 +40,7 @@ opencl_srcs = [
4040
"src/core/CL/cl_kernels/common/floor.cl",
4141
"src/core/CL/cl_kernels/common/gather.cl",
4242
"src/core/CL/cl_kernels/common/gemm.cl",
43+
"src/core/CL/cl_kernels/common/gemm_utils.cl",
4344
"src/core/CL/cl_kernels/common/gemmlowp.cl",
4445
"src/core/CL/cl_kernels/common/gemv.cl",
4546
"src/core/CL/cl_kernels/common/generate_proposals.cl",
@@ -80,17 +81,13 @@ opencl_srcs = [
8081
"src/core/CL/cl_kernels/nchw/channel_shuffle.cl",
8182
"src/core/CL/cl_kernels/nchw/depth_to_space.cl",
8283
"src/core/CL/cl_kernels/nchw/dequantization_layer.cl",
83-
"src/core/CL/cl_kernels/nchw/direct_convolution1x1.cl",
84-
"src/core/CL/cl_kernels/nchw/direct_convolution3x3.cl",
85-
"src/core/CL/cl_kernels/nchw/direct_convolution5x5.cl",
86-
"src/core/CL/cl_kernels/nchw/direct_convolution_quantized.cl",
84+
"src/core/CL/cl_kernels/nchw/direct_convolution.cl",
8785
"src/core/CL/cl_kernels/nchw/im2col.cl",
8886
"src/core/CL/cl_kernels/nchw/normalization_layer.cl",
8987
"src/core/CL/cl_kernels/nchw/normalize_planar_yuv_layer.cl",
9088
"src/core/CL/cl_kernels/nchw/normalize_planar_yuv_layer_quantized.cl",
9189
"src/core/CL/cl_kernels/nchw/pooling_layer.cl",
9290
"src/core/CL/cl_kernels/nchw/prior_box_layer.cl",
93-
"src/core/CL/cl_kernels/nchw/remap.cl",
9491
"src/core/CL/cl_kernels/nchw/reorg_layer.cl",
9592
"src/core/CL/cl_kernels/nchw/scale.cl",
9693
"src/core/CL/cl_kernels/nchw/space_to_batch.cl",
@@ -114,7 +111,6 @@ opencl_srcs = [
114111
"src/core/CL/cl_kernels/nhwc/normalize_planar_yuv_layer_quantized.cl",
115112
"src/core/CL/cl_kernels/nhwc/pooling_layer.cl",
116113
"src/core/CL/cl_kernels/nhwc/pooling_layer_quantized.cl",
117-
"src/core/CL/cl_kernels/nhwc/remap.cl",
118114
"src/core/CL/cl_kernels/nhwc/reorg_layer.cl",
119115
"src/core/CL/cl_kernels/nhwc/scale.cl",
120116
"src/core/CL/cl_kernels/nhwc/space_to_batch.cl",
@@ -241,7 +237,6 @@ cc_library_static {
241237
"src/core/CL/kernels/CLROIPoolingLayerKernel.cpp",
242238
"src/core/CL/kernels/CLRangeKernel.cpp",
243239
"src/core/CL/kernels/CLReductionOperationKernel.cpp",
244-
"src/core/CL/kernels/CLRemapKernel.cpp",
245240
"src/core/CL/kernels/CLReorgLayerKernel.cpp",
246241
"src/core/CL/kernels/CLReverseKernel.cpp",
247242
"src/core/CL/kernels/CLSelectKernel.cpp",
@@ -293,7 +288,6 @@ cc_library_static {
293288
"src/core/NEON/kernels/NEROIPoolingLayerKernel.cpp",
294289
"src/core/NEON/kernels/NERangeKernel.cpp",
295290
"src/core/NEON/kernels/NEReductionOperationKernel.cpp",
296-
"src/core/NEON/kernels/NERemapKernel.cpp",
297291
"src/core/NEON/kernels/NEReorgLayerKernel.cpp",
298292
"src/core/NEON/kernels/NEReverseKernel.cpp",
299293
"src/core/NEON/kernels/NESelectKernel.cpp",
@@ -336,8 +330,6 @@ cc_library_static {
336330
"src/core/NEON/kernels/arm_gemm/transform.cpp",
337331
"src/core/NEON/kernels/batchnormalization/impl/NEON/fp16.cpp",
338332
"src/core/NEON/kernels/batchnormalization/impl/NEON/fp32.cpp",
339-
"src/core/NEON/kernels/batchnormalization/impl/SVE/fp16.cpp",
340-
"src/core/NEON/kernels/batchnormalization/impl/SVE/fp32.cpp",
341333
"src/core/NEON/kernels/convolution/common/padding.cpp",
342334
"src/core/NEON/kernels/convolution/common/qasymm8.cpp",
343335
"src/core/NEON/kernels/convolution/common/qsymm8.cpp",
@@ -431,44 +423,79 @@ cc_library_static {
431423
"src/cpu/kernels/CpuTransposeKernel.cpp",
432424
"src/cpu/kernels/CpuWeightsReshapeKernel.cpp",
433425
"src/cpu/kernels/CpuWinogradConv2dKernel.cpp",
434-
"src/cpu/kernels/activation/neon/fp16.cpp",
435-
"src/cpu/kernels/activation/neon/fp32.cpp",
436-
"src/cpu/kernels/activation/neon/qasymm8.cpp",
437-
"src/cpu/kernels/activation/neon/qasymm8_signed.cpp",
438-
"src/cpu/kernels/activation/neon/qsymm16.cpp",
439-
"src/cpu/kernels/activation/sve/fp16.cpp",
440-
"src/cpu/kernels/activation/sve/fp32.cpp",
441-
"src/cpu/kernels/activation/sve/qasymm8.cpp",
442-
"src/cpu/kernels/activation/sve/qasymm8_signed.cpp",
443-
"src/cpu/kernels/activation/sve/qsymm16.cpp",
444-
"src/cpu/kernels/add/neon/qasymm8.cpp",
445-
"src/cpu/kernels/add/neon/qasymm8_signed.cpp",
446-
"src/cpu/kernels/add/neon/qsymm16.cpp",
447-
"src/cpu/kernels/add/sve/impl.cpp",
448-
"src/cpu/kernels/add/sve/qasymm8.cpp",
449-
"src/cpu/kernels/add/sve/qasymm8_signed.cpp",
450-
"src/cpu/kernels/add/sve/qsymm16.cpp",
451-
"src/cpu/kernels/elementwise/sve/elementwise.cpp",
452-
"src/cpu/kernels/elementwise/sve/elementwise_unary.cpp",
426+
"src/cpu/kernels/activation/generic/neon/fp16.cpp",
427+
"src/cpu/kernels/activation/generic/neon/fp32.cpp",
428+
"src/cpu/kernels/activation/generic/neon/qasymm8.cpp",
429+
"src/cpu/kernels/activation/generic/neon/qasymm8_signed.cpp",
430+
"src/cpu/kernels/activation/generic/neon/qsymm16.cpp",
431+
"src/cpu/kernels/add/generic/neon/fp16.cpp",
432+
"src/cpu/kernels/add/generic/neon/fp32.cpp",
433+
"src/cpu/kernels/add/generic/neon/impl.cpp",
434+
"src/cpu/kernels/add/generic/neon/integer.cpp",
435+
"src/cpu/kernels/add/generic/neon/qasymm8.cpp",
436+
"src/cpu/kernels/add/generic/neon/qasymm8_signed.cpp",
437+
"src/cpu/kernels/add/generic/neon/qsymm16.cpp",
438+
"src/cpu/kernels/boundingboxtransform/generic/neon/fp16.cpp",
439+
"src/cpu/kernels/boundingboxtransform/generic/neon/fp32.cpp",
440+
"src/cpu/kernels/boundingboxtransform/generic/neon/impl.cpp",
441+
"src/cpu/kernels/boundingboxtransform/generic/neon/qsymm16.cpp",
442+
"src/cpu/kernels/crop/generic/neon/fp16.cpp",
443+
"src/cpu/kernels/crop/generic/neon/fp32.cpp",
444+
"src/cpu/kernels/crop/generic/neon/impl.cpp",
445+
"src/cpu/kernels/crop/generic/neon/integer.cpp",
446+
"src/cpu/kernels/elementwise_binary/generic/neon/fp16.cpp",
447+
"src/cpu/kernels/elementwise_binary/generic/neon/fp32.cpp",
448+
"src/cpu/kernels/elementwise_binary/generic/neon/integer.cpp",
449+
"src/cpu/kernels/elementwise_binary/generic/neon/qasymm8.cpp",
450+
"src/cpu/kernels/elementwise_binary/generic/neon/qasymm8_signed.cpp",
451+
"src/cpu/kernels/elementwise_unary/generic/neon/fp16.cpp",
452+
"src/cpu/kernels/elementwise_unary/generic/neon/fp32.cpp",
453+
"src/cpu/kernels/elementwise_unary/generic/neon/integer.cpp",
453454
"src/cpu/kernels/floor/neon/fp16.cpp",
454455
"src/cpu/kernels/floor/neon/fp32.cpp",
456+
"src/cpu/kernels/genproposals/generic/neon/fp16.cpp",
457+
"src/cpu/kernels/genproposals/generic/neon/fp32.cpp",
458+
"src/cpu/kernels/genproposals/generic/neon/impl.cpp",
459+
"src/cpu/kernels/genproposals/generic/neon/qsymm16.cpp",
460+
"src/cpu/kernels/instancenorm/generic/neon/fp16.cpp",
461+
"src/cpu/kernels/instancenorm/generic/neon/fp32.cpp",
462+
"src/cpu/kernels/instancenorm/generic/neon/impl.cpp",
455463
"src/cpu/kernels/internal/CpuDepthwiseConv2dAssemblyWrapperKernel.cpp",
456464
"src/cpu/kernels/internal/CpuPool2dAssemblyWrapperKernel.cpp",
465+
"src/cpu/kernels/maxunpool/generic/neon/fp16.cpp",
466+
"src/cpu/kernels/maxunpool/generic/neon/fp32.cpp",
467+
"src/cpu/kernels/maxunpool/generic/neon/impl.cpp",
468+
"src/cpu/kernels/maxunpool/generic/neon/qasymm8.cpp",
469+
"src/cpu/kernels/maxunpool/generic/neon/qasymm8_signed.cpp",
470+
"src/cpu/kernels/meanstddevnorm/generic/neon/fp16.cpp",
471+
"src/cpu/kernels/meanstddevnorm/generic/neon/fp32.cpp",
472+
"src/cpu/kernels/meanstddevnorm/generic/neon/impl.cpp",
457473
"src/cpu/kernels/pool2d/neon/fp16.cpp",
458474
"src/cpu/kernels/pool2d/neon/fp32.cpp",
459475
"src/cpu/kernels/pool2d/neon/nchw/all.cpp",
460476
"src/cpu/kernels/pool2d/neon/qasymm8.cpp",
461477
"src/cpu/kernels/pool2d/neon/qasymm8_signed.cpp",
478+
"src/cpu/kernels/range/generic/neon/fp16.cpp",
479+
"src/cpu/kernels/range/generic/neon/fp32.cpp",
480+
"src/cpu/kernels/range/generic/neon/impl.cpp",
481+
"src/cpu/kernels/range/generic/neon/integer.cpp",
482+
"src/cpu/kernels/roialign/generic/neon/fp16.cpp",
483+
"src/cpu/kernels/roialign/generic/neon/fp32.cpp",
484+
"src/cpu/kernels/roialign/generic/neon/impl.cpp",
485+
"src/cpu/kernels/roialign/generic/neon/qasymm8.cpp",
486+
"src/cpu/kernels/roialign/generic/neon/qasymm8_signed.cpp",
462487
"src/cpu/kernels/scale/neon/fp16.cpp",
463488
"src/cpu/kernels/scale/neon/integer.cpp",
464489
"src/cpu/kernels/scale/neon/qasymm8.cpp",
465490
"src/cpu/kernels/scale/neon/qasymm8_signed.cpp",
466-
"src/cpu/kernels/scale/sve/fp16.cpp",
467-
"src/cpu/kernels/scale/sve/fp32.cpp",
468-
"src/cpu/kernels/scale/sve/integer.cpp",
469-
"src/cpu/kernels/scale/sve/qasymm8.cpp",
470-
"src/cpu/kernels/scale/sve/qasymm8_signed.cpp",
471-
"src/cpu/kernels/softmax/impl/sve/impl.cpp",
491+
"src/cpu/kernels/select/generic/neon/fp16.cpp",
492+
"src/cpu/kernels/select/generic/neon/fp32.cpp",
493+
"src/cpu/kernels/select/generic/neon/impl.cpp",
494+
"src/cpu/kernels/select/generic/neon/integer.cpp",
495+
"src/cpu/kernels/softmax/generic/neon/fp16.cpp",
496+
"src/cpu/kernels/softmax/generic/neon/fp32.cpp",
497+
"src/cpu/kernels/softmax/generic/neon/qasymm8.cpp",
498+
"src/cpu/kernels/softmax/generic/neon/qasymm8_signed.cpp",
472499
"src/cpu/kernels/sub/neon/qasymm8.cpp",
473500
"src/cpu/kernels/sub/neon/qasymm8_signed.cpp",
474501
"src/cpu/kernels/sub/neon/qsymm16.cpp",
@@ -683,7 +710,6 @@ cc_library_static {
683710
"src/runtime/CL/functions/CLRange.cpp",
684711
"src/runtime/CL/functions/CLReduceMean.cpp",
685712
"src/runtime/CL/functions/CLReductionOperation.cpp",
686-
"src/runtime/CL/functions/CLRemap.cpp",
687713
"src/runtime/CL/functions/CLReorgLayer.cpp",
688714
"src/runtime/CL/functions/CLReshapeLayer.cpp",
689715
"src/runtime/CL/functions/CLReverse.cpp",
@@ -794,7 +820,6 @@ cc_library_static {
794820
"src/runtime/NEON/functions/NERange.cpp",
795821
"src/runtime/NEON/functions/NEReduceMean.cpp",
796822
"src/runtime/NEON/functions/NEReductionOperation.cpp",
797-
"src/runtime/NEON/functions/NERemap.cpp",
798823
"src/runtime/NEON/functions/NEReorgLayer.cpp",
799824
"src/runtime/NEON/functions/NEReshapeLayer.cpp",
800825
"src/runtime/NEON/functions/NEReverse.cpp",
@@ -1013,8 +1038,11 @@ cc_library_static {
10131038
"src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8u32_mmla_6x16/generic.cpp",
10141039
"src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_dot_8x12/generic.cpp",
10151040
"src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_dot_8x12/x1.cpp",
1041+
"src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_mmla_8x12/a510.cpp",
10161042
"src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_mmla_8x12/generic.cpp",
1043+
"src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_s8s32_mmla_8x12/a510.cpp",
10171044
"src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_s8s32_mmla_8x12/generic.cpp",
1045+
"src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_u8u32_mmla_8x12/a510.cpp",
10181046
"src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_u8u32_mmla_8x12/generic.cpp",
10191047
"src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x12/a53.cpp",
10201048
"src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x12/a55.cpp",

‎LICENSE‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
MIT License
22

3-
Copyright (c) 2017-2021 Arm Limited
3+
Copyright (c) 2017-2022 Arm Limited
44

55
Permission is hereby granted, free of charge, to any person obtaining a copy
66
of this software and associated documentation files (the "Software"), to deal

0 commit comments

Comments
 (0)