Skip to content

Commit a175e88

Browse files
author
Jenkins
committed
Compute Library v22.05
1 parent 8f587de commit a175e88

530 files changed

Lines changed: 48050 additions & 25029 deletions

File tree

Some content is hidden

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

‎Android.bp‎

Lines changed: 63 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
//
2-
// Copyright © 2020-2021 Arm Ltd. All rights reserved.
2+
// Copyright © 2020-2022 Arm Ltd. All rights reserved.
33
// SPDX-License-Identifier: MIT
44
//
55

@@ -109,6 +109,8 @@ opencl_srcs = [
109109
"src/core/CL/cl_kernels/nhwc/normalization_layer.cl",
110110
"src/core/CL/cl_kernels/nhwc/normalize_planar_yuv_layer.cl",
111111
"src/core/CL/cl_kernels/nhwc/normalize_planar_yuv_layer_quantized.cl",
112+
"src/core/CL/cl_kernels/nhwc/pooling_3d_layer.cl",
113+
"src/core/CL/cl_kernels/nhwc/pooling_3d_layer_quantized.cl",
112114
"src/core/CL/cl_kernels/nhwc/pooling_layer.cl",
113115
"src/core/CL/cl_kernels/nhwc/pooling_layer_quantized.cl",
114116
"src/core/CL/cl_kernels/nhwc/reorg_layer.cl",
@@ -278,7 +280,6 @@ cc_library_static {
278280
"src/core/NEON/kernels/NEInstanceNormalizationLayerKernel.cpp",
279281
"src/core/NEON/kernels/NEL2NormalizeLayerKernel.cpp",
280282
"src/core/NEON/kernels/NELogicalKernel.cpp",
281-
"src/core/NEON/kernels/NEMaxUnpoolingLayerKernel.cpp",
282283
"src/core/NEON/kernels/NEMeanStdDevNormalizationKernel.cpp",
283284
"src/core/NEON/kernels/NENormalizationLayerKernel.cpp",
284285
"src/core/NEON/kernels/NEPadLayerKernel.cpp",
@@ -296,12 +297,16 @@ cc_library_static {
296297
"src/core/NEON/kernels/NEStackLayerKernel.cpp",
297298
"src/core/NEON/kernels/NEStridedSliceKernel.cpp",
298299
"src/core/NEON/kernels/NETileKernel.cpp",
300+
"src/core/NEON/kernels/arm_conv/addressing.cpp",
299301
"src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp",
300302
"src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp32.cpp",
301303
"src/core/NEON/kernels/arm_conv/depthwise/depthwise_s8q.cpp",
304+
"src/core/NEON/kernels/arm_conv/depthwise/depthwise_strategies_common.cpp",
302305
"src/core/NEON/kernels/arm_conv/depthwise/depthwise_u8q.cpp",
303306
"src/core/NEON/kernels/arm_conv/depthwise/depthwise_u8s8u8q.cpp",
304307
"src/core/NEON/kernels/arm_conv/depthwise/interleaves/8b_mla.cpp",
308+
"src/core/NEON/kernels/arm_conv/depthwise/interleaves/generic.cpp",
309+
"src/core/NEON/kernels/arm_conv/depthwise/interleaves/generic_quantized_dot_product.cpp",
305310
"src/core/NEON/kernels/arm_conv/pooling/kernels/cpp_nhwc_1x1_stride_any_depthfirst/generic.cpp",
306311
"src/core/NEON/kernels/arm_conv/pooling/pooling_fp16.cpp",
307312
"src/core/NEON/kernels/arm_conv/pooling/pooling_fp32.cpp",
@@ -363,6 +368,16 @@ cc_library_static {
363368
"src/core/Utils.cpp",
364369
"src/core/Validate.cpp",
365370
"src/core/Version.cpp",
371+
"src/core/experimental/dynamic_fusion/ClKernelBuildingAPI.cpp",
372+
"src/core/experimental/dynamic_fusion/ClKernelBuildingImpl/components/ClDirectConvolutionKernelComponent.cpp",
373+
"src/core/experimental/dynamic_fusion/ClKernelBuildingImpl/components/ClElementwiseAddKernelComponent.cpp",
374+
"src/core/experimental/dynamic_fusion/ClKernelBuildingImpl/components/ClStoreKernelComponents.cpp",
375+
"src/core/experimental/dynamic_fusion/OperatorGraph.cpp",
376+
"src/core/experimental/dynamic_fusion/WorkloadImpl/ClFusedKernelGraph.cpp",
377+
"src/core/experimental/dynamic_fusion/WorkloadImpl/ClKernelGraph.cpp",
378+
"src/core/experimental/dynamic_fusion/WorkloadImpl/ClWorkload.cpp",
379+
"src/core/experimental/dynamic_fusion/WorkloadImpl/DependencyGraph.cpp",
380+
"src/core/experimental/dynamic_fusion/WorkloadImpl/OperatorGraphImpl.cpp",
366381
"src/core/helpers/SoftmaxHelpers.cpp",
367382
"src/core/helpers/WindowHelpers.cpp",
368383
"src/core/utils/AssemblyUtils.cpp",
@@ -412,9 +427,11 @@ cc_library_static {
412427
"src/cpu/kernels/CpuGemmMatrixMultiplyKernel.cpp",
413428
"src/cpu/kernels/CpuGemmTranspose1xWKernel.cpp",
414429
"src/cpu/kernels/CpuIm2ColKernel.cpp",
430+
"src/cpu/kernels/CpuMaxUnpoolingLayerKernel.cpp",
415431
"src/cpu/kernels/CpuMulKernel.cpp",
416432
"src/cpu/kernels/CpuPermuteKernel.cpp",
417433
"src/cpu/kernels/CpuPool2dKernel.cpp",
434+
"src/cpu/kernels/CpuPool3dKernel.cpp",
418435
"src/cpu/kernels/CpuQuantizeKernel.cpp",
419436
"src/cpu/kernels/CpuReshapeKernel.cpp",
420437
"src/cpu/kernels/CpuScaleKernel.cpp",
@@ -439,20 +456,44 @@ cc_library_static {
439456
"src/cpu/kernels/boundingboxtransform/generic/neon/fp32.cpp",
440457
"src/cpu/kernels/boundingboxtransform/generic/neon/impl.cpp",
441458
"src/cpu/kernels/boundingboxtransform/generic/neon/qsymm16.cpp",
459+
"src/cpu/kernels/cast/generic/neon/bfloat16.cpp",
460+
"src/cpu/kernels/cast/generic/neon/fp16.cpp",
442461
"src/cpu/kernels/crop/generic/neon/fp16.cpp",
443462
"src/cpu/kernels/crop/generic/neon/fp32.cpp",
444463
"src/cpu/kernels/crop/generic/neon/impl.cpp",
445464
"src/cpu/kernels/crop/generic/neon/integer.cpp",
465+
"src/cpu/kernels/depthwiseconv2d/generic/neon/fp16.cpp",
466+
"src/cpu/kernels/depthwiseconv2d/generic/neon/fp32.cpp",
467+
"src/cpu/kernels/depthwiseconv2d/generic/neon/impl.cpp",
468+
"src/cpu/kernels/depthwiseconv2d/generic/neon/qasymm8.cpp",
469+
"src/cpu/kernels/depthwiseconv2d/generic/neon/qasymm8_signed.cpp",
470+
"src/cpu/kernels/directconv2d/nchw/all.cpp",
471+
"src/cpu/kernels/directconv2d/nhwc/neon/fp32.cpp",
472+
"src/cpu/kernels/directconv2d/nhwc/neon/impl.cpp",
446473
"src/cpu/kernels/elementwise_binary/generic/neon/fp16.cpp",
447474
"src/cpu/kernels/elementwise_binary/generic/neon/fp32.cpp",
448475
"src/cpu/kernels/elementwise_binary/generic/neon/integer.cpp",
449476
"src/cpu/kernels/elementwise_binary/generic/neon/qasymm8.cpp",
450477
"src/cpu/kernels/elementwise_binary/generic/neon/qasymm8_signed.cpp",
451478
"src/cpu/kernels/elementwise_unary/generic/neon/fp16.cpp",
452479
"src/cpu/kernels/elementwise_unary/generic/neon/fp32.cpp",
480+
"src/cpu/kernels/elementwise_unary/generic/neon/impl.cpp",
453481
"src/cpu/kernels/elementwise_unary/generic/neon/integer.cpp",
454482
"src/cpu/kernels/floor/neon/fp16.cpp",
455483
"src/cpu/kernels/floor/neon/fp32.cpp",
484+
"src/cpu/kernels/fuse_batch_normalization/generic/fp16.cpp",
485+
"src/cpu/kernels/fuse_batch_normalization/generic/fp32.cpp",
486+
"src/cpu/kernels/fuse_batch_normalization/generic/impl.cpp",
487+
"src/cpu/kernels/fuse_batch_normalization/nchw/all.cpp",
488+
"src/cpu/kernels/fuse_batch_normalization/nhwc/neon/fp16.cpp",
489+
"src/cpu/kernels/fuse_batch_normalization/nhwc/neon/fp32.cpp",
490+
"src/cpu/kernels/fuse_batch_normalization/nhwc/neon/impl.cpp",
491+
"src/cpu/kernels/gemm_matrix_add/generic/neon/fp16.cpp",
492+
"src/cpu/kernels/gemm_matrix_add/generic/neon/fp32.cpp",
493+
"src/cpu/kernels/gemm_matrix_add/generic/neon/impl.cpp",
494+
"src/cpu/kernels/gemm_matrix_mul/generic/neon/fp16.cpp",
495+
"src/cpu/kernels/gemm_matrix_mul/generic/neon/fp32.cpp",
496+
"src/cpu/kernels/gemm_matrix_mul/generic/neon/impl.cpp",
456497
"src/cpu/kernels/genproposals/generic/neon/fp16.cpp",
457498
"src/cpu/kernels/genproposals/generic/neon/fp32.cpp",
458499
"src/cpu/kernels/genproposals/generic/neon/impl.cpp",
@@ -462,6 +503,9 @@ cc_library_static {
462503
"src/cpu/kernels/instancenorm/generic/neon/impl.cpp",
463504
"src/cpu/kernels/internal/CpuDepthwiseConv2dAssemblyWrapperKernel.cpp",
464505
"src/cpu/kernels/internal/CpuPool2dAssemblyWrapperKernel.cpp",
506+
"src/cpu/kernels/l2normlayer/generic/neon/fp16.cpp",
507+
"src/cpu/kernels/l2normlayer/generic/neon/fp32.cpp",
508+
"src/cpu/kernels/l2normlayer/generic/neon/impl.cpp",
465509
"src/cpu/kernels/maxunpool/generic/neon/fp16.cpp",
466510
"src/cpu/kernels/maxunpool/generic/neon/fp32.cpp",
467511
"src/cpu/kernels/maxunpool/generic/neon/impl.cpp",
@@ -475,6 +519,11 @@ cc_library_static {
475519
"src/cpu/kernels/pool2d/neon/nchw/all.cpp",
476520
"src/cpu/kernels/pool2d/neon/qasymm8.cpp",
477521
"src/cpu/kernels/pool2d/neon/qasymm8_signed.cpp",
522+
"src/cpu/kernels/pool3d/neon/fp16.cpp",
523+
"src/cpu/kernels/pool3d/neon/fp32.cpp",
524+
"src/cpu/kernels/pool3d/neon/impl.cpp",
525+
"src/cpu/kernels/pool3d/neon/qasymm8.cpp",
526+
"src/cpu/kernels/pool3d/neon/qasymm8_signed.cpp",
478527
"src/cpu/kernels/range/generic/neon/fp16.cpp",
479528
"src/cpu/kernels/range/generic/neon/fp32.cpp",
480529
"src/cpu/kernels/range/generic/neon/impl.cpp",
@@ -494,6 +543,7 @@ cc_library_static {
494543
"src/cpu/kernels/select/generic/neon/integer.cpp",
495544
"src/cpu/kernels/softmax/generic/neon/fp16.cpp",
496545
"src/cpu/kernels/softmax/generic/neon/fp32.cpp",
546+
"src/cpu/kernels/softmax/generic/neon/impl.cpp",
497547
"src/cpu/kernels/softmax/generic/neon/qasymm8.cpp",
498548
"src/cpu/kernels/softmax/generic/neon/qasymm8_signed.cpp",
499549
"src/cpu/kernels/sub/neon/qasymm8.cpp",
@@ -522,9 +572,11 @@ cc_library_static {
522572
"src/cpu/operators/CpuGemmDirectConv2d.cpp",
523573
"src/cpu/operators/CpuGemmLowpMatrixMultiplyCore.cpp",
524574
"src/cpu/operators/CpuGemmLowpOutputStage.cpp",
575+
"src/cpu/operators/CpuMaxUnpooling.cpp",
525576
"src/cpu/operators/CpuMul.cpp",
526577
"src/cpu/operators/CpuPermute.cpp",
527578
"src/cpu/operators/CpuPool2d.cpp",
579+
"src/cpu/operators/CpuPool3d.cpp",
528580
"src/cpu/operators/CpuQuantize.cpp",
529581
"src/cpu/operators/CpuReshape.cpp",
530582
"src/cpu/operators/CpuScale.cpp",
@@ -571,6 +623,7 @@ cc_library_static {
571623
"src/gpu/cl/kernels/ClMulKernel.cpp",
572624
"src/gpu/cl/kernels/ClPermuteKernel.cpp",
573625
"src/gpu/cl/kernels/ClPool2dKernel.cpp",
626+
"src/gpu/cl/kernels/ClPool3dKernel.cpp",
574627
"src/gpu/cl/kernels/ClQuantizeKernel.cpp",
575628
"src/gpu/cl/kernels/ClReshapeKernel.cpp",
576629
"src/gpu/cl/kernels/ClScaleKernel.cpp",
@@ -583,6 +636,7 @@ cc_library_static {
583636
"src/gpu/cl/kernels/ClWinogradFilterTransformKernel.cpp",
584637
"src/gpu/cl/kernels/ClWinogradInputTransformKernel.cpp",
585638
"src/gpu/cl/kernels/ClWinogradOutputTransformKernel.cpp",
639+
"src/gpu/cl/kernels/experimental/dynamic_fusion/ClCompositeKernel.cpp",
586640
"src/gpu/cl/kernels/gemm/ClGemmHelpers.cpp",
587641
"src/gpu/cl/kernels/gemm/native/ClGemmDefaultConfigNativeBifrost.cpp",
588642
"src/gpu/cl/kernels/gemm/native/ClGemmDefaultConfigNativeMidgard.cpp",
@@ -617,13 +671,15 @@ cc_library_static {
617671
"src/gpu/cl/operators/ClPRelu.cpp",
618672
"src/gpu/cl/operators/ClPermute.cpp",
619673
"src/gpu/cl/operators/ClPool2d.cpp",
674+
"src/gpu/cl/operators/ClPool3d.cpp",
620675
"src/gpu/cl/operators/ClQuantize.cpp",
621676
"src/gpu/cl/operators/ClReshape.cpp",
622677
"src/gpu/cl/operators/ClScale.cpp",
623678
"src/gpu/cl/operators/ClSoftmax.cpp",
624679
"src/gpu/cl/operators/ClSub.cpp",
625680
"src/gpu/cl/operators/ClTranspose.cpp",
626681
"src/gpu/cl/operators/ClWinogradConv2d.cpp",
682+
"src/gpu/cl/operators/experimental/dynamic_fusion/ClCompositeOperator.cpp",
627683
"src/runtime/Allocator.cpp",
628684
"src/runtime/BlobLifetimeManager.cpp",
629685
"src/runtime/BlobMemoryPool.cpp",
@@ -700,6 +756,7 @@ cc_library_static {
700756
"src/runtime/CL/functions/CLPadLayer.cpp",
701757
"src/runtime/CL/functions/CLPermute.cpp",
702758
"src/runtime/CL/functions/CLPixelWiseMultiplication.cpp",
759+
"src/runtime/CL/functions/CLPooling3dLayer.cpp",
703760
"src/runtime/CL/functions/CLPoolingLayer.cpp",
704761
"src/runtime/CL/functions/CLPriorBoxLayer.cpp",
705762
"src/runtime/CL/functions/CLQLSTMLayer.cpp",
@@ -810,6 +867,7 @@ cc_library_static {
810867
"src/runtime/NEON/functions/NEPadLayer.cpp",
811868
"src/runtime/NEON/functions/NEPermute.cpp",
812869
"src/runtime/NEON/functions/NEPixelWiseMultiplication.cpp",
870+
"src/runtime/NEON/functions/NEPooling3dLayer.cpp",
813871
"src/runtime/NEON/functions/NEPoolingLayer.cpp",
814872
"src/runtime/NEON/functions/NEPriorBoxLayer.cpp",
815873
"src/runtime/NEON/functions/NEQLSTMLayer.cpp",
@@ -913,6 +971,9 @@ cc_library_static {
913971
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_packed_to_nhwc_3x3_s2_with_multiplier_output2x4_dot_depthfirst/generic.cpp",
914972
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_packed_to_nhwc_5x5_s1_with_multiplier_output4x2_dot_depthfirst/generic.cpp",
915973
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst/generic.cpp",
974+
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8qa_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp",
975+
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8qa_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp",
976+
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8qa_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp",
916977
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8s8u8q_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp",
917978
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8s8u8q_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp",
918979
"src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8s8u8q_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp",

0 commit comments

Comments
 (0)