arm_compute v19.02

Change-Id: I853a3ecf38f206da13c1b03640c8adf73c20477c

Author: Jenkins

README.md

@@ -1,3 +1,7 @@
+<div align="center">
+ <img src="https://raw.githubusercontent.com/ARM-software/ComputeLibrary/gh-pages/ACL_logo.png"><br><br>
+</div>
+
 Release repository: https://github.com/arm-software/ComputeLibrary
 
 Development repository: https://review.mlplatform.org/#/admin/projects/ml/ComputeLibrary
@@ -19,6 +23,7 @@ Related projects:
 
 Documentation available here:
 
+- [v19.02](https://arm-software.github.io/ComputeLibrary/v19.02/)
 - [v18.11](https://arm-software.github.io/ComputeLibrary/v18.11/index.xhtml)
 - [v18.08](https://arm-software.github.io/ComputeLibrary/v18.08/)
 - [v18.05](https://arm-software.github.io/ComputeLibrary/v18.05/)
@@ -35,6 +40,8 @@ Documentation available here:
 
 Binaries available here:
 
+- [v19.02-linux](https://github.com/ARM-software/ComputeLibrary/releases/download/v19.02/arm_compute-v19.02-bin-linux.tar.gz)
+- [v19.02-android](https://github.com/ARM-software/ComputeLibrary/releases/download/v19.02/arm_compute-v19.02-bin-android.tar.gz)
 - [v18.11-linux](https://github.com/ARM-software/ComputeLibrary/releases/download/v18.11/arm_compute-v18.11-bin-linux.tar.gz)
 - [v18.11-android](https://github.com/ARM-software/ComputeLibrary/releases/download/v18.11/arm_compute-v18.11-bin-android.tar.gz)
 - [v18.08-linux](https://github.com/ARM-software/ComputeLibrary/releases/download/v18.08/arm_compute-v18.08-bin-linux.tar.gz)
@@ -54,6 +61,18 @@ Binaries available here:
 - [v17.04](https://github.com/ARM-software/ComputeLibrary/releases/download/v17.04/arm_compute-v17.04-bin.tar.gz)
 - [v17.03.1](https://github.com/ARM-software/ComputeLibrary/releases/download/v17.03.1/arm_compute-v17.03.1-bin.tar.gz)
 
-Contact: [email protected]
-
 License & Contributions: The software is provided under MIT license. Contributions to this project are accepted under the same license.
+
+### Public mailing list
+For technical discussion, the ComputeLibrary project has a public mailing list: [email protected]
+The list is open to anyone inside or outside of Arm to self subscribe.  In order to subscribe, please visit the following website:
+https://lists.linaro.org/mailman/listinfo/acl-dev
+
+### Developer Certificate of Origin (DCO)
+Before the ComputeLibrary project accepts your contribution, you need to certify its origin and give us your permission. To manage this process we use the Developer Certificate of Origin (DCO) V1.1 (https://developercertificate.org/)
+
+To indicate that you agree to the the terms of the DCO, you "sign off" your contribution by adding a line with your name and e-mail address to every git commit message:
+
+```Signed-off-by: John Doe <[email protected]>```
+
+You must use your real name, no pseudonyms or anonymous contributions are accepted.

SConscript

@@ -24,8 +24,8 @@ import os.path
 import re
 import subprocess
 
-VERSION = "v18.11"
-SONAME_VERSION="13.0.0"
+VERSION = "v19.02"
+SONAME_VERSION="14.0.0"
 
 Import('env')
 Import('vars')
@@ -161,7 +161,11 @@ arm_compute_env.Append(LIBS = ['dl'])
 core_files = Glob('src/core/*.cpp')
 core_files += Glob('src/core/CPP/*.cpp')
 core_files += Glob('src/core/CPP/kernels/*.cpp')
-core_files += Glob('src/core/utils/*/*.cpp')
+core_files += Glob('src/core/utils/helpers/*.cpp')
+core_files += Glob('src/core/utils/io/*.cpp')
+core_files += Glob('src/core/utils/quantization/*.cpp')
+if env["logging"]:
+    core_files += Glob('src/core/utils/logging/*.cpp')
 
 runtime_files = Glob('src/runtime/*.cpp')
 runtime_files += Glob('src/runtime/CPP/ICPPSimpleFunction.cpp')
@@ -186,6 +190,7 @@ if env['opencl']:
     runtime_files += Glob('src/runtime/CL/*.cpp')
     runtime_files += Glob('src/runtime/CL/functions/*.cpp')
     runtime_files += Glob('src/runtime/CL/tuners/*.cpp')
+    runtime_files += Glob('src/runtime/CL/gemm_reshaped/*.cpp')
 
     graph_files += Glob('src/graph/backends/CL/*.cpp')
 

SConstruct

@@ -55,6 +55,8 @@ vars.AddVariables(
     BoolVariable("cppthreads", "Enable C++11 threads backend", True),
     PathVariable("build_dir", "Specify sub-folder for the build", ".", PathVariable.PathAccept),
     PathVariable("install_dir", "Specify sub-folder for the install", "", PathVariable.PathAccept),
+    BoolVariable("exceptions", "Enable/disable C++ exception support", True),
+    ("toolchain_prefix", "Override the toolchain prefix", ""),
     ("extra_cxx_flags", "Extra CXX flags to be appended to the build command", ""),
     ("extra_link_flags", "Extra LD flags to be appended to the build command", ""),
     ("compiler_cache", "Command to prefix to the C and C++ compiler (e.g ccache)", "")
@@ -116,6 +118,14 @@ if env['os'] == 'bare_metal':
          print("ERROR: OpenMP and C++11 threads not supported in bare_metal. Use cppthreads=0 openmp=0")
          Exit(1)
 
+if not env['exceptions']:
+    if env['opencl'] or env['gles_compute']:
+         print("ERROR: OpenCL and GLES are not supported when building without exceptions. Use opencl=0 gles_compute=0")
+         Exit(1)
+
+    env.Append(CPPDEFINES = ['ARM_COMPUTE_EXCEPTIONS_DISABLED'])
+    env.Append(CXXFLAGS = ['-fno-exceptions'])
+
 env.Append(CXXFLAGS = ['-Wno-deprecated-declarations','-Wall','-DARCH_ARM',
          '-Wextra','-Wno-unused-parameter','-pedantic','-Wdisabled-optimization','-Wformat=2',
          '-Winit-self','-Wstrict-overflow=2','-Wswitch-default',
@@ -200,6 +210,9 @@ elif env['arch'] == 'x86_64':
 if env['build'] == 'native':
     prefix = ""
 
+if env["toolchain_prefix"] != "":
+    prefix = env["toolchain_prefix"]
+
 env['CC'] = env['compiler_cache']+" "+prefix + c_compiler
 env['CXX'] = env['compiler_cache']+" "+prefix + cpp_compiler
 env['LD'] = prefix + "ld"
@@ -295,8 +308,8 @@ if env['gles_compute'] and env['os'] != 'android':
 
 SConscript('./SConscript', variant_dir=build_path, duplicate=0)
 
-if env['examples'] and env['os'] != 'bare_metal':
+if env['examples'] and env['os'] != 'bare_metal' and env['exceptions']:
     SConscript('./examples/SConscript', variant_dir='%s/examples' % build_path, duplicate=0)
 
-if env['os'] != 'bare_metal':
+if env['os'] != 'bare_metal' and env['exceptions']:
     SConscript('./tests/SConscript', variant_dir='%s/tests' % build_path, duplicate=0)

arm_compute/core/CL/CLHelpers.h

@@ -135,5 +135,14 @@ bool dot8_acc_supported(const cl::Device &device);
  * @return True if the configuration is supported
  */
 bool cl_winograd_convolution_layer_supported(const Size2D &output_tile, const Size2D &kernel_size, DataLayout data_layout);
+
+/** Helper function to get the preferred native vector width size for built-in scalar types that can be put into vectors
+ *
+ * @param[in] device A CL device
+ * @param[in] dt     data type
+ *
+ * @return preferred vector width
+ */
+size_t preferred_vector_width(const cl::Device &device, DataType dt);
 }
 #endif /* __ARM_COMPUTE_CLHELPERS_H__ */

arm_compute/core/CL/CLKernels.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-2018 ARM Limited.
+ * Copyright (c) 2016-2019 ARM Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -28,9 +28,6 @@
 #include "arm_compute/core/CL/kernels/CLAbsoluteDifferenceKernel.h"
 #include "arm_compute/core/CL/kernels/CLAccumulateKernel.h"
 #include "arm_compute/core/CL/kernels/CLActivationLayerKernel.h"
-#include "arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h"
-#include "arm_compute/core/CL/kernels/CLArithmeticDivisionKernel.h"
-#include "arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h"
 #include "arm_compute/core/CL/kernels/CLBatchNormalizationLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLBatchToSpaceLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLBitwiseAndKernel.h"
@@ -45,6 +42,7 @@
 #include "arm_compute/core/CL/kernels/CLChannelShuffleLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLCol2ImKernel.h"
 #include "arm_compute/core/CL/kernels/CLColorConvertKernel.h"
+#include "arm_compute/core/CL/kernels/CLComparisonKernel.h"
 #include "arm_compute/core/CL/kernels/CLConvertFullyConnectedWeightsKernel.h"
 #include "arm_compute/core/CL/kernels/CLConvolutionKernel.h"
 #include "arm_compute/core/CL/kernels/CLCopyKernel.h"
@@ -53,14 +51,17 @@
 #include "arm_compute/core/CL/kernels/CLDepthConvertLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLDepthwiseConvolutionLayer3x3NCHWKernel.h"
 #include "arm_compute/core/CL/kernels/CLDepthwiseConvolutionLayer3x3NHWCKernel.h"
+#include "arm_compute/core/CL/kernels/CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel.h"
+#include "arm_compute/core/CL/kernels/CLDepthwiseConvolutionLayerReshapeWeightsKernel.h"
 #include "arm_compute/core/CL/kernels/CLDepthwiseIm2ColKernel.h"
 #include "arm_compute/core/CL/kernels/CLDepthwiseVectorToTensorKernel.h"
-#include "arm_compute/core/CL/kernels/CLDepthwiseWeightsReshapeKernel.h"
 #include "arm_compute/core/CL/kernels/CLDequantizationLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLDerivativeKernel.h"
 #include "arm_compute/core/CL/kernels/CLDilateKernel.h"
 #include "arm_compute/core/CL/kernels/CLDirectConvolutionLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLDirectConvolutionLayerOutputStageKernel.h"
+#include "arm_compute/core/CL/kernels/CLElementWiseUnaryLayerKernel.h"
+#include "arm_compute/core/CL/kernels/CLElementwiseOperationKernel.h"
 #include "arm_compute/core/CL/kernels/CLErodeKernel.h"
 #include "arm_compute/core/CL/kernels/CLFastCornersKernel.h"
 #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h"
@@ -69,6 +70,7 @@
 #include "arm_compute/core/CL/kernels/CLFuseBatchNormalizationKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMInterleave4x4Kernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMLowpMatrixMultiplyKernel.h"
+#include "arm_compute/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMLowpOffsetContributionKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMLowpOffsetContributionOutputStageKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel.h"
@@ -78,8 +80,12 @@
 #include "arm_compute/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMMatrixAdditionKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMMatrixMultiplyKernel.h"
+#include "arm_compute/core/CL/kernels/CLGEMMMatrixMultiplyReshapedKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMMatrixVectorMultiplyKernel.h"
+#include "arm_compute/core/CL/kernels/CLGEMMReshapeLHSMatrixKernel.h"
+#include "arm_compute/core/CL/kernels/CLGEMMReshapeRHSMatrixKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMTranspose1xWKernel.h"
+#include "arm_compute/core/CL/kernels/CLGatherKernel.h"
 #include "arm_compute/core/CL/kernels/CLGaussian3x3Kernel.h"
 #include "arm_compute/core/CL/kernels/CLGaussian5x5Kernel.h"
 #include "arm_compute/core/CL/kernels/CLGaussianPyramidKernel.h"
@@ -110,20 +116,25 @@
 #include "arm_compute/core/CL/kernels/CLQuantizationLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLROIAlignLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLROIPoolingLayerKernel.h"
+#include "arm_compute/core/CL/kernels/CLRangeKernel.h"
 #include "arm_compute/core/CL/kernels/CLReductionOperationKernel.h"
 #include "arm_compute/core/CL/kernels/CLRemapKernel.h"
 #include "arm_compute/core/CL/kernels/CLReorgLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLReshapeLayerKernel.h"
+#include "arm_compute/core/CL/kernels/CLReverseKernel.h"
 #include "arm_compute/core/CL/kernels/CLScaleKernel.h"
 #include "arm_compute/core/CL/kernels/CLScharr3x3Kernel.h"
+#include "arm_compute/core/CL/kernels/CLSelectKernel.h"
 #include "arm_compute/core/CL/kernels/CLSobel3x3Kernel.h"
 #include "arm_compute/core/CL/kernels/CLSobel5x5Kernel.h"
 #include "arm_compute/core/CL/kernels/CLSobel7x7Kernel.h"
 #include "arm_compute/core/CL/kernels/CLSoftmaxLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLSpaceToBatchLayerKernel.h"
+#include "arm_compute/core/CL/kernels/CLStackLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLStridedSliceKernel.h"
 #include "arm_compute/core/CL/kernels/CLTableLookupKernel.h"
 #include "arm_compute/core/CL/kernels/CLThresholdKernel.h"
+#include "arm_compute/core/CL/kernels/CLTileKernel.h"
 #include "arm_compute/core/CL/kernels/CLTransposeKernel.h"
 #include "arm_compute/core/CL/kernels/CLUpsampleLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLWarpAffineKernel.h"

arm_compute/core/CL/ICLArray.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-2018 ARM Limited.
+ * Copyright (c) 2016-2019 ARM Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -121,8 +121,6 @@ using ICLKeyPointArray = ICLArray<KeyPoint>;
 using ICLCoordinates2DArray = ICLArray<Coordinates2D>;
 /** Interface for OpenCL Array of Detection Windows. */
 using ICLDetectionWindowArray = ICLArray<DetectionWindow>;
-/** Interface for OpenCL Array of ROIs. */
-using ICLROIArray = ICLArray<ROI>;
 /** Interface for OpenCL Array of 2D Sizes. */
 using ICLSize2DArray = ICLArray<Size2D>;
 /** Interface for OpenCL Array of uint8s. */