Blame - docs/00_introduction.dox - ml/ComputeLibrary

blob: fa6c22713f982443b5b49ff9f499ab5900bb6a71 [file] [log] [blame]

Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	1	/** @mainpage Introduction
				2
				3	@tableofcontents
				4
				5	The Computer Vision and Machine Learning library is a set of functions optimised for both ARM CPUs and GPUs using SIMD technologies.
				6
				7	Several builds of the library are available using various configurations:
				8	- OS: Linux, Android or bare metal.
				9	- Architecture: armv7a (32bit) or arm64-v8a (64bit)
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	10	- Technology: NEON / OpenCL / GLES_COMPUTE / NEON and OpenCL and GLES_COMPUTE
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	11	- Debug / Asserts / Release: Use a build with asserts enabled to debug your application and enable extra validation. Once you are sure your application works as expected you can switch to a release build of the library for maximum performance.
				12
				13	@section S0_1_contact Contact / Support
				14
				15	Please email developer@arm.com
				16
				17	In order to facilitate the work of the support team please provide the build information of the library you are using. To get the version of the library you are using simply run:
				18
				19	$ strings android-armv7a-cl-asserts/libarm_compute.so \| grep arm_compute_version
				20	arm_compute_version=v16.12 Build options: {'embed_kernels': '1', 'opencl': '1', 'arch': 'armv7a', 'neon': '0', 'asserts': '1', 'debug': '0', 'os': 'android', 'Werror': '1'} Git hash=f51a545d4ea12a9059fe4e598a092f1fd06dc858
				21
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	22	@section S0_2_prebuilt_binaries Pre-built binaries
				23
				24	For each release we provide some pre-built binaries of the library [here](https://github.com/ARM-software/ComputeLibrary/releases)
				25
				26	These binaries have been built using the following toolchains:
				27	- Linux armv7a: gcc-linaro-arm-linux-gnueabihf-4.9-2014.07_linux
				28	- Linux arm64-v8a: gcc-linaro-4.9-2016.02-x86_64_aarch64-linux-gnu
				29	- Android armv7a: clang++ / gnustl NDK r14
				30	- Android am64-v8a: clang++ / gnustl NDK r14
				31
				32	@warning Make sure to use a compatible toolchain to build your application or you will get some std::bad_alloc errors at runtime.
				33
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	34	@section S1_file_organisation File organisation
				35
				36	This archive contains:
				37	- The arm_compute header and source files
				38	- The latest Khronos OpenCL 1.2 C headers from the <a href="https://www.khronos.org/registry/cl/">Khronos OpenCL registry</a>
				39	- The latest Khronos cl2.hpp from the <a href="https://www.khronos.org/registry/cl/">Khronos OpenCL registry</a> (API version 2.1 when this document was written)
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	40	- The latest Khronos OpenGL ES 3.1 C headers from the <a href="https://www.khronos.org/registry/gles/">Khronos OpenGL ES registry</a>
				41	- The latest Khronos EGL 1.5 C headers from the <a href="https://www.khronos.org/registry/gles/">Khronos EGL registry</a>
				42	- The sources for a stub version of libOpenCL.so, libGLESv1_CM.so, libGLESv2.so and libEGL.so to help you build your application.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	43	- An examples folder containing a few examples to compile and link against the library.
				44	- A @ref utils folder containing headers with some boiler plate code used by the examples.
				45	- This documentation.
				46
				47	You should have the following file organisation:
				48
				49	.
				50	├── arm_compute --> All the arm_compute headers
				51	│ ├── core
				52	│ │ ├── CL
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	53	│ │ │ ├── CLKernelLibrary.h --> Manages all the OpenCL kernels compilation and caching, provides accessors for the OpenCL Context.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	54	│ │ │ ├── CLKernels.h --> Includes all the OpenCL kernels at once
				55	│ │ │ ├── CL specialisation of all the generic objects interfaces (ICLTensor, ICLImage, etc.)
				56	│ │ │ ├── kernels --> Folder containing all the OpenCL kernels
				57	│ │ │ │ └── CL*Kernel.h
				58	│ │ │ └── OpenCL.h --> Wrapper to configure the Khronos OpenCL C++ header
				59	│ │ ├── CPP
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	60	│ │ │ ├── CPPKernels.h --> Includes all the CPP kernels at once
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	61	│ │ │ └── kernels --> Folder containing all the CPP kernels
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	62	│ │ │ └── CPP*Kernel.h
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	63	│ │ ├── GLES_COMPUTE
				64	│ │ │ ├── GCKernelLibrary.h --> Manages all the GLES kernels compilation and caching, provides accessors for the GLES Context.
				65	│ │ │ ├── GCKernels.h --> Includes all the GLES kernels at once
				66	│ │ │ ├── GLES specialisation of all the generic objects interfaces (IGCTensor, IGCImage, etc.)
				67	│ │ │ ├── kernels --> Folder containing all the GLES kernels
				68	│ │ │ │ └── GC*Kernel.h
				69	│ │ │ └── OpenGLES.h --> Wrapper to configure the Khronos EGL and OpenGL ES C header
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	70	│ │ ├── NEON
				71	│ │ │ ├── kernels --> Folder containing all the NEON kernels
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	72	│ │ │ │ ├── arm64 --> Folder containing the interfaces for the assembly arm64 NEON kernels
				73	│ │ │ │ ├── arm32 --> Folder containing the interfaces for the assembly arm32 NEON kernels
				74	│ │ │ │ ├── assembly --> Folder containing the NEON assembly routines.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	75	│ │ │ │ └── NE*Kernel.h
				76	│ │ │ └── NEKernels.h --> Includes all the NEON kernels at once
				77	│ │ ├── All common basic types (Types.h, Window, Coordinates, Iterator, etc.)
				78	│ │ ├── All generic objects interfaces (ITensor, IImage, etc.)
				79	│ │ └── Objects metadata classes (ImageInfo, TensorInfo, MultiImageInfo)
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	80	│ ├── graph
				81	│ │ ├── CL --> OpenCL specific operations
				82	│ │ │ └── CLMap.h / CLUnmap.h
				83	│ │ ├── nodes
				84	│ │ │ └── The various nodes supported by the graph API
				85	│ │ ├── Nodes.h --> Includes all the Graph nodes at once.
				86	│ │ └── Graph objects ( INode, ITensorAccessor, Graph, etc.)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	87	│ └── runtime
				88	│ ├── CL
				89	│ │ ├── CL objects & allocators (CLArray, CLImage, CLTensor, etc.)
				90	│ │ ├── functions --> Folder containing all the OpenCL functions
				91	│ │ │ └── CL*.h
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	92	│ │ ├── CLScheduler.h --> Interface to enqueue OpenCL kernels and get/set the OpenCL CommandQueue and ICLTuner.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	93	│ │ └── CLFunctions.h --> Includes all the OpenCL functions at once
				94	│ ├── CPP
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	95	│ │ ├── CPPKernels.h --> Includes all the CPP functions at once.
				96	│ │ └── CPPScheduler.h --> Basic pool of threads to execute CPP/NEON code on several cores in parallel
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	97	│ ├── GLES_COMPUTE
				98	│ │ ├── GLES objects & allocators (GCArray, GCImage, GCTensor, etc.)
				99	│ │ ├── functions --> Folder containing all the GLES functions
				100	│ │ │ └── GC*.h
				101	│ │ ├── GCScheduler.h --> Interface to enqueue GLES kernels and get/set the GLES CommandQueue.
				102	│ │ └── GCFunctions.h --> Includes all the GLES functions at once
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	103	│ ├── NEON
				104	│ │ ├── functions --> Folder containing all the NEON functions
				105	│ │ │ └── NE*.h
				106	│ │ └── NEFunctions.h --> Includes all the NEON functions at once
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	107	│ ├── OMP
				108	│ │ └── OMPScheduler.h --> OpenMP scheduler (Alternative to the CPPScheduler)
				109	│ ├── Memory manager files (LifetimeManager, PoolManager, etc.)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	110	│ └── Basic implementations of the generic object interfaces (Array, Image, Tensor, etc.)
				111	├── documentation
				112	│ ├── index.xhtml
				113	│ └── ...
				114	├── documentation.xhtml -> documentation/index.xhtml
				115	├── examples
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	116	│ ├── cl_*.cpp --> OpenCL examples
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	117	│ ├── gc_*.cpp --> GLES compute shaders examples
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	118	│ ├── graph_*.cpp --> Graph examples
				119	│ ├── neoncl_*.cpp --> NEON / OpenCL interoperability examples
				120	│ └── neon_*.cpp --> NEON examples
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	121	├── include
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	122	│ ├── CL
				123	│ │ └── Khronos OpenCL C headers and C++ wrapper
				124	│ ├── half --> FP16 library available from http://half.sourceforge.net
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	125	│ ├── libnpy --> Library to load / write npy buffers, available from https://github.com/llohse/libnpy
				126	│ └── linux --> Headers only needed for Linux builds
				127	│ └── Khronos EGL and OpenGLES headers
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	128	├── opencl-1.2-stubs
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	129	│ └── opencl_stubs.c --> OpenCL stubs implementation
				130	├── opengles-3.1-stubs
				131	│ ├── EGL.c --> EGL stubs implementation
				132	│ └── GLESv2.c --> GLESv2 stubs implementation
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	133	├── scripts
				134	│ ├── caffe_data_extractor.py --> Basic script to export weights from Caffe to npy files
				135	│ └── tensorflow_data_extractor.py --> Basic script to export weights from Tensor Flow to npy files
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	136	├── src
				137	│ ├── core
				138	│ │ └── ... (Same structure as headers)
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	139	│ │ ├── CL
				140	│ │ │ └── cl_kernels --> All the OpenCL kernels
				141	│ │ └── GLES_COMPUTE
				142	│ │ └── cs_shaders --> All the OpenGL ES Compute Shaders
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	143	│ ├── graph
				144	│ │ └── ... (Same structure as headers)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	145	│ └── runtime
				146	│ └── ... (Same structure as headers)
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	147	├── support
				148	│ └── Various headers to work around toolchains / platform issues.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	149	├── tests
				150	│ ├── All test related files shared between validation and benchmark
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	151	│ ├── CL --> OpenCL accessors
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	152	│ ├── GLES_COMPUTE --> GLES accessors
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	153	│ ├── NEON --> NEON accessors
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	154	│ ├── benchmark --> Sources for benchmarking
				155	│ │ ├── Benchmark specific files
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	156	│ │ ├── CL --> OpenCL benchmarking tests
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	157	│ │ ├── GLES_COMPUTE --> GLES benchmarking tests
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	158	│ │ └── NEON --> NEON benchmarking tests
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	159	│ ├── datasets
				160	│ │ └── Datasets for all the validation / benchmark tests, layer configurations for various networks, etc.
				161	│ ├── framework
				162	│ │ └── Boiler plate code for both validation and benchmark test suites (Command line parsers, instruments, output loggers, etc.)
				163	│ ├── networks
				164	│ │ └── Examples of how to instantiate networks.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	165	│ ├── validation --> Sources for validation
				166	│ │ ├── Validation specific files
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	167	│ │ ├── CL --> OpenCL validation tests
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	168	│ │ ├── GLES_COMPUTE --> GLES validation tests
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	169	│ │ ├── CPP --> C++ reference implementations
				170	│ │ ├── fixtures
				171	│ │ │ └── Fixtures to initialise and run the runtime Functions.
				172	│ │ └── NEON --> NEON validation tests
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	173	│ └── dataset --> Datasets defining common sets of input parameters
				174	└── utils --> Boiler plate code used by examples
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	175	└── Various utilities to print types, load / store assets, etc.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	176
				177	@section S2_versions_changelog Release versions and changelog
				178
				179	@subsection S2_1_versions Release versions
				180
				181	All releases are numbered vYY.MM Where YY are the last two digits of the year, and MM the month number.
				182	If there is more than one release in a month then an extra sequential number is appended at the end:
				183
				184	v17.03 (First release of March 2017)
				185	v17.03.1 (Second release of March 2017)
				186	v17.04 (First release of April 2017)
				187
				188	@note We're aiming at releasing one major public release with new features per quarter. All releases in between will only contain bug fixes.
				189
				190	@subsection S2_2_changelog Changelog
				191
Anthony Barbier	64c95a0	2018-01-22 18:48:55 +0000	[diff] [blame^]	192	v18.01 Public maintenance release
				193	- Various bug fixes
				194	- Added some of the missing validate() methods
				195	- Added @ref arm_compute::CLDeconvolutionLayerUpsampleKernel / @ref arm_compute::CLDeconvolutionLayer @ref arm_compute::CLDeconvolutionLayerUpsample
				196	- Added @ref arm_compute::CLPermuteKernel / @ref arm_compute::CLPermute
				197	- Added method to clean the programs cache in the CL Kernel library.
				198	- Added @ref arm_compute::GCArithmeticAdditionKernel / @ref arm_compute::GCArithmeticAddition
				199	- Added @ref arm_compute::GCDepthwiseConvolutionLayer3x3Kernel / @ref arm_compute::GCDepthwiseConvolutionLayer3x3
				200	- Added @ref arm_compute::GCNormalizePlanarYUVLayerKernel / @ref arm_compute::GCNormalizePlanarYUVLayer
				201	- Added @ref arm_compute::GCScaleKernel / @ref arm_compute::GCScale
				202	- Added @ref arm_compute::GCWeightsReshapeKernel / @ref arm_compute::GCConvolutionLayer
				203	- Added FP16 support to the following GLES compute kernels:
				204	- @ref arm_compute::GCCol2ImKernel
				205	- @ref arm_compute::GCGEMMInterleave4x4Kernel
				206	- @ref arm_compute::GCGEMMTranspose1xWKernel
				207	- @ref arm_compute::GCIm2ColKernel
				208	- Refactored NEON Winograd (@ref arm_compute::NEWinogradLayerKernel)
				209	- Added @ref arm_compute::NEDirectConvolutionLayerOutputStageKernel
				210	- Added QASYMM8 support to the following NEON kernels:
				211	- @ref arm_compute::NEDepthwiseConvolutionLayer3x3Kernel
				212	- @ref arm_compute::NEFillBorderKernel
				213	- @ref arm_compute::NEPoolingLayerKernel
				214	- Added new examples:
				215	- graph_cl_mobilenet_qasymm8.cpp
				216	- graph_inception_v3.cpp
				217	- gc_dc.cpp
				218	- More tests added to both validation and benchmarking suites.
				219
Gian Marco	ff85093	2017-12-11 12:37:17 +0000	[diff] [blame]	220	v17.12 Public major release
				221	- Most machine learning functions on OpenCL support the new data type QASYMM8
				222	- Introduced logging interface
				223	- Introduced opencl timer
				224	- Reworked GEMMLowp interface
				225	- Added new NEON assembly kernels for GEMMLowp, SGEMM and HGEMM
				226	- Added validation method for most Machine Learning kernels / functions
				227	- Added new graph examples such as googlenet, mobilenet, squeezenet, vgg16 and vgg19
				228	- Added sgemm example for OpenCL
				229	- Added absolute difference example for GLES compute
				230	- Added new tests and benchmarks in validation and benchmark frameworks
				231	- Added new kernels / functions for GLES compute
				232
				233	- New OpenGL ES kernels / functions
				234	- @ref arm_compute::GCAbsoluteDifferenceKernel / @ref arm_compute::GCAbsoluteDifference
				235	- @ref arm_compute::GCActivationLayerKernel / @ref arm_compute::GCActivationLayer
				236	- @ref arm_compute::GCBatchNormalizationLayerKernel / @ref arm_compute::GCBatchNormalizationLayer
				237	- @ref arm_compute::GCCol2ImKernel
				238	- @ref arm_compute::GCDepthConcatenateLayerKernel / @ref arm_compute::GCDepthConcatenateLayer
				239	- @ref arm_compute::GCDirectConvolutionLayerKernel / @ref arm_compute::GCDirectConvolutionLayer
				240	- @ref arm_compute::GCDropoutLayerKernel / @ref arm_compute::GCDropoutLayer
				241	- @ref arm_compute::GCFillBorderKernel / @ref arm_compute::GCFillBorder
				242	- @ref arm_compute::GCGEMMInterleave4x4Kernel / @ref arm_compute::GCGEMMInterleave4x4
				243	- @ref arm_compute::GCGEMMMatrixAccumulateBiasesKernel / @ref arm_compute::GCGEMMMatrixAdditionKernel / @ref arm_compute::GCGEMMMatrixMultiplyKernel / @ref arm_compute::GCGEMM
				244	- @ref arm_compute::GCGEMMTranspose1xWKernel / @ref arm_compute::GCGEMMTranspose1xW
				245	- @ref arm_compute::GCIm2ColKernel
				246	- @ref arm_compute::GCNormalizationLayerKernel / @ref arm_compute::GCNormalizationLayer
				247	- @ref arm_compute::GCPixelWiseMultiplicationKernel / @ref arm_compute::GCPixelWiseMultiplication
				248	- @ref arm_compute::GCPoolingLayerKernel / @ref arm_compute::GCPoolingLayer
				249	- @ref arm_compute::GCLogits1DMaxKernel / @ref arm_compute::GCLogits1DShiftExpSumKernel / @ref arm_compute::GCLogits1DNormKernel / @ref arm_compute::GCSoftmaxLayer
				250	- @ref arm_compute::GCTransposeKernel / @ref arm_compute::GCTranspose
				251
				252	- New NEON kernels / functions
				253	- @ref arm_compute::NEGEMMLowpAArch64A53Kernel / @ref arm_compute::NEGEMMLowpAArch64Kernel / @ref arm_compute::NEGEMMLowpAArch64V8P4Kernel / arm_compute::NEGEMMInterleavedBlockedKernel / @ref arm_compute::NEGEMMLowpAssemblyMatrixMultiplyCore
				254	- @ref arm_compute::NEHGEMMAArch64FP16Kernel
				255	- @ref arm_compute::NEDepthwiseConvolutionLayer3x3Kernel / @ref arm_compute::NEDepthwiseIm2ColKernel / @ref arm_compute::NEGEMMMatrixVectorMultiplyKernel / @ref arm_compute::NEDepthwiseVectorToTensorKernel / @ref arm_compute::NEDepthwiseConvolutionLayer
				256	- @ref arm_compute::NEGEMMLowpOffsetContributionKernel / @ref arm_compute::NEGEMMLowpMatrixAReductionKernel / @ref arm_compute::NEGEMMLowpMatrixBReductionKernel / @ref arm_compute::NEGEMMLowpMatrixMultiplyCore
				257	- @ref arm_compute::NEGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel / @ref arm_compute::NEGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint
				258	- @ref arm_compute::NEGEMMLowpQuantizeDownInt32ToUint8ScaleKernel / @ref arm_compute::NEGEMMLowpQuantizeDownInt32ToUint8Scale
				259	- @ref arm_compute::NEWinogradLayerKernel / @ref arm_compute::NEWinogradLayer
				260
				261	- New OpenCL kernels / functions
				262	- @ref arm_compute::CLGEMMLowpOffsetContributionKernel / @ref arm_compute::CLGEMMLowpMatrixAReductionKernel / @ref arm_compute::CLGEMMLowpMatrixBReductionKernel / @ref arm_compute::CLGEMMLowpMatrixMultiplyCore
				263	- @ref arm_compute::CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel / @ref arm_compute::CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint
				264	- @ref arm_compute::CLGEMMLowpQuantizeDownInt32ToUint8ScaleKernel / @ref arm_compute::CLGEMMLowpQuantizeDownInt32ToUint8Scale
				265
				266	- New graph nodes for NEON and OpenCL
				267	- @ref arm_compute::graph::BranchLayer
				268	- @ref arm_compute::graph::DepthConvertLayer
				269	- @ref arm_compute::graph::DepthwiseConvolutionLayer
				270	- @ref arm_compute::graph::DequantizationLayer
				271	- @ref arm_compute::graph::FlattenLayer
				272	- @ref arm_compute::graph::QuantizationLayer
				273	- @ref arm_compute::graph::ReshapeLayer
				274
Anthony Barbier	3c5b4ff	2017-10-12 13:20:52 +0100	[diff] [blame]	275	v17.10 Public maintenance release
				276	- Bug fixes:
				277	- Check the maximum local workgroup size supported by OpenCL devices
				278	- Minor documentation updates (Fixed instructions to build the examples)
				279	- Introduced a arm_compute::graph::GraphContext
				280	- Added a few new Graph nodes, support for branches and grouping.
				281	- Automatically enable cl_printf in debug builds
				282	- Fixed bare metal builds for armv7a
				283	- Added AlexNet and cartoon effect examples
				284	- Fixed library builds: libraries are no longer built as supersets of each other.(It means application using the Runtime part of the library now need to link against both libarm_compute_core and libarm_compute)
				285
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	286	v17.09 Public major release
				287	- Experimental Graph support: initial implementation of a simple stream API to easily chain machine learning layers.
				288	- Memory Manager (@ref arm_compute::BlobLifetimeManager, @ref arm_compute::BlobMemoryPool, @ref arm_compute::ILifetimeManager, @ref arm_compute::IMemoryGroup, @ref arm_compute::IMemoryManager, @ref arm_compute::IMemoryPool, @ref arm_compute::IPoolManager, @ref arm_compute::MemoryManagerOnDemand, @ref arm_compute::PoolManager)
				289	- New validation and benchmark frameworks (Boost and Google frameworks replaced by homemade framework).
				290	- Most machine learning functions support both fixed point 8 and 16 bit (QS8, QS16) for both NEON and OpenCL.
				291	- New NEON kernels / functions:
				292	- @ref arm_compute::NEGEMMAssemblyBaseKernel @ref arm_compute::NEGEMMAArch64Kernel
				293	- @ref arm_compute::NEDequantizationLayerKernel / @ref arm_compute::NEDequantizationLayer
				294	- @ref arm_compute::NEFloorKernel / @ref arm_compute::NEFloor
Giorgio Arena	04a8f8c	2017-11-23 11:45:24 +0000	[diff] [blame]	295	- @ref arm_compute::NEL2NormalizeLayerKernel / @ref arm_compute::NEL2NormalizeLayer
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	296	- @ref arm_compute::NEQuantizationLayerKernel @ref arm_compute::NEMinMaxLayerKernel / @ref arm_compute::NEQuantizationLayer
				297	- @ref arm_compute::NEROIPoolingLayerKernel / @ref arm_compute::NEROIPoolingLayer
				298	- @ref arm_compute::NEReductionOperationKernel / @ref arm_compute::NEReductionOperation
				299	- @ref arm_compute::NEReshapeLayerKernel / @ref arm_compute::NEReshapeLayer
				300
				301	- New OpenCL kernels / functions:
Giorgio Arena	04a8f8c	2017-11-23 11:45:24 +0000	[diff] [blame]	302	- @ref arm_compute::CLDepthwiseConvolutionLayer3x3Kernel @ref arm_compute::CLDepthwiseIm2ColKernel @ref arm_compute::CLDepthwiseVectorToTensorKernel @ref arm_compute::CLDepthwiseWeightsReshapeKernel / @ref arm_compute::CLDepthwiseConvolutionLayer3x3 @ref arm_compute::CLDepthwiseConvolutionLayer @ref arm_compute::CLDepthwiseSeparableConvolutionLayer
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	303	- @ref arm_compute::CLDequantizationLayerKernel / @ref arm_compute::CLDequantizationLayer
				304	- @ref arm_compute::CLDirectConvolutionLayerKernel / @ref arm_compute::CLDirectConvolutionLayer
				305	- @ref arm_compute::CLFlattenLayer
				306	- @ref arm_compute::CLFloorKernel / @ref arm_compute::CLFloor
				307	- @ref arm_compute::CLGEMMTranspose1xW
				308	- @ref arm_compute::CLGEMMMatrixVectorMultiplyKernel
Giorgio Arena	04a8f8c	2017-11-23 11:45:24 +0000	[diff] [blame]	309	- @ref arm_compute::CLL2NormalizeLayerKernel / @ref arm_compute::CLL2NormalizeLayer
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	310	- @ref arm_compute::CLQuantizationLayerKernel @ref arm_compute::CLMinMaxLayerKernel / @ref arm_compute::CLQuantizationLayer
				311	- @ref arm_compute::CLROIPoolingLayerKernel / @ref arm_compute::CLROIPoolingLayer
				312	- @ref arm_compute::CLReductionOperationKernel / @ref arm_compute::CLReductionOperation
				313	- @ref arm_compute::CLReshapeLayerKernel / @ref arm_compute::CLReshapeLayer
				314
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	315	v17.06 Public major release
				316	- Various bug fixes
				317	- Added support for fixed point 8 bit (QS8) to the various NEON machine learning kernels.
				318	- Added unit tests and benchmarks (AlexNet, LeNet)
				319	- Added support for sub tensors.
				320	- Added infrastructure to provide GPU specific optimisation for some OpenCL kernels.
				321	- Added @ref arm_compute::OMPScheduler (OpenMP) scheduler for NEON
				322	- Added @ref arm_compute::SingleThreadScheduler scheduler for NEON (For bare metal)
				323	- User can specify his own scheduler by implementing the @ref arm_compute::IScheduler interface.
				324	- New OpenCL kernels / functions:
				325	- @ref arm_compute::CLBatchNormalizationLayerKernel / @ref arm_compute::CLBatchNormalizationLayer
Giorgio Arena	04a8f8c	2017-11-23 11:45:24 +0000	[diff] [blame]	326	- @ref arm_compute::CLDepthConcatenateLayerKernel / @ref arm_compute::CLDepthConcatenateLayer
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	327	- @ref arm_compute::CLHOGOrientationBinningKernel @ref arm_compute::CLHOGBlockNormalizationKernel, @ref arm_compute::CLHOGDetectorKernel / @ref arm_compute::CLHOGDescriptor @ref arm_compute::CLHOGDetector @ref arm_compute::CLHOGGradient @ref arm_compute::CLHOGMultiDetection
				328	- @ref arm_compute::CLLocallyConnectedMatrixMultiplyKernel / @ref arm_compute::CLLocallyConnectedLayer
				329	- @ref arm_compute::CLWeightsReshapeKernel / @ref arm_compute::CLConvolutionLayerReshapeWeights
				330	- New C++ kernels:
				331	- @ref arm_compute::CPPDetectionWindowNonMaximaSuppressionKernel
				332	- New NEON kernels / functions:
				333	- @ref arm_compute::NEBatchNormalizationLayerKernel / @ref arm_compute::NEBatchNormalizationLayer
Giorgio Arena	04a8f8c	2017-11-23 11:45:24 +0000	[diff] [blame]	334	- @ref arm_compute::NEDepthConcatenateLayerKernel / @ref arm_compute::NEDepthConcatenateLayer
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	335	- @ref arm_compute::NEDirectConvolutionLayerKernel / @ref arm_compute::NEDirectConvolutionLayer
				336	- @ref arm_compute::NELocallyConnectedMatrixMultiplyKernel / @ref arm_compute::NELocallyConnectedLayer
				337	- @ref arm_compute::NEWeightsReshapeKernel / @ref arm_compute::NEConvolutionLayerReshapeWeights
				338
				339	v17.05 Public bug fixes release
				340	- Various bug fixes
				341	- Remaining of the functions ported to use accurate padding.
				342	- Library does not link against OpenCL anymore (It uses dlopen / dlsym at runtime instead to determine whether or not OpenCL is available).
				343	- Added "free" method to allocator.
				344	- Minimum version of g++ required for armv7 Linux changed from 4.8 to 4.9
				345
				346	v17.04 Public bug fixes release
				347
				348	The following functions have been ported to use the new accurate padding:
				349	- @ref arm_compute::CLColorConvertKernel
				350	- @ref arm_compute::CLEdgeNonMaxSuppressionKernel
				351	- @ref arm_compute::CLEdgeTraceKernel
				352	- @ref arm_compute::CLGaussianPyramidHorKernel
				353	- @ref arm_compute::CLGaussianPyramidVertKernel
				354	- @ref arm_compute::CLGradientKernel
				355	- @ref arm_compute::NEChannelCombineKernel
				356	- @ref arm_compute::NEFillArrayKernel
				357	- @ref arm_compute::NEGaussianPyramidHorKernel
				358	- @ref arm_compute::NEGaussianPyramidVertKernel
				359	- @ref arm_compute::NEHarrisScoreFP16Kernel
				360	- @ref arm_compute::NEHarrisScoreKernel
				361	- @ref arm_compute::NEHOGDetectorKernel
				362	- @ref arm_compute::NELogits1DMaxKernel
				363	- @ref arm_compute::NELogits1DShiftExpSumKernel
				364	- @ref arm_compute::NELogits1DNormKernel
				365	- @ref arm_compute::NENonMaximaSuppression3x3FP16Kernel
				366	- @ref arm_compute::NENonMaximaSuppression3x3Kernel
				367
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	368	v17.03.1 First Major public release of the sources
				369	- Renamed the library to arm_compute
				370	- New CPP target introduced for C++ kernels shared between NEON and CL functions.
				371	- New padding calculation interface introduced and ported most kernels / functions to use it.
				372	- New OpenCL kernels / functions:
Gian Marco	05288a2	2017-11-21 10:57:50 +0000	[diff] [blame]	373	- @ref arm_compute::CLGEMMLowpMatrixMultiplyKernel / arm_compute::CLGEMMLowp
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	374	- New NEON kernels / functions:
				375	- @ref arm_compute::NENormalizationLayerKernel / @ref arm_compute::NENormalizationLayer
				376	- @ref arm_compute::NETransposeKernel / @ref arm_compute::NETranspose
				377	- @ref arm_compute::NELogits1DMaxKernel, @ref arm_compute::NELogits1DShiftExpSumKernel, @ref arm_compute::NELogits1DNormKernel / @ref arm_compute::NESoftmaxLayer
				378	- @ref arm_compute::NEIm2ColKernel, @ref arm_compute::NECol2ImKernel, arm_compute::NEConvolutionLayerWeightsReshapeKernel / @ref arm_compute::NEConvolutionLayer
				379	- @ref arm_compute::NEGEMMMatrixAccumulateBiasesKernel / @ref arm_compute::NEFullyConnectedLayer
Gian Marco	e75a02b	2017-11-08 12:24:09 +0000	[diff] [blame]	380	- @ref arm_compute::NEGEMMLowpMatrixMultiplyKernel / arm_compute::NEGEMMLowp
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	381
				382	v17.03 Sources preview
				383	- New OpenCL kernels / functions:
				384	- @ref arm_compute::CLGradientKernel, @ref arm_compute::CLEdgeNonMaxSuppressionKernel, @ref arm_compute::CLEdgeTraceKernel / @ref arm_compute::CLCannyEdge
				385	- GEMM refactoring + FP16 support: @ref arm_compute::CLGEMMInterleave4x4Kernel, @ref arm_compute::CLGEMMTranspose1xWKernel, @ref arm_compute::CLGEMMMatrixMultiplyKernel, @ref arm_compute::CLGEMMMatrixAdditionKernel / @ref arm_compute::CLGEMM
				386	- @ref arm_compute::CLGEMMMatrixAccumulateBiasesKernel / @ref arm_compute::CLFullyConnectedLayer
				387	- @ref arm_compute::CLTransposeKernel / @ref arm_compute::CLTranspose
				388	- @ref arm_compute::CLLKTrackerInitKernel, @ref arm_compute::CLLKTrackerStage0Kernel, @ref arm_compute::CLLKTrackerStage1Kernel, @ref arm_compute::CLLKTrackerFinalizeKernel / @ref arm_compute::CLOpticalFlow
				389	- @ref arm_compute::CLNormalizationLayerKernel / @ref arm_compute::CLNormalizationLayer
				390	- @ref arm_compute::CLLaplacianPyramid, @ref arm_compute::CLLaplacianReconstruct
				391	- New NEON kernels / functions:
				392	- @ref arm_compute::NEActivationLayerKernel / @ref arm_compute::NEActivationLayer
				393	- GEMM refactoring + FP16 support (Requires armv8.2 CPU): @ref arm_compute::NEGEMMInterleave4x4Kernel, @ref arm_compute::NEGEMMTranspose1xWKernel, @ref arm_compute::NEGEMMMatrixMultiplyKernel, @ref arm_compute::NEGEMMMatrixAdditionKernel / @ref arm_compute::NEGEMM
				394	- @ref arm_compute::NEPoolingLayerKernel / @ref arm_compute::NEPoolingLayer
				395
				396	v17.02.1 Sources preview
				397	- New OpenCL kernels / functions:
				398	- @ref arm_compute::CLLogits1DMaxKernel, @ref arm_compute::CLLogits1DShiftExpSumKernel, @ref arm_compute::CLLogits1DNormKernel / @ref arm_compute::CLSoftmaxLayer
				399	- @ref arm_compute::CLPoolingLayerKernel / @ref arm_compute::CLPoolingLayer
Gian Marco Iodice	5cb4c42	2017-06-23 10:38:25 +0100	[diff] [blame]	400	- @ref arm_compute::CLIm2ColKernel, @ref arm_compute::CLCol2ImKernel, arm_compute::CLConvolutionLayerWeightsReshapeKernel / @ref arm_compute::CLConvolutionLayer
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	401	- @ref arm_compute::CLRemapKernel / @ref arm_compute::CLRemap
				402	- @ref arm_compute::CLGaussianPyramidHorKernel, @ref arm_compute::CLGaussianPyramidVertKernel / @ref arm_compute::CLGaussianPyramid, @ref arm_compute::CLGaussianPyramidHalf, @ref arm_compute::CLGaussianPyramidOrb
				403	- @ref arm_compute::CLMinMaxKernel, @ref arm_compute::CLMinMaxLocationKernel / @ref arm_compute::CLMinMaxLocation
				404	- @ref arm_compute::CLNonLinearFilterKernel / @ref arm_compute::CLNonLinearFilter
				405	- New NEON FP16 kernels (Requires armv8.2 CPU)
				406	- @ref arm_compute::NEAccumulateWeightedFP16Kernel
				407	- @ref arm_compute::NEBox3x3FP16Kernel
				408	- @ref arm_compute::NENonMaximaSuppression3x3FP16Kernel
				409
				410	v17.02 Sources preview
				411	- New OpenCL kernels / functions:
				412	- @ref arm_compute::CLActivationLayerKernel / @ref arm_compute::CLActivationLayer
				413	- @ref arm_compute::CLChannelCombineKernel / @ref arm_compute::CLChannelCombine
				414	- @ref arm_compute::CLDerivativeKernel / @ref arm_compute::CLChannelExtract
				415	- @ref arm_compute::CLFastCornersKernel / @ref arm_compute::CLFastCorners
				416	- @ref arm_compute::CLMeanStdDevKernel / @ref arm_compute::CLMeanStdDev
				417	- New NEON kernels / functions:
				418	- HOG / SVM: @ref arm_compute::NEHOGOrientationBinningKernel, @ref arm_compute::NEHOGBlockNormalizationKernel, @ref arm_compute::NEHOGDetectorKernel, arm_compute::NEHOGNonMaximaSuppressionKernel / @ref arm_compute::NEHOGDescriptor, @ref arm_compute::NEHOGDetector, @ref arm_compute::NEHOGGradient, @ref arm_compute::NEHOGMultiDetection
				419	- @ref arm_compute::NENonLinearFilterKernel / @ref arm_compute::NENonLinearFilter
				420	- Introduced a CLScheduler to manage the default context and command queue used by the runtime library and create synchronisation events.
				421	- Switched all the kernels / functions to use tensors instead of images.
				422	- Updated documentation to include instructions to build the library from sources.
				423
				424	v16.12 Binary preview release
				425	- Original release
				426
				427	@section S3_how_to_build How to build the library and the examples
				428
				429	@subsection S3_1_build_options Build options
				430
				431	scons 2.3 or above is required to build the library.
				432	To see the build options available simply run ```scons -h```:
				433
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	434	debug: Debug (yes\|no)
				435	default: False
				436	actual: False
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	437
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	438	asserts: Enable asserts (this flag is forced to 1 for debug=1) (yes\|no)
				439	default: False
				440	actual: False
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	441
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	442	arch: Target Architecture (armv7a\|arm64-v8a\|arm64-v8.2-a\|x86_32\|x86_64)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	443	default: armv7a
				444	actual: armv7a
				445
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	446	os: Target OS (linux\|android\|bare_metal)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	447	default: linux
				448	actual: linux
				449
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	450	build: Build type (native\|cross_compile)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	451	default: cross_compile
				452	actual: cross_compile
				453
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	454	examples: Build example programs (yes\|no)
				455	default: True
				456	actual: True
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	457
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	458	Werror: Enable/disable the -Werror compilation flag (yes\|no)
				459	default: True
				460	actual: True
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	461
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	462	opencl: Enable OpenCL support (yes\|no)
				463	default: True
				464	actual: True
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	465
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	466	neon: Enable Neon support (yes\|no)
				467	default: False
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	468	actual: False
				469
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	470	gles_compute: Enable OpenGL ES Compute Shader support (yes\|no)
				471	default: False
				472	actual: False
				473
				474	embed_kernels: Embed OpenCL kernels and OpenGL ES compute shader in library binary (yes\|no)
Anthony Barbier	cc0a80b	2017-12-15 11:37:29 +0000	[diff] [blame]	475	default: True
				476	actual: True
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	477
				478	set_soname: Set the library's soname and shlibversion (requires SCons 2.4 or above) (yes\|no)
				479	default: False
				480	actual: False
				481
				482	openmp: Enable OpenMP backend (yes\|no)
				483	default: False
				484	actual: False
				485
				486	cppthreads: Enable C++11 threads backend (yes\|no)
				487	default: True
				488	actual: True
				489
				490	build_dir: Specify sub-folder for the build ( /path/to/build_dir )
				491	default: .
				492	actual: .
				493
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	494	extra_cxx_flags: Extra CXX flags to be appended to the build command
				495	default:
				496	actual:
				497
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	498	pmu: Enable PMU counters (yes\|no)
				499	default: False
				500	actual: False
				501
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	502	mali: Enable Mali hardware counters (yes\|no)
				503	default: False
				504	actual: False
				505
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	506	validation_tests: Build validation test programs (yes\|no)
				507	default: False
				508	actual: False
				509
				510	benchmark_tests: Build benchmark test programs (yes\|no)
				511	default: False
				512	actual: False
				513
				514	@b debug / @b asserts:
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	515	- With debug=1 asserts are enabled, and the library is built with symbols and no optimisations enabled.
				516	- With debug=0 and asserts=1: Optimisations are enabled and symbols are removed, however all the asserts are still present (This is about 20% slower than the release build)
				517	- With debug=0 and asserts=0: All optimisations are enable and no validation is performed, if the application misuses the library it is likely to result in a crash. (Only use this mode once you are sure your application is working as expected).
				518
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	519	@b arch: The x86_32 and x86_64 targets can only be used with neon=0 and opencl=1.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	520
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	521	@b os: Choose the operating system you are targeting: Linux, Android or bare metal.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	522	@note bare metal can only be used for NEON (not OpenCL), only static libraries get built and NEON's multi-threading support is disabled.
				523
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	524	@b build: you can either build directly on your device (native) or cross compile from your desktop machine (cross-compile). In both cases make sure the compiler is available in your path.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	525
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	526	@note If you want to natively compile for 32bit on a 64bit ARM device running a 64bit OS then you will have to use cross-compile too.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	527
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	528	@b Werror: If you are compiling using the same toolchains as the ones used in this guide then there shouldn't be any warning and therefore you should be able to keep Werror=1. If with a different compiler version the library fails to build because of warnings interpreted as errors then, if you are sure the warnings are not important, you might want to try to build with Werror=0 (But please do report the issue either on Github or by an email to developer@arm.com so that the issue can be addressed).
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	529
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	530	@b opencl / @b neon / @b gles_compute: Choose which SIMD technology you want to target. (NEON for ARM Cortex-A CPUs or OpenCL / GLES_COMPUTE for ARM Mali GPUs)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	531
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	532	@b embed_kernels: For OpenCL / GLES_COMPUTE only: set embed_kernels=1 if you want the OpenCL / GLES_COMPUTE kernels to be built in the library's binaries instead of being read from separate ".cl" / ".cs" files. If embed_kernels is set to 0 then the application can set the path to the folder containing the OpenCL / GLES_COMPUTE kernel files by calling CLKernelLibrary::init() / GCKernelLibrary::init(). By default the path is set to "./cl_kernels" / "./cs_shaders".
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	533
				534	@b set_soname: Do you want to build the versioned version of the library ?
				535
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	536	If enabled the library will contain a SONAME and SHLIBVERSION and some symlinks will automatically be created between the objects.
				537	Example:
				538	libarm_compute_core.so -> libarm_compute_core.so.1.0.0
				539	libarm_compute_core.so.1 -> libarm_compute_core.so.1.0.0
				540	libarm_compute_core.so.1.0.0
				541
				542	@note This options is disabled by default as it requires SCons version 2.4 or above.
				543
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	544	@b extra_cxx_flags: Custom CXX flags which will be appended to the end of the build command.
				545
				546	@b build_dir: Build the library in a subfolder of the "build" folder. (Allows to build several configurations in parallel).
				547
				548	@b examples: Build or not the examples
				549
				550	@b validation_tests: Enable the build of the validation suite.
				551
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	552	@b benchmark_tests: Enable the build of the benchmark tests
				553
				554	@b pmu: Enable the PMU cycle counter to measure execution time in benchmark tests. (Your device needs to support it)
				555
Anthony Barbier	6a5627a	2017-09-26 14:42:02 +0100	[diff] [blame]	556	@b mali: Enable the collection of Mali hardware counters to measure execution time in benchmark tests. (Your device needs to have a Mali driver that supports it)
				557
Anthony Barbier	79c6178	2017-06-23 11:48:24 +0100	[diff] [blame]	558	@b openmp Build in the OpenMP scheduler for NEON.
				559
				560	@note Only works when building with g++ not clang++
				561
				562	@b cppthreads Build in the C++11 scheduler for NEON.
				563
				564	@sa arm_compute::Scheduler::set
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	565
Moritz Pflanzer	07674de	2017-07-21 09:39:36 +0100	[diff] [blame]	566	@subsection S3_2_linux Building for Linux
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	567
				568	@subsubsection S3_2_1_library How to build the library ?
				569
				570	For Linux, the library was successfully built and tested using the following Linaro GCC toolchain:
				571
				572	- gcc-linaro-arm-linux-gnueabihf-4.9-2014.07_linux
				573	- gcc-linaro-4.9-2016.02-x86_64_aarch64-linux-gnu
				574	- gcc-linaro-6.3.1-2017.02-i686_aarch64-linux-gnu
				575
				576	@note If you are building with opencl=1 then scons will expect to find libOpenCL.so either in the current directory or in "build" (See the section below if you need a stub OpenCL library to link against)
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	577	@note If you are building with gles_compute=1 then scons will expect to find libEGL.so / libGLESv1_CM.so / libGLESv2.so either in the current directory or in "build" (See the section below if you need a stub OpenCL library to link against)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	578
				579	To cross-compile the library in debug mode, with NEON only support, for Linux 32bit:
				580
				581	scons Werror=1 -j8 debug=1 neon=1 opencl=0 os=linux arch=armv7a
				582
				583	To cross-compile the library in asserts mode, with OpenCL only support, for Linux 64bit:
				584
				585	scons Werror=1 -j8 debug=0 asserts=1 neon=0 opencl=1 embed_kernels=1 os=linux arch=arm64-v8a
				586
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	587	To cross-compile the library in asserts mode, with GLES_COMPUTE only support, for Linux 64bit:
				588
				589	scons Werror=1 -j8 debug=0 asserts=1 neon=0 opencl=0 gles_compute=1 embed_kernels=1 os=linux arch=arm64-v8a
				590
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	591	You can also compile the library natively on an ARM device by using <b>build=native</b>:
				592
				593	scons Werror=1 -j8 debug=0 neon=1 opencl=0 os=linux arch=arm64-v8a build=native
				594	scons Werror=1 -j8 debug=0 neon=1 opencl=0 os=linux arch=armv7a build=native
				595
				596	@note g++ for ARM is mono-arch, therefore if you want to compile for Linux 32bit on a Linux 64bit platform you will have to use a cross compiler.
				597
				598	For example on a 64bit Debian based system you would have to install <b>g++-arm-linux-gnueabihf</b>
				599
				600	apt-get install g++-arm-linux-gnueabihf
				601
				602	Then run
				603
				604	scons Werror=1 -j8 debug=0 neon=1 opencl=0 os=linux arch=armv7a build=cross_compile
				605
				606	or simply remove the build parameter as build=cross_compile is the default value:
				607
				608	scons Werror=1 -j8 debug=0 neon=1 opencl=0 os=linux arch=armv7a
				609
				610	@attention To cross compile with opencl=1 you need to make sure to have a version of libOpenCL matching your target architecture.
				611
				612	@subsubsection S3_2_2_examples How to manually build the examples ?
				613
				614	The examples get automatically built by scons as part of the build process of the library described above. This section just describes how you can build and link your own application against our library.
				615
				616	@note The following command lines assume the arm_compute and libOpenCL binaries are present in the current directory or in the system library path. If this is not the case you can specify the location of the pre-built library with the compiler option -L. When building the OpenCL example the commands below assume that the CL headers are located in the include folder where the command is executed.
				617
				618	To cross compile a NEON example for Linux 32bit:
				619
Anthony Barbier	b2881fc	2017-09-29 17:12:12 +0100	[diff] [blame]	620	arm-linux-gnueabihf-g++ examples/neon_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -mfpu=neon -L. -larm_compute -larm_compute_core -o neon_convolution
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	621
				622	To cross compile a NEON example for Linux 64bit:
				623
Anthony Barbier	b2881fc	2017-09-29 17:12:12 +0100	[diff] [blame]	624	aarch64-linux-gnu-g++ examples/neon_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -L. -larm_compute -larm_compute_core -o neon_convolution
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	625
				626	(notice the only difference with the 32 bit command is that we don't need the -mfpu option and the compiler's name is different)
				627
				628	To cross compile an OpenCL example for Linux 32bit:
				629
Anthony Barbier	b2881fc	2017-09-29 17:12:12 +0100	[diff] [blame]	630	arm-linux-gnueabihf-g++ examples/cl_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -mfpu=neon -L. -larm_compute -larm_compute_core -lOpenCL -o cl_convolution -DARM_COMPUTE_CL
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	631
				632	To cross compile an OpenCL example for Linux 64bit:
				633
Anthony Barbier	b2881fc	2017-09-29 17:12:12 +0100	[diff] [blame]	634	aarch64-linux-gnu-g++ examples/cl_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -L. -larm_compute -larm_compute_core -lOpenCL -o cl_convolution -DARM_COMPUTE_CL
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	635
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	636	To cross compile a GLES example for Linux 32bit:
				637
				638	arm-linux-gnueabihf-g++ examples/gc_absdiff.cpp utils/Utils.cpp -I. -Iinclude/ -L. -larm_compute -larm_compute_core -std=c++11 -mfpu=neon -DARM_COMPUTE_GC -Iinclude/linux/ -o gc_absdiff
				639
				640	To cross compile a GLES example for Linux 64bit:
				641
				642	aarch64-linux-gnu-g++ examples/gc_absdiff.cpp utils/Utils.cpp -I. -Iinclude/ -L. -larm_compute -larm_compute_core -std=c++11 -DARM_COMPUTE_GC -Iinclude/linux/ -o gc_absdiff
				643
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	644	(notice the only difference with the 32 bit command is that we don't need the -mfpu option and the compiler's name is different)
				645
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	646	To cross compile the examples with the Graph API, such as graph_lenet.cpp, you need to link the examples against arm_compute_graph.so too.
				647
				648	@note The compute library must currently be built with both neon and opencl enabled - neon=1 and opencl=1
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	649
				650	i.e. to cross compile the "graph_lenet" example for Linux 32bit:
				651
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	652	arm-linux-gnueabihf-g++ examples/graph_lenet.cpp utils/Utils.cpp utils/GraphUtils.cpp -I. -Iinclude -std=c++11 -mfpu=neon -L. -larm_compute_graph -larm_compute -larm_compute_core -Wl,--allow-shlib-undefined -o graph_lenet
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	653
				654	i.e. to cross compile the "graph_lenet" example for Linux 64bit:
				655
Isabella Gottardi	b28f29d	2017-11-09 17:05:07 +0000	[diff] [blame]	656	aarch64-linux-gnu-g++ examples/graph_lenet.cpp utils/Utils.cpp utils/GraphUtils.cpp -I. -Iinclude -std=c++11 -L. -larm_compute_graph -larm_compute -larm_compute_core -Wl,--allow-shlib-undefined -o graph_lenet
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	657
				658	(notice the only difference with the 32 bit command is that we don't need the -mfpu option and the compiler's name is different)
				659
Anthony Barbier	e500747	2017-10-27 15:01:44 +0100	[diff] [blame]	660	@note If compiling using static libraries, this order must be followed when linking: arm_compute_graph_static, arm_compute, arm_compute_core
				661
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	662	To compile natively (i.e directly on an ARM device) for NEON for Linux 32bit:
				663
Anthony Barbier	b2881fc	2017-09-29 17:12:12 +0100	[diff] [blame]	664	g++ examples/neon_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -mfpu=neon -larm_compute -larm_compute_core -o neon_convolution
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	665
				666	To compile natively (i.e directly on an ARM device) for NEON for Linux 64bit:
				667
Anthony Barbier	b2881fc	2017-09-29 17:12:12 +0100	[diff] [blame]	668	g++ examples/neon_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -larm_compute -larm_compute_core -o neon_convolution
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	669
				670	(notice the only difference with the 32 bit command is that we don't need the -mfpu option)
				671
				672	To compile natively (i.e directly on an ARM device) for OpenCL for Linux 32bit or Linux 64bit:
				673
Anthony Barbier	b2881fc	2017-09-29 17:12:12 +0100	[diff] [blame]	674	g++ examples/cl_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -larm_compute -larm_compute_core -lOpenCL -o cl_convolution -DARM_COMPUTE_CL
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	675
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	676	To compile natively (i.e directly on an ARM device) for GLES for Linux 32bit or Linux 64bit:
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	677
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	678	g++ examples/gc_absdiff.cpp utils/Utils.cpp -I. -Iinclude/ -L. -larm_compute -larm_compute_core -std=c++11 -DARM_COMPUTE_GC -Iinclude/linux/ -o gc_absdiff
				679
				680	To compile natively the examples with the Graph API, such as graph_lenet.cpp, you need to link the examples against arm_compute_graph.so too.
				681	@note The compute library must currently be built with both neon and opencl enabled - neon=1 and opencl=1
				682
				683	i.e. to natively compile the "graph_lenet" example for Linux 32bit:
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	684
Isabella Gottardi	b28f29d	2017-11-09 17:05:07 +0000	[diff] [blame]	685	g++ examples/graph_lenet.cpp utils/Utils.cpp utils/GraphUtils.cpp -I. -Iinclude -std=c++11 -mfpu=neon -L. -larm_compute_graph -larm_compute -larm_compute_core -Wl,--allow-shlib-undefined -o graph_lenet
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	686
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	687	i.e. to natively compile the "graph_lenet" example for Linux 64bit:
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	688
Isabella Gottardi	b28f29d	2017-11-09 17:05:07 +0000	[diff] [blame]	689	g++ examples/graph_lenet.cpp utils/Utils.cpp utils/GraphUtils.cpp -I. -Iinclude -std=c++11 L. -larm_compute_graph -larm_compute -larm_compute_core -Wl,--allow-shlib-undefined -o graph_lenet
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	690
				691	(notice the only difference with the 32 bit command is that we don't need the -mfpu option)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	692
Anthony Barbier	e500747	2017-10-27 15:01:44 +0100	[diff] [blame]	693	@note If compiling using static libraries, this order must be followed when linking: arm_compute_graph_static, arm_compute, arm_compute_core
				694
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	695	@note These two commands assume libarm_compute.so is available in your library path, if not add the path to it using -L
				696
				697	To run the built executable simply run:
				698
				699	LD_LIBRARY_PATH=build ./neon_convolution
				700
				701	or
				702
				703	LD_LIBRARY_PATH=build ./cl_convolution
				704
Moritz Pflanzer	07674de	2017-07-21 09:39:36 +0100	[diff] [blame]	705	@subsection S3_3_android Building for Android
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	706
				707	For Android, the library was successfully built and tested using Google's standalone toolchains:
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	708	- NDK r14 arm-linux-androideabi-4.9 for armv7a (clang++)
				709	- NDK r14 aarch64-linux-android-4.9 for arm64-v8a (clang++)
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	710
				711	Here is a guide to <a href="https://developer.android.com/ndk/guides/standalone_toolchain.html">create your Android standalone toolchains from the NDK</a>
				712
				713	- Download the NDK r14 from here: https://developer.android.com/ndk/downloads/index.html
				714	- Make sure you have Python 2 installed on your machine.
				715	- Generate the 32 and/or 64 toolchains by running the following commands:
				716
				717
Anthony Barbier	e500747	2017-10-27 15:01:44 +0100	[diff] [blame]	718	$NDK/build/tools/make_standalone_toolchain.py --arch arm64 --install-dir $MY_TOOLCHAINS/aarch64-linux-android-4.9 --stl gnustl --api 21
				719	$NDK/build/tools/make_standalone_toolchain.py --arch arm --install-dir $MY_TOOLCHAINS/arm-linux-androideabi-4.9 --stl gnustl --api 21
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	720
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	721	@attention Due to some NDK issues make sure you use clang++ & gnustl
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	722
				723	@note Make sure to add the toolchains to your PATH: export PATH=$PATH:$MY_TOOLCHAINS/aarch64-linux-android-4.9/bin:$MY_TOOLCHAINS/arm-linux-androideabi-4.9/bin
				724
				725	@subsubsection S3_3_1_library How to build the library ?
				726
				727	@note If you are building with opencl=1 then scons will expect to find libOpenCL.so either in the current directory or in "build" (See the section below if you need a stub OpenCL library to link against)
				728
				729	To cross-compile the library in debug mode, with NEON only support, for Android 32bit:
				730
				731	CXX=clang++ CC=clang scons Werror=1 -j8 debug=1 neon=1 opencl=0 os=android arch=armv7a
				732
				733	To cross-compile the library in asserts mode, with OpenCL only support, for Android 64bit:
				734
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	735	CXX=clang++ CC=clang scons Werror=1 -j8 debug=0 asserts=1 neon=0 opencl=1 embed_kernels=1 os=android arch=arm64-v8a
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	736
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	737	To cross-compile the library in asserts mode, with GLES_COMPUTE only support, for Android 64bit:
				738
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	739	CXX=clang++ CC=clang scons Werror=1 -j8 debug=0 asserts=1 neon=0 opencl=0 gles_compute=1 embed_kernels=1 os=android arch=arm64-v8a
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	740
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	741	@subsubsection S3_3_2_examples How to manually build the examples ?
				742
				743	The examples get automatically built by scons as part of the build process of the library described above. This section just describes how you can build and link your own application against our library.
				744
Anthony Barbier	fabb038	2017-06-23 14:42:52 +0100	[diff] [blame]	745	@note The following command lines assume the arm_compute and libOpenCL binaries are present in the current directory or in the system library path. If this is not the case you can specify the location of the pre-built library with the compiler option -L. When building the OpenCL example the commands below assume that the CL headers are located in the include folder where the command is executed.
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	746
				747	Once you've got your Android standalone toolchain built and added to your path you can do the following:
				748
				749	To cross compile a NEON example:
				750
				751	#32 bit:
Georgios Pinitas	9873ea3	2017-12-05 15:28:55 +0000	[diff] [blame]	752	arm-linux-androideabi-clang++ examples/neon_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -larm_compute-static -larm_compute_core-static -L. -o neon_convolution_arm -static-libstdc++ -pie
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	753	#64 bit:
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	754	aarch64-linux-android-clang++ examples/neon_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -larm_compute-static -larm_compute_core-static -L. -o neon_convolution_aarch64 -static-libstdc++ -pie
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	755
				756	To cross compile an OpenCL example:
				757
				758	#32 bit:
Georgios Pinitas	9873ea3	2017-12-05 15:28:55 +0000	[diff] [blame]	759	arm-linux-androideabi-clang++ examples/cl_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -larm_compute-static -larm_compute_core-static -L. -o cl_convolution_arm -static-libstdc++ -pie -lOpenCL -DARM_COMPUTE_CL
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	760	#64 bit:
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	761	aarch64-linux-android-clang++ examples/cl_convolution.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -larm_compute-static -larm_compute_core-static -L. -o cl_convolution_aarch64 -static-libstdc++ -pie -lOpenCL -DARM_COMPUTE_CL
				762
				763	To cross compile a GLES example:
Anthony Barbier	cc0a80b	2017-12-15 11:37:29 +0000	[diff] [blame]	764
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	765	#32 bit:
				766	arm-linux-androideabi-clang++ examples/gc_absdiff.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -larm_compute-static -larm_compute_core-static -L. -o gc_absdiff_arm -static-libstdc++ -pie -DARM_COMPUTE_GC
				767	#64 bit:
				768	aarch64-linux-android-clang++ examples/gc_absdiff.cpp utils/Utils.cpp -I. -Iinclude -std=c++11 -larm_compute-static -larm_compute_core-static -L. -o gc_absdiff_aarch64 -static-libstdc++ -pie -DARM_COMPUTE_GC
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	769
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	770	To cross compile the examples with the Graph API, such as graph_lenet.cpp, you need to link the library arm_compute_graph also.
				771	(notice the compute library has to be built with both neon and opencl enabled - neon=1 and opencl=1)
				772
				773	#32 bit:
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	774	arm-linux-androideabi-clang++ examples/graph_lenet.cpp utils/Utils.cpp utils/GraphUtils.cpp -I. -Iinclude -std=c++11 -Wl,--whole-archive -larm_compute_graph-static -Wl,--no-whole-archive -larm_compute-static -larm_compute_core-static -L. -o graph_lenet_arm -static-libstdc++ -pie -lOpenCL -DARM_COMPUTE_CL
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	775	#64 bit:
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	776	aarch64-linux-android-clang++ examples/graph_lenet.cpp utils/Utils.cpp utils/GraphUtils.cpp -I. -Iinclude -std=c++11 -Wl,--whole-archive -larm_compute_graph-static -Wl,--no-whole-archive -larm_compute-static -larm_compute_core-static -L. -o graph_lenet_aarch64 -static-libstdc++ -pie -lOpenCL -DARM_COMPUTE_CL
Gian Marco Iodice	daec1aa	2017-09-29 12:03:18 +0100	[diff] [blame]	777
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	778	@note Due to some issues in older versions of the Mali OpenCL DDK (<= r13p0), we recommend to link arm_compute statically on Android.
Anthony Barbier	20dbb82	2017-12-13 21:19:39 +0000	[diff] [blame]	779	@note When linked statically the arm_compute_graph library currently needs the --whole-archive linker flag in order to work properly
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	780
				781	Then you need to do is upload the executable and the shared library to the device using ADB:
				782
				783	adb push neon_convolution_arm /data/local/tmp/
				784	adb push cl_convolution_arm /data/local/tmp/
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	785	adb push gc_absdiff_arm /data/local/tmp/
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	786	adb shell chmod 777 -R /data/local/tmp/
				787
				788	And finally to run the example:
				789
				790	adb shell /data/local/tmp/neon_convolution_arm
				791	adb shell /data/local/tmp/cl_convolution_arm
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	792	adb shell /data/local/tmp/gc_absdiff_arm
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	793
				794	For 64bit:
				795
				796	adb push neon_convolution_aarch64 /data/local/tmp/
				797	adb push cl_convolution_aarch64 /data/local/tmp/
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	798	adb push gc_absdiff_aarch64 /data/local/tmp/
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	799	adb shell chmod 777 -R /data/local/tmp/
				800
				801	And finally to run the example:
				802
				803	adb shell /data/local/tmp/neon_convolution_aarch64
				804	adb shell /data/local/tmp/cl_convolution_aarch64
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	805	adb shell /data/local/tmp/gc_absdiff_aarch64
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	806
Michalis Spyrou	6e52ba3	2017-10-04 15:40:38 +0100	[diff] [blame]	807	@subsection S3_4_bare_metal Building for bare metal
				808
				809	For bare metal, the library was successfully built using linaros's latest (gcc-linaro-6.3.1-2017.05) bare metal toolchains:
				810	- arm-eabi for armv7a
				811	- aarch64-elf for arm64-v8a
				812
				813	Download linaro for <a href="https://releases.linaro.org/components/toolchain/binaries/6.3-2017.05/arm-eabi/">armv7a</a> and <a href="https://releases.linaro.org/components/toolchain/binaries/6.3-2017.05/aarch64-elf/">arm64-v8a</a>.
				814
				815	@note Make sure to add the toolchains to your PATH: export PATH=$PATH:$MY_TOOLCHAINS/gcc-linaro-6.3.1-2017.05-x86_64_aarch64-elf/bin:$MY_TOOLCHAINS/gcc-linaro-6.3.1-2017.05-x86_64_arm-eabi/bin
				816
				817	@subsubsection S3_4_1_library How to build the library ?
				818
				819	To cross-compile the library with NEON support for baremetal arm64-v8a:
				820
				821	scons Werror=1 -j8 debug=0 neon=1 opencl=0 os=bare_metal arch=arm64-v8a build=cross_compile cppthreads=0 openmp=0 standalone=1
				822
				823	@subsubsection S3_4_2_examples How to manually build the examples ?
				824
				825	Examples are disabled when building for bare metal. If you want to build the examples you need to provide a custom bootcode depending on the target architecture and link against the compute library. More information about bare metal bootcode can be found <a href="http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dai0527a/index.html">here</a>.
				826
				827	@subsection S3_5_windows_host Building on a Windows host system
Moritz Pflanzer	07674de	2017-07-21 09:39:36 +0100	[diff] [blame]	828
				829	Using `scons` directly from the Windows command line is known to cause
				830	problems. The reason seems to be that if `scons` is setup for cross-compilation
				831	it gets confused about Windows style paths (using backslashes). Thus it is
				832	recommended to follow one of the options outlined below.
				833
Michalis Spyrou	6e52ba3	2017-10-04 15:40:38 +0100	[diff] [blame]	834	@subsubsection S3_5_1_ubuntu_on_windows Bash on Ubuntu on Windows
Moritz Pflanzer	07674de	2017-07-21 09:39:36 +0100	[diff] [blame]	835
				836	The best and easiest option is to use
				837	<a href="https://msdn.microsoft.com/en-gb/commandline/wsl/about">Ubuntu on Windows</a>.
				838	This feature is still marked as beta and thus might not be available.
				839	However, if it is building the library is as simple as opening a *Bash on
				840	Ubuntu on Windows* shell and following the general guidelines given above.
				841
Michalis Spyrou	6e52ba3	2017-10-04 15:40:38 +0100	[diff] [blame]	842	@subsubsection S3_5_2_cygwin Cygwin
Moritz Pflanzer	07674de	2017-07-21 09:39:36 +0100	[diff] [blame]	843
				844	If the Windows subsystem for Linux is not available <a href="https://www.cygwin.com/">Cygwin</a>
				845	can be used to install and run `scons`. In addition to the default packages
				846	installed by Cygwin `scons` has to be selected in the installer. (`git` might
				847	also be useful but is not strictly required if you already have got the source
				848	code of the library.) Linaro provides pre-built versions of
				849	<a href="http://releases.linaro.org/components/toolchain/binaries/">GCC cross-compilers</a>
				850	that can be used from the Cygwin terminal. When building for Android the
				851	compiler is included in the Android standalone toolchain. After everything has
				852	been set up in the Cygwin terminal the general guide on building the library
				853	can be followed.
				854
Michalis Spyrou	6e52ba3	2017-10-04 15:40:38 +0100	[diff] [blame]	855	@subsection S3_6_cl_stub_library The OpenCL stub library
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	856
				857	In the opencl-1.2-stubs folder you will find the sources to build a stub OpenCL library which then can be used to link your application or arm_compute against.
				858
				859	If you preferred you could retrieve the OpenCL library from your device and link against this one but often this library will have dependencies on a range of system libraries forcing you to link your application against those too even though it is not using them.
				860
				861	@warning This OpenCL library provided is a stub and not a real implementation. You can use it to resolve OpenCL's symbols in arm_compute while building the example but you must make sure the real libOpenCL.so is in your PATH when running the example or it will not work.
				862
				863	To cross-compile the stub OpenCL library simply run:
				864
				865	<target-prefix>-gcc -o libOpenCL.so -Iinclude opencl-1.2-stubs/opencl_stubs.c -fPIC -shared
				866
				867	For example:
				868
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	869	#Linux 32bit
				870	arm-linux-gnueabihf-gcc -o libOpenCL.so -Iinclude opencl-1.2-stubs/opencl_stubs.c -fPIC -shared
				871	#Linux 64bit
				872	aarch64-linux-gnu-gcc -o libOpenCL.so -Iinclude -shared opencl-1.2-stubs/opencl_stubs.c -fPIC
				873	#Android 32bit
				874	arm-linux-androideabi-clang -o libOpenCL.so -Iinclude -shared opencl-1.2-stubs/opencl_stubs.c -fPIC -shared
				875	#Android 64bit
Anthony Barbier	14c86a9	2017-12-14 16:27:41 +0000	[diff] [blame]	876	aarch64-linux-android-clang -o libOpenCL.so -Iinclude -shared opencl-1.2-stubs/opencl_stubs.c -fPIC -shared
				877
				878	@subsection S3_7_gles_stub_library The Linux OpenGLES and EGL stub libraries
				879
				880	In the opengles-3.1-stubs folder you will find the sources to build stub EGL and OpenGLES libraries which then can be used to link your Linux application of arm_compute against.
				881
				882	@note The stub libraries are only needed on Linux. For Android, the NDK toolchains already provide the meta-EGL and meta-GLES libraries.
				883
				884	To cross-compile the stub OpenGLES and EGL libraries simply run:
				885
				886	<target-prefix>-gcc -o libEGL.so -Iinclude/linux opengles-3.1-stubs/EGL.c -fPIC -shared
				887	<target-prefix>-gcc -o libGLESv2.so -Iinclude/linux opengles-3.1-stubs/GLESv2.c -fPIC -shared
				888
				889	#Linux 32bit
				890	arm-linux-gnueabihf-gcc -o libEGL.so -Iinclude/linux opengles-3.1-stubs/EGL.c -fPIC -shared
				891	arm-linux-gnueabihf-gcc -o libGLESv2.so -Iinclude/linux opengles-3.1-stubs/GLESv2.c -fPIC -shared
				892
				893	#Linux 64bit
				894	aarch64-linux-gnu-gcc -o libEGL.so -Iinclude/linux opengles-3.1-stubs/EGL.c -fPIC -shared
				895	aarch64-linux-gnu-gcc -o libGLESv2.so -Iinclude/linux opengles-3.1-stubs/GLESv2.c -fPIC -shared
Anthony Barbier	6ff3b19	2017-09-04 18:44:23 +0100	[diff] [blame]	896	*/