commit | c5ab4df0c11dc66db47f2070edc719923af3367e | [log] [tgz] |
---|---|---|
author | SiCong Li <sicong.li@arm.com> | Tue Oct 17 17:38:57 2023 +0100 |
committer | SiCong Li <sicong.li@arm.com> | Wed Nov 08 09:49:56 2023 +0000 |
tree | c04bdac32528e628b2a9b9a1c1653e300328fc1b | |
parent | 4a9dbedfbfa66c2612c7461e60cd867b8aea825b [diff] |
Optimize CpuGemmConv2d start-up time When weight has no holes, we can replace CpuWeightsReshapeKernel with: - Collapse by reinterpreting weight's 3 spatial dimensions - Perform CpuTranspose For more details see the documentation in src/cpu/operators/CpuGemmConv2d.cpp This is one optimization since the CpuTranspose is better performing than CpuWeightsReshapeKernel A second optimization is to fuse this transpose with other weight transformations (e.g. pretranspose_B_array in CpuGemmAssemblyDispatch) However this second optimization depends on how the underlying gemm methods (the fall back path: CpuGemmMatrixMultiplyKernel or the assembly path: CpuGemmAssemblyDispatch) chooses to fuse the transpose. Therefore, this patch moves the transpose down from CpuGemmConv2d, to the individual gemm operators where the fusion decision needs to be made, by passing an extra "transpose_b" flag to CpuGemm New transpose_b flag in different scopes (they are all the same, but with different names because pretranspose_b has a different meaning in GemmAssemblyDispatch): GEMMInfo::pretranspose_B -> AsmGemmInfo::transpose_b New auxilliary tensors holding the transposed b result: - CpuGemm optimized path: CpuGemmAssemblyDispatch::PrePretransposedB - CpuGemm fallback path: CpuGemm::PreTransposedRHS Note that this patch does not yet have the second optimization (COMPMID-6595), but it prepares for it. Relates to COMPMID-6595 Resolves COMPMID-6499 Change-Id: I999a2da9da4b2b15369a3cc06d7872c86e0190ea Signed-off-by: SiCong Li <sicong.li@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10526 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Anitha Raj <Anitha.Raj@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
⚠ Important From release 22.05: 'master' branch has been replaced with 'main' following our inclusive language update, more information here.
⚠ Important From release 22.08: armv7a with Android build will no longer be tested or maintained.
⚠ Important From release 23.02: The 23.02 release introduces a change to the default tensor extend padding behavior. To remain compatible with previous behavior, users will need to set the new flag
ITensorInfo::lock_paddings()
on tensors for which paddings should not be extended, such as the input and output of the model that need to be mapped to a camera frame or frame buffer.
The Compute Library is a collection of low-level machine learning functions optimized for Arm® Cortex®-A, Arm® Neoverse® and Arm® Mali™ GPUs architectures.
The library provides superior performance to other open source alternatives and immediate support for new Arm® technologies e.g. SVE2.
Key Features:
Repository | Link |
---|---|
Release | https://github.com/arm-software/ComputeLibrary |
Development | https://review.mlplatform.org/#/admin/projects/ml/ComputeLibrary |
Note: The documentation includes the reference API, changelogs, build guide, contribution guide, errata, etc.
All the binaries can be downloaded from here or from the tables below.
Platform | Operating System | Release archive (Download) |
---|---|---|
Raspberry Pi 4 | Linux® 32bit | |
Raspberry Pi 4 | Linux® 64bit | |
Odroid N2 | Linux® 64bit | |
HiKey960 | Linux® 64bit |
Architecture | Operating System | Release archive (Download) |
---|---|---|
armv7 | Linux® | |
arm64-v8a | Android™ | |
arm64-v8a | Linux® | |
arm64-v8.2-a | Android™ | |
arm64-v8.2-a | Linux® |
Please refer to the following link for more pre-built binaries:
Pre-build binaries are generated with the following security / good coding practices related flags:
-Wall, -Wextra, -Wformat=2, -Winit-self, -Wstrict-overflow=2, -Wswitch-default, -Woverloaded-virtual, -Wformat-security, -Wctor-dtor-privacy, -Wsign-promo, -Weffc++, -pedantic, -fstack-protector-strong
Arm® CPUs:
Arm® Mali™ GPUs:
x86
⚠ Important Bazel and CMake builds are experimental CPU only builds, please see the documentation for more details.
Contributions to the Compute Library are more than welcome. If you are interested on contributing, please have a look at our how to contribute guidelines.
Before the Compute Library 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 <john.doe@example.org>
You must use your real name, no pseudonyms or anonymous contributions are accepted.
For technical discussion, the ComputeLibrary project has a public mailing list: acl-dev@lists.linaro.org 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/mailman3/lists/acl-dev.lists.linaro.org/
The software is provided under MIT license. Contributions to this project are accepted under the same license.
This project contains code from other projects as listed below. The original license text is included in those source files.
The OpenCL header library is licensed under Apache License, Version 2.0, which is a permissive license compatible with MIT license.
The half library is licensed under MIT license.
The libnpy library is licensed under MIT license.
The stb image library is either licensed under MIT license or is in Public Domain. It is used by this project under the terms of MIT license.
Android is a trademark of Google LLC.
Arm, Cortex, Mali and Neon are registered trademarks or trademarks of Arm Limited (or its subsidiaries) in the US and/or elsewhere.
Bazel is a trademark of Google LLC., registered in the U.S. and other countries.
CMake is a trademark of Kitware, Inc., registered in the U.S. and other countries.
Linux® is the registered trademark of Linus Torvalds in the U.S. and other countries.
Mac and macOS are trademarks of Apple Inc., registered in the U.S. and other countries.
Tizen is a registered trademark of The Linux Foundation.
Windows® is a trademark of the Microsoft group of companies.