| commit | daaad47d9cc74991bc3f840664df32e67c54d2ac | [log] [tgz] |
|---|---|---|
| author | Shunting Zhang <shunting@fb.com> | Thu Nov 04 20:51:23 2021 -0700 |
| committer | Facebook GitHub Bot <facebook-github-bot@users.noreply.github.com> | Thu Nov 04 20:52:57 2021 -0700 |
| tree | b25b6fdb1c185297e5c094ba904cd2f24535eb63 | |
| parent | 5a48868d3961545a74ce611abf464b6ee8e34b90 [diff] |
Allow torch::deploy unity embed xar file of any size (#67814) Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/67814 There was a limitation on the xar file size we can embed into the binary previously. The payload (xar file here) is added to .data section by default using 'ld -b binary -r' command (which section the payload goes is hardcoded in ld BTW. Check code pointer [here](https://github.com/bminor/binutils-gdb/blob/binutils-2_32/bfd/binary.c#L80) ) . When we link the object file containing the payload to other parts of the executable, we will get relocation out of range error if the overall size of .test, .data, .bss etc sections exceed 2G. Some relocation entries uses 32 bit singed integer, thus the limit is 2G here. To solve the issue and mitigate the risk, we designed a mechanism to put the payload in a customized payload section (.torch_deploy_payload.unity here). The payload section does not join the party of relocating and symbol resolution, thus in theory it can be as large as the disk space... Since we don't do relocation for the payload section, the start/end/size symbols are no longer available/valid, we have to parse the ELF file ourselves to figure out those. The mechanism can be used to embed interprter.so as well. The interpreter.so is currently 0.5G. That will limit the other .test/.data/.bss sections of the executable to be at most 1.5G. Using this mechanim in this diff avoid the interpreter.so taking any budgets. We could also use this mechanism to ship python scripts with our binary rather than freeze them before hand. These use cases are not handled in this diff. This diff also improves experience for those simple use cases that does not depends on extra shared libraries in the XAR file (except the shared libraries for python extensions themselves). This is mainly for fixing the stress test right now, but it also makes other simple cases easier. ghstack-source-id: 142483327 Test Plan: # Verify the relocation out of range issue is fixed Add //caffe2:torch as a dependency to the macro build_unity(name="example", …) in torch/csrc/deploy/unity/TARGETS and run 'buck run mode/opt :unity_demo', it's expected to get the relocation errors like: ``` ld.lld: error: caffe2/c10/util/intrusive_ptr.h:325:(.text._ZN11ska_ordered8detailv317sherwood_v3_tableISt4pairIN3c106IValueES4_ES4_NS3_6detail11DictKeyHashENS0_16KeyOrValueHasherIS4_S5_S7_EENS6_14DictKeyEqualToENS0_18KeyOrValueEqualityIS4_S5_SA_EESaIS5_ESaINS0_17sherwood_v3_entryIS5_EEEE15emplace_new_keyIS5_JEEES2_INSH_18templated_iteratorIS5_EEbEaPSF_OT_DpOT0_+0x4E9): relocation R_X86_64_32S out of range: 2345984168 is not in [-2147483648, 2147483647]; references c10::UndefinedTensorImpl::_singleton >>> defined in /data/sandcastle/boxes/fbsource/fbcode/buck-out/opt/gen/caffe2/c10/c10#platform009-clang,static/libc10.a(../c10#compile-UndefinedTensorImpl.cpp.o44c44c4c,platform009-clang/core/UndefinedTensorImpl.cpp.o) ``` With the diff, the error above is resolved. # Pass Stress Test Also pass existing unit tests for unity. buck test mode/opt //caffe2/torch/csrc/deploy/unity/tests:test_unity_sum -- --exact 'caffe2/torch/csrc/deploy/unity/tests:test_unity_sum - UnityTest.TestUnitySum' --run-disabled --jobs 18 --stress-runs 10 --record-results buck test mode/opt //caffe2/torch/csrc/deploy/unity/tests:test_unity_simple_model -- --exact 'caffe2/torch/csrc/deploy/unity/tests:test_unity_simple_model - UnityTest.TestUnitySimpleModel' --run-disabled --jobs 18 --stress-runs 10 --record-results # Verify debug sections are not messed up Verified that debug sections are not messed up and GDB still works: `gdb ~/fbcode/buck-out/gen/caffe2/torch/csrc/deploy/unity/unity_demo` ``` b main run l c ``` Reviewed By: suo Differential Revision: D32159644 fbshipit-source-id: a133513261b73551a71acc257f4019f7b5af34a8
PyTorch is a Python package that provides two high-level features:
You can reuse your favorite Python packages such as NumPy, SciPy, and Cython to extend PyTorch when needed.
| System | 3.6 | 3.7 | 3.8 |
|---|---|---|---|
| Linux CPU | — | ||
| Linux GPU | — | ||
| Windows CPU / GPU | — | — | |
| Linux (ppc64le) CPU | — | — | |
| Linux (ppc64le) GPU | — | — | |
| Linux (aarch64) CPU |
See also the CI HUD at hud.pytorch.org.
At a granular level, PyTorch is a library that consists of the following components:
| Component | Description |
|---|---|
| torch | a Tensor library like NumPy, with strong GPU support |
| torch.autograd | a tape-based automatic differentiation library that supports all differentiable Tensor operations in torch |
| torch.jit | a compilation stack (TorchScript) to create serializable and optimizable models from PyTorch code |
| torch.nn | a neural networks library deeply integrated with autograd designed for maximum flexibility |
| torch.multiprocessing | Python multiprocessing, but with magical memory sharing of torch Tensors across processes. Useful for data loading and Hogwild training |
| torch.utils | DataLoader and other utility functions for convenience |
Usually, PyTorch is used either as:
Elaborating Further:
If you use NumPy, then you have used Tensors (a.k.a. ndarray).
PyTorch provides Tensors that can live either on the CPU or the GPU and accelerates the computation by a huge amount.
We provide a wide variety of tensor routines to accelerate and fit your scientific computation needs such as slicing, indexing, math operations, linear algebra, reductions. And they are fast!
PyTorch has a unique way of building neural networks: using and replaying a tape recorder.
Most frameworks such as TensorFlow, Theano, Caffe, and CNTK have a static view of the world. One has to build a neural network and reuse the same structure again and again. Changing the way the network behaves means that one has to start from scratch.
With PyTorch, we use a technique called reverse-mode auto-differentiation, which allows you to change the way your network behaves arbitrarily with zero lag or overhead. Our inspiration comes from several research papers on this topic, as well as current and past work such as torch-autograd, autograd, Chainer, etc.
While this technique is not unique to PyTorch, it's one of the fastest implementations of it to date. You get the best of speed and flexibility for your crazy research.
PyTorch is not a Python binding into a monolithic C++ framework. It is built to be deeply integrated into Python. You can use it naturally like you would use NumPy / SciPy / scikit-learn etc. You can write your new neural network layers in Python itself, using your favorite libraries and use packages such as Cython and Numba. Our goal is to not reinvent the wheel where appropriate.
PyTorch is designed to be intuitive, linear in thought, and easy to use. When you execute a line of code, it gets executed. There isn't an asynchronous view of the world. When you drop into a debugger or receive error messages and stack traces, understanding them is straightforward. The stack trace points to exactly where your code was defined. We hope you never spend hours debugging your code because of bad stack traces or asynchronous and opaque execution engines.
PyTorch has minimal framework overhead. We integrate acceleration libraries such as Intel MKL and NVIDIA (cuDNN, NCCL) to maximize speed. At the core, its CPU and GPU Tensor and neural network backends are mature and have been tested for years.
Hence, PyTorch is quite fast – whether you run small or large neural networks.
The memory usage in PyTorch is extremely efficient compared to Torch or some of the alternatives. We've written custom memory allocators for the GPU to make sure that your deep learning models are maximally memory efficient. This enables you to train bigger deep learning models than before.
Writing new neural network modules, or interfacing with PyTorch's Tensor API was designed to be straightforward and with minimal abstractions.
You can write new neural network layers in Python using the torch API or your favorite NumPy-based libraries such as SciPy.
If you want to write your layers in C/C++, we provide a convenient extension API that is efficient and with minimal boilerplate. No wrapper code needs to be written. You can see a tutorial here and an example here.
Commands to install binaries via Conda or pip wheels are on our website: https://pytorch.org/get-started/locally/
Python wheels for NVIDIA's Jetson Nano, Jetson TX2, and Jetson AGX Xavier are available via the following URLs:
They require JetPack 4.2 and above, and @dusty-nv maintains them
If you are installing from source, you will need Python 3.6.2 or later and a C++14 compiler. Also, we highly recommend installing an Anaconda environment. You will get a high-quality BLAS library (MKL) and you get controlled dependency versions regardless of your Linux distro.
Once you have Anaconda installed, here are the instructions.
If you want to compile with CUDA support, install
If you want to disable CUDA support, export environment variable USE_CUDA=0. Other potentially useful environment variables may be found in setup.py.
If you are building for NVIDIA's Jetson platforms (Jetson Nano, TX1, TX2, AGX Xavier), Instructions to install PyTorch for Jetson Nano are available here
If you want to compile with ROCm support, install
If you want to disable ROCm support, export environment variable USE_ROCM=0. Other potentially useful environment variables may be found in setup.py.
Common
conda install astunparse numpy ninja pyyaml mkl mkl-include setuptools cmake cffi typing_extensions future six requests dataclasses
On Linux
# CUDA only: Add LAPACK support for the GPU if needed conda install -c pytorch magma-cuda110 # or the magma-cuda* that matches your CUDA version from https://anaconda.org/pytorch/repo
On MacOS
# Add these packages if torch.distributed is needed conda install pkg-config libuv
On Windows
# Add these packages if torch.distributed is needed. # Distributed package support on Windows is a prototype feature and is subject to changes. conda install -c conda-forge libuv=1.39
git clone --recursive https://github.com/pytorch/pytorch cd pytorch # if you are updating an existing checkout git submodule sync git submodule update --init --recursive --jobs 0
On Linux
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"} python setup.py install
Note that if you are compiling for ROCm, you must run this command first:
python tools/amd_build/build_amd.py
Note that if you are using Anaconda, you may experience an error caused by the linker:
build/temp.linux-x86_64-3.7/torch/csrc/stub.o: file not recognized: file format not recognized collect2: error: ld returned 1 exit status error: command 'g++' failed with exit status 1
This is caused by ld from Conda environment shadowing the system ld. You should use a newer version of Python that fixes this issue. The recommended Python version is 3.6.10+, 3.7.6+ and 3.8.1+.
On macOS
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"} MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py install
CUDA is not supported on macOS.
On Windows
Choose Correct Visual Studio Version.
Sometimes there are regressions in new versions of Visual Studio, so it‘s best to use the same Visual Studio Version 16.8.5 as Pytorch CI’s. You can use Visual Studio Enterprise, Professional or Community though PyTorch CI uses Visual Studio BuildTools.
If you want to build legacy python code, please refer to Building on legacy code and CUDA
Build with CPU
It's fairly easy to build with CPU.
Note on OpenMP: The desired OpenMP implementation is Intel OpenMP (iomp). In order to link against iomp, you'll need to manually download the library and set up the building environment by tweaking CMAKE_INCLUDE_PATH and LIB. The instruction here is an example for setting up both MKL and Intel OpenMP. Without these configurations for CMake, Microsoft Visual C OpenMP runtime (vcomp) will be used.
Build with CUDA
NVTX is needed to build Pytorch with CUDA. NVTX is a part of CUDA distributive, where it is called “Nsight Compute”. To install it onto already installed CUDA run CUDA installation once again and check the corresponding checkbox. Make sure that CUDA with Nsight Compute is installed after Visual Studio.
Currently, VS 2017 / 2019, and Ninja are supported as the generator of CMake. If ninja.exe is detected in PATH, then Ninja will be used as the default generator, otherwise, it will use VS 2017 / 2019.
If Ninja is selected as the generator, the latest MSVC will get selected as the underlying toolchain.
Additional libraries such as Magma, oneDNN, a.k.a MKLDNN or DNNL, and Sccache are often needed. Please refer to the installation-helper to install them.
You can refer to the build_pytorch.bat script for some other environment variables configurations
cmd :: Set the environment variables after you have downloaded and upzipped the mkl package, :: else CMake would throw error as `Could NOT find OpenMP`. set CMAKE_INCLUDE_PATH={Your directory}\mkl\include set LIB={Your directory}\mkl\lib;%LIB% :: Read the content in the previous section carefully before you proceed. :: [Optional] If you want to override the underlying toolset used by Ninja and Visual Studio with CUDA, please run the following script block. :: "Visual Studio 2019 Developer Command Prompt" will be run automatically. :: Make sure you have CMake >= 3.12 before you do this when you use the Visual Studio generator. set CMAKE_GENERATOR_TOOLSET_VERSION=14.27 set DISTUTILS_USE_SDK=1 for /f "usebackq tokens=*" %i in (`"%ProgramFiles(x86)%\Microsoft Visual Studio\Installer\vswhere.exe" -version [15^,17^) -products * -latest -property installationPath`) do call "%i\VC\Auxiliary\Build\vcvarsall.bat" x64 -vcvars_ver=%CMAKE_GENERATOR_TOOLSET_VERSION% :: [Optional] If you want to override the CUDA host compiler set CUDAHOSTCXX=C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.27.29110\bin\HostX64\x64\cl.exe python setup.py install
You can adjust the configuration of cmake variables optionally (without building first), by doing the following. For example, adjusting the pre-detected directories for CuDNN or BLAS can be done with such a step.
On Linux
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"} python setup.py build --cmake-only ccmake build # or cmake-gui build
On macOS
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"} MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py build --cmake-only ccmake build # or cmake-gui build
You can also pull a pre-built docker image from Docker Hub and run with docker v19.03+
docker run --gpus all --rm -ti --ipc=host pytorch/pytorch:latest
Please note that PyTorch uses shared memory to share data between processes, so if torch multiprocessing is used (e.g. for multithreaded data loaders) the default shared memory segment size that container runs with is not enough, and you should increase shared memory size either with --ipc=host or --shm-size command line options to nvidia-docker run.
NOTE: Must be built with a docker version > 18.06
The Dockerfile is supplied to build images with CUDA 11.1 support and cuDNN v8. You can pass PYTHON_VERSION=x.y make variable to specify which Python version is to be used by Miniconda, or leave it unset to use the default.
make -f docker.Makefile # images are tagged as docker.io/${your_docker_username}/pytorch
To build documentation in various formats, you will need Sphinx and the readthedocs theme.
cd docs/ pip install -r requirements.txt
You can then build the documentation by running make <format> from the docs/ folder. Run make to get a list of all available output formats.
If you get a katex error run npm install katex. If it persists, try npm install -g katex
Installation instructions and binaries for previous PyTorch versions may be found on our website.
Three-pointers to get you started:
PyTorch has a 90-day release cycle (major releases). Please let us know if you encounter a bug by filing an issue.
We appreciate all contributions. If you are planning to contribute back bug-fixes, please do so without any further discussion.
If you plan to contribute new features, utility functions, or extensions to the core, please first open an issue and discuss the feature with us. Sending a PR without discussion might end up resulting in a rejected PR because we might be taking the core in a different direction than you might be aware of.
To learn more about making a contribution to Pytorch, please see our Contribution page.
PyTorch is a community-driven project with several skillful engineers and researchers contributing to it.
PyTorch is currently maintained by Adam Paszke, Sam Gross, Soumith Chintala and Gregory Chanan with major contributions coming from hundreds of talented individuals in various forms and means. A non-exhaustive but growing list needs to mention: Trevor Killeen, Sasank Chilamkurthy, Sergey Zagoruyko, Adam Lerer, Francisco Massa, Alykhan Tejani, Luca Antiga, Alban Desmaison, Andreas Koepf, James Bradbury, Zeming Lin, Yuandong Tian, Guillaume Lample, Marat Dukhan, Natalia Gimelshein, Christian Sarofeen, Martin Raison, Edward Yang, Zachary Devito.
Note: This project is unrelated to hughperkins/pytorch with the same name. Hugh is a valuable contributor to the Torch community and has helped with many things Torch and PyTorch.
PyTorch has a BSD-style license, as found in the LICENSE file.