This repository contains modifications to the PyTorch repository to provide a working Caffe2 C++ library (not requiring python) for the Aurora VE.
For example, you might train a PyTorch model on another processor, export it as ONNX, and save it as a Caffe2 model. This library allows you to write C++ code on the VE that loads and runs the model.
An example of how to use the C++ Caffe2 API, including training, is available in the intro and mnist tutorials in cdmalon/caffe2_cpp_tutorial. Note that a lot of the API has changed (upstream) after Caffe2 was incorporated into PyTorch, and these tutorials cover some of those changes.
Use clang from llvm-ve for the following steps.
To build protobuf for VE, first you must build protobuf for the host x86 environment, so that you have an executable protoc compiler.
Then clone protocolbuffers/protobuf into a build directory for Aurora, and:
./autogen.sh
/opt/nec/ve/bin/autoreconf -ifv
./configure --build=x86_64-unknown-linux-gnu --host=ve-unknown-linux-gnu CC=/path/to/clang CXX=/path/to/clang++ FC=nfort CCAS=nas LD=nld CFLAGS="-target ve-linux -O3 -fno-vectorize -fno-slp-vectorize -fno-crash-diagnostics" CXXFLAGS="-target ve-linux -O3 -fno-vectorize -fno-slp-vectorize -fno-crash-diagnostics" LDFLAGS="-target ve-linux -O3 -fno-vectorize -fno-slp-vectorize -fno-crash-diagnostics -Wl,-z,max-page-size=0x200000 -Wl,-rpath=/path/to/lib/clang/9.0.0/lib/linux/ve --with-protoc=/path/to/x86/protoc --prefix=/your/protobuf/install/dir
export VE_LD_LIBRARY_PATH=/path/to/lib/clang/9.0.0/lib/linux/ve
vi libtoolIn ./libtool, set:
postdeps="-lc++ -lc++abi -lunwind -lpthread -ldl -lm -lc /path/to/lib/clang/9.0.0/lib/linux/libclang_rt.builtins-ve.a" # in two places
archive_cmds # add '-target ve-linux' to the existing setting, wherever this is set
build_libtool_libs=yes # just the first occurrenceThen you can make and make install.
git clone --recursive https://github.com/necla-ml/pytorch and set the following environment variables:
export VE_LD_LIBRARY_PATH=/opt/nec/ve/lib:/path/to/lib/clang/9.0.0/lib/linux/ve:/your/protobuf/install/dir/lib
export NLC_HOME=/opt/nec/ve/nlc/1.0.0
export Protobuf_INSTALL_DIR=/your/protobuf/install/dir
export SNAPPY_ROOT_DIR=/your/snappy/install/dir # if desired; see below
export LEVELDB_ROOT=/your/leveldb/install/dir # if desired; see below
export CC=/path/to/clang
export CXX=/path/to/clang++These can be cloned from google/snappy and google/leveldb. Make a build subdirectory and:
cmake -DCMAKE_TOOLCHAIN_FILE=/path/to/pytorch-checkout/cmake/Modules/ve.cmake -DCMAKE_BUILD_TYPE=Debug -DCMAKE_INSTALL_PREFIX=/your/snappy/install/dir -DBUILD_SHARED_LIBS=ON ..
make
make installand similarly for LevelDB.
With the environment variables set as above, go to your PyTorch clone
directory. Enter an environment where Python (for the host architecture)
and PyTorch's dependent modules are installed, and execute scripts/build_ve.sh.
Your libraries will be built in the subdirectory build_ve/lib.
We recommend CMake 3.14. If earlier CMake is used, CMake might only configure and not
build, so that you need to go to build_ve and execute make yourself.
PyTorch is a Python package that provides two high-level features:
- Tensor computation (like NumPy) with strong GPU acceleration
- Deep neural networks built on a tape-based autograd system
You can reuse your favorite Python packages such as NumPy, SciPy and Cython to extend PyTorch when needed.
| System | 2.7 | 3.5 | 3.6 |
|---|---|---|---|
| Linux CPU | — | ||
| Linux GPU | — | ||
| Windows GPU | — | — | |
| Linux (ppc64le) CPU | — | ||
| Linux (ppc64le) GPU | — |
See also the ci.pytorch.org HUD.
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.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, Trainer and other utility functions for convenience |
Usually one uses PyTorch either as:
- a replacement for NumPy to use the power of GPUs.
- a deep learning research platform that provides maximum flexibility and speed.
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 (TH, THC, THNN, THCUNN) 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. There is no wrapper code that needs to be written. You can see a tutorial here and an example here.
Commands to install from binaries via Conda or pip wheels are on our website: https://pytorch.org
If you are installing from source, we highly recommend installing an Anaconda environment. You will get a high-quality BLAS library (MKL) and you get a controlled compiler version regardless of your Linux distro.
Once you have Anaconda installed, here are the instructions.
If you want to compile with CUDA support, install
- NVIDIA CUDA 7.5 or above
- NVIDIA cuDNN v6.x or above
If you want to disable CUDA support, export environment variable NO_CUDA=1.
Other potentially useful environment variables may be found in setup.py.
If you want to build on Windows, Visual Studio 2017 14.11 toolset and NVTX are also needed. Especially, for CUDA 8 build on Windows, there will be an additional requirement for VS 2015 Update 3 and a patch for it. The details of the patch can be found out here.
Common
conda install numpy pyyaml mkl mkl-include setuptools cmake cffi typing
On Linux
# Add LAPACK support for the GPU if needed
conda install -c pytorch magma-cuda92 # or [magma-cuda80 | magma-cuda91] depending on your cuda versiongit clone --recursive https://github.com/pytorch/pytorch
cd pytorchOn Linux
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
python setup.py installOn macOS
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py installOn Windows
set "VS150COMNTOOLS=C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Auxiliary\Build"
set CMAKE_GENERATOR=Visual Studio 15 2017 Win64
set DISTUTILS_USE_SDK=1
REM The following two lines are needed for Python 2.7, but the support for it is very experimental.
set MSSdk=1
set FORCE_PY27_BUILD=1
REM As for CUDA 8, VS2015 Update 3 is also required to build PyTorch. Use the following line.
set "CUDAHOSTCXX=%VS140COMNTOOLS%\..\..\VC\bin\amd64\cl.exe"
call "%VS150COMNTOOLS%\vcvarsall.bat" x64 -vcvars_ver=14.11
python setup.py installDockerfile is supplied to build images with cuda support and cudnn v7. You can pass -e PYTHON_VERSION=x.y flag to specify which Python version is to be used by Miniconda, or leave it unset to use the default. Build from pytorch repo directory as docker needs to copy git repo into docker filesystem while building the image.
docker build -t pytorch -f docker/pytorch/Dockerfile .
You can also pull a pre-built docker image from Docker Hub and run with nvidia-docker, but this is not currently maintained and will pull PyTorch 0.2.
nvidia-docker run --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.
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.
Installation instructions and binaries for previous PyTorch versions may be found on our website.
Three pointers to get you started:
- Tutorials: get you started with understanding and using PyTorch
- Examples: easy to understand pytorch code across all domains
- The API Reference
- forums: discuss implementations, research, etc. https://discuss.pytorch.org
- GitHub issues: bug reports, feature requests, install issues, RFCs, thoughts, etc.
- Slack: general chat, online discussions, collaboration etc. https://pytorch.slack.com/ . Our slack channel is invite-only to promote a healthy balance between power-users and beginners. If you need a slack invite, ping us at slack@pytorch.org
- newsletter: no-noise, one-way email newsletter with important announcements about pytorch. You can sign-up here: https://eepurl.com/cbG0rv
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.
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 10s 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 Kopf, 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 in the Torch community and has helped with many things Torch and PyTorch.
PyTorch is BSD-style licensed, as found in the LICENSE file.