Other languages:

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

PyTorch has a distant connection with Torch, but for all practical purposes you can treat them as separate projects.

PyTorch developers also offer LibTorch, which allows one to implement extensions to PyTorch using C++, and to implement pure C++ machine learning applications. Models written in Python using PyTorch can be converted and used in pure C++ through TorchScript.

Installation

Latest available wheels

To see the latest version of PyTorch that we have built:

[name@server ~]$ avail_wheels "torch*"

For more information on listing wheels, see listing available wheels.

Installing Compute Canada wheel

The preferred option is to install it using the Python wheel as follows:

1. Load a Python module, either python/2.7, python/3.5, python/3.6 or python/3.7

2. Create and start a virtual environment.

3. Install PyTorch in the virtual environment with pip install.

GPU and CPU

(venv) [name@server ~] pip install torch --no-index

Extra

In addition to torch, you can install torchvision, torchtext and torchaudio:

(venv) [name@server ~] pip install torch torchvision torchtext torchaudio --no-index

Job submission

Here is an example of a job submission script using the python wheel, with a virtual environment inside a job:

File : pytorch-test.sh

#!/bin/bash
#SBATCH --gres=gpu:1       # Request GPU "generic resources"
#SBATCH --cpus-per-task=6  # Cores proportional to GPUs: 6 on Cedar, 16 on Graham.
#SBATCH --mem=32000M       # Memory proportional to GPUs: 32000 Cedar, 64000 Graham.
#SBATCH --time=0-03:00
#SBATCH --output=%N-%j.out

module load python/3.6
virtualenv --no-download $SLURM_TMPDIR/env
source $SLURM_TMPDIR/env/bin/activate
pip install torch --no-index

python pytorch-test.py

The Python script pytorch-test.py has the form

File : pytorch-test.py

import torch
x = torch.Tensor(5, 3)
print(x)
y = torch.rand(5, 3)
print(y)
# let us run the following only if CUDA is available
if torch.cuda.is_available():
    x = x.cuda()
    y = y.cuda()
    print(x + y)

You can then submit a PyTorch job with:

[name@server ~]$ sbatch pytorch-test.sh

Troubleshooting

Memory leak

On AVX512 hardware (Béluga, Skylake or V100 nodes), older versions of Pytorch (less than v1.0.1) using older libraries (cuDNN < v7.5 or MAGMA < v2.5) may considerably leak memory resulting in an out-of-memory exception and death of your tasks. Please upgrade to the latest torch version.

LibTorch

LibTorch allows one to implement both C++ extensions to PyTorch and pure C++ machine learning applications. It contains "all headers, libraries and CMake configuration files required to depend on PyTorch" (as mentioned in the docs).

How to use LibTorch

Get the library

wget https://download.pytorch.org/libtorch/cu100/libtorch-shared-with-deps-latest.zip
unzip libtorch-shared-with-deps-latest.zip
cd libtorch
export LIBTORCH_ROOT=$(pwd)  # this variable is used in the example below

Patch the library (this workaround is needed for compiling on Compute Canada clusters):

sed -i -e 's/\/usr\/local\/cuda\/lib64\/libculibos.a;dl;\/usr\/local\/cuda\/lib64\/libculibos.a;//g' share/cmake/Caffe2/Caffe2Targets.cmake

The library is also included in the PyTorch wheel, but this is not the recommended way to acquire it. Once Pytorch is installed in a virtual environment, you can find it at: $VIRTUAL_ENV/lib/python3.6/site-packages/torch/lib/libtorch.so.

Compile a minimal example

Create the following two files:

File : example-app.cpp

#include <torch/torch.h>
#include <iostream>

int main() {
    torch::Device device(torch::kCPU);
    if (torch::cuda::is_available()) {
        std::cout << "CUDA is available! Using GPU." << std::endl;
        device = torch::Device(torch::kCUDA);
    }

    torch::Tensor tensor = torch::rand({2, 3}).to(device);
    std::cout << tensor << std::endl;
}

File : CMakeLists.txt

cmake_minimum_required(VERSION 3.0 FATAL_ERROR)
project(example-app)

find_package(Torch REQUIRED)

add_executable(example-app example-app.cpp)
target_link_libraries(example-app "${TORCH_LIBRARIES}")
set_property(TARGET example-app PROPERTY CXX_STANDARD 11)

Load the necessary modules:

module load cmake intel/2018.3 cuda/10 cudnn

Compile the program:

mkdir build
cd build
cmake -DCMAKE_PREFIX_PATH="$LIBTORCH_ROOT;$EBROOTCUDA;$EBROOTCUDNN" ..
make

Run the program:

./example-app

To test an application with CUDA, request an interactive job with a GPU.

Resources

https://pytorch.org/cppdocs/