LibTorch: Difference between revisions

From Alliance Doc
Jump to navigation Jump to search
No edit summary
(Redirected page to PyTorch#LibTorch)
Tag: New redirect
 
(2 intermediate revisions by one other user not shown)
Line 1: Line 1:
<languages />
#REDIRECT [[PyTorch#LibTorch]]
[[Category:Software]]
<translate>
 
{{Draft}}
 
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."<ref>https://pytorch.org/cppdocs/installing.html (Retrieved 2019-07-12)</ref>
 
== How to use LibTorch ==
 
=== Download ===
 
<syntaxhighlight>
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
</syntaxhighlight>
 
Patch the library:
<syntaxhighlight>
sed -i -e 's/\/usr\/local\/cuda\/lib64\/libculibos.a;dl;\/usr\/local\/cuda\/lib64\/libculibos.a;//g' share/cmake/Caffe2/Caffe2Targets.cmake
</syntaxhighlight>
 
=== Compile a minimal example ===
 
Create <tt>example-app.cpp</tt>:
 
<syntaxhighlight>
#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;
}
</syntaxhighlight>
 
Create <tt>CMakeLists.txt</tt>:
 
<syntaxhighlight>
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)
</syntaxhighlight>
 
Load the necessary modules:
 
<syntaxhighlight>
module load cmake intel/2018.3 cuda/10 cudnn
</syntaxhighlight>
 
Compile the program:
 
<syntaxhighlight>
mkdir build
cd build
cmake -DCMAKE_PREFIX_PATH="$LIBTORCH_ROOT;$EBROOTCUDA;$EBROOTCUDNN" ..
make
</syntaxhighlight>
 
Run the program:
 
<syntaxhighlight>
./example-app
</syntaxhighlight>
 
To test CUDA, request an interactive job with a GPU.
 
== Resources ==
 
https://pytorch.org/cppdocs/
 
</translate>

Latest revision as of 15:34, 23 July 2019

Redirect to:

Retrieved from "https://docs.alliancecan.ca/mediawiki/index.php?title=LibTorch&oldid=74296"