mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-06 12:20:52 +01:00
4aee92833c
Summary: Fixes https://github.com/pytorch/pytorch/issues/20271 Pull Request resolved: https://github.com/pytorch/pytorch/pull/21150 Differential Revision: D15559590 Pulled By: pjh5 fbshipit-source-id: 4063bf91464425e8efe4765dc17bb7e9b7bfccc7
37 lines
1.2 KiB
ReStructuredText
37 lines
1.2 KiB
ReStructuredText
libtorch (C++-only)
|
|
===================
|
|
|
|
The core of pytorch does not depend on Python. A
|
|
CMake-based build system compiles the C++ source code into a shared
|
|
object, libtorch.so.
|
|
|
|
Building libtorch
|
|
-----------------
|
|
|
|
You can use a python script/module located in tools package to build libtorch
|
|
::
|
|
cd <pytorch_root>
|
|
|
|
# Make a new folder to build in to avoid polluting the source directories
|
|
mkdir build_libtorch && cd build_libtorch
|
|
|
|
# You might need to export some required environment variables here.
|
|
Normally setup.py sets good default env variables, but you'll have to do
|
|
that manually.
|
|
python ../tools/build_libtorch.py
|
|
|
|
|
|
Alternatively, you can call setup.py normally and then copy the built cpp libraries. This method may have side effects to your active Python installation.
|
|
::
|
|
cd <pytorch_root>
|
|
python setup.py build
|
|
|
|
ls torch/lib/tmp_install # output is produced here
|
|
ls torch/lib/tmp_install/lib/libtorch.so # of particular interest
|
|
|
|
To produce libtorch.a rather than libtorch.so, set the environment variable `BUILD_SHARED_LIBS=OFF`.
|
|
|
|
To use ninja rather than make, set `CMAKE_GENERATOR="-GNinja" CMAKE_INSTALL="ninja install"`.
|
|
|
|
Note that we are working on eliminating tools/build_pytorch_libs.sh in favor of a unified cmake build.
|