pytorch/tools
Michael Suo d321be61c0 [ci] remove dead code related to test selection (#81163)
Since we are using Rockset for all this now, remove the code that used
the S3 path.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/81163
Approved by: https://github.com/janeyx99
2022-07-12 04:50:19 +00:00
..
amd_build Revert "To add hipify_torch as a submodule in pytorch/third_party (#74704)" 2022-06-21 23:54:00 +00:00
autograd [forward ad] Fix codegen to ignore undefined outputs (#81114) 2022-07-11 15:01:39 +00:00
bazel_tools [bazel] Add --config=shell for easier debugging (#79350) 2022-06-25 02:00:40 +00:00
build_defs Consolidate all python targets in the tools folder (#80408) 2022-06-29 23:27:47 +00:00
code_analyzer
code_coverage
config
coverage_plugins_package
fast_nvcc
gdb
iwyu
jit Consolidate all python targets in the tools folder (#80408) 2022-06-29 23:27:47 +00:00
linter Add ufmt to unify black and usort (#81157) 2022-07-11 19:20:22 +00:00
lite_interpreter Consolidate all python targets in the tools folder (#80408) 2022-06-29 23:27:47 +00:00
lldb
onnx
pyi Revert "Remove split functional wrapper (#74727)" 2022-07-11 18:29:45 +00:00
rules [AutoAccept][Codemod][FBSourceBuckFormatLinter] Daily arc lint --take BUCKFORMAT (#79692) 2022-06-16 17:01:17 +00:00
setup_helpers Consolidate all python targets in the tools folder (#80408) 2022-06-29 23:27:47 +00:00
shared
stats [ci] remove dead code related to test selection (#81163) 2022-07-12 04:50:19 +00:00
test [torchgen] Support multiple namespace in NativeFunctions.h (#79733) 2022-07-08 21:56:52 +00:00
testing [ci] remove dead code related to test selection (#81163) 2022-07-12 04:50:19 +00:00
__init__.py
bazel.bzl move gen_aten and gen_aten_hip into shared build structure 2022-06-15 18:22:52 +00:00
BUCK.bzl Consolidate all python targets in the tools folder (#80408) 2022-06-29 23:27:47 +00:00
BUCK.oss Consolidate all python targets in the tools folder (#80408) 2022-06-29 23:27:47 +00:00
build_libtorch.py
build_pytorch_libs.py
cpuinfo_target_definition.bzl
download_mnist.py
extract_scripts.py
gen_flatbuffers.sh
gen_vulkan_spv.py Consolidate all python targets in the tools folder (#80408) 2022-06-29 23:27:47 +00:00
generate_torch_version.py
generated_dirs.txt
git_add_generated_dirs.sh
git_reset_generated_dirs.sh
miniz_target_definition.bzl Updating miniz library from version 2.0.8 -> 2.1.0 (#79636) 2022-06-22 15:02:16 +00:00
nightly.py
nvcc_fix_deps.py
perf_kernel_defs.bzl [caffe2/perfkernels] Avoid native.host_info() in build files (#80812) 2022-07-06 22:52:59 +00:00
pytorch.version
README.md
render_junit.py
sgx_aten_target_definitions.bzl
sgx_caffe2_target_definitions.bzl [caffe2/perfkernels] Avoid native.host_info() in build files (#80812) 2022-07-06 22:52:59 +00:00
sgx_target_definitions.bzl
substitute.py move substitute lib to open source (#79093) 2022-06-08 23:00:47 +00:00
target_definitions.bzl Update llvm deps for Buck build (#79919) 2022-06-21 17:54:01 +00:00
update_masked_docs.py
vscode_settings.py

This folder contains a number of scripts which are used as part of the PyTorch build process. This directory also doubles as a Python module hierarchy (thus the __init__.py).

Overview

Modern infrastructure:

  • autograd - Code generation for autograd. This includes definitions of all our derivatives.
  • jit - Code generation for JIT
  • shared - Generic infrastructure that scripts in tools may find useful.
    • module_loader.py - Makes it easier to import arbitrary Python files in a script, without having to add them to the PYTHONPATH first.

Build system pieces:

  • setup_helpers - Helper code for searching for third-party dependencies on the user system.
  • build_pytorch_libs.py - cross-platform script that builds all of the constituent libraries of PyTorch, but not the PyTorch Python extension itself.
  • build_libtorch.py - Script for building libtorch, a standalone C++ library without Python support. This build script is tested in CI.
  • fast_nvcc - Mostly-transparent wrapper over nvcc that parallelizes compilation when used to build CUDA files for multiple architectures at once.
    • fast_nvcc.py - Python script, entrypoint to the fast nvcc wrapper.

Developer tools which you might find useful:

Important if you want to run on AMD GPU:

  • amd_build - HIPify scripts, for transpiling CUDA into AMD HIP. Right now, PyTorch and Caffe2 share logic for how to do this transpilation, but have separate entry-points for transpiling either PyTorch or Caffe2 code.
    • build_amd.py - Top-level entry point for HIPifying our codebase.

Tools which are only situationally useful: