Thanks to visit codestin.com
Credit goes to github.com

Skip to content

build extensions in parallel #1882

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
crcrpar merged 4 commits into from
Feb 25, 2025
Merged

build extensions in parallel #1882

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
f0df092
build extensions in parallel
yuantailing Feb 20, 2025
d9c3507
fix: setup.py develop supports --parallel
yuantailing Feb 21, 2025
f1ee5b1
update README.md
yuantailing Feb 21, 2025
fd28caf
Update README.md: Remove an `export` command
yuantailing Feb 24, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,7 +130,7 @@ CUDA and C++ extensions via
git clone https://github.com/NVIDIA/apex
cd apex
# if pip >= 23.1 (ref: https://pip.pypa.io/en/stable/news/#v23-1) which supports multiple `--config-settings` with the same key...
pip install -v --disable-pip-version-check --no-cache-dir --no-build-isolation --config-settings "--build-option=--cpp_ext" --config-settings "--build-option=--cuda_ext" ./
pip install -v --disable-pip-version-check --no-cache-dir --no-build-isolation --config-settings "--build-option=--cpp_ext --cuda_ext --parallel 4" ./
Copy link
Copy Markdown
Collaborator

@crcrpar crcrpar Feb 21, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Likewise --thread option, this would increase CPU mem usage, so could you separately add the example command with --thread and --parallel?

Copy link
Copy Markdown
Contributor Author

@yuantailing yuantailing Feb 21, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Updated README.md

# otherwise
pip install -v --disable-pip-version-check --no-cache-dir --no-build-isolation --global-option="--cpp_ext" --global-option="--cuda_ext" ./
```
Expand Down
36 changes: 35 additions & 1 deletion setup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
import sys
import warnings
import os
import threading
import glob
from packaging.version import parse, Version

Expand Down Expand Up @@ -859,6 +860,39 @@ def check_cudnn_version_and_warn(global_option: str, required_cudnn_version: int
)


# Patch because `setup.py bdist_wheel` does not accept the `parallel` option
parallel = None
if "--parallel" in sys.argv:
idx = sys.argv.index("--parallel")
parallel = int(sys.argv[idx + 1])
sys.argv.pop(idx + 1)
sys.argv.pop(idx)


# Prevent file conflicts when multiple extensions are compiled simultaneously
class BuildExtensionSeparateDir(BuildExtension):
build_extension_patch_lock = threading.Lock()
thread_ext_name_map = {}

def build_extension(self, ext):
with self.build_extension_patch_lock:
if not getattr(self.compiler, "_compile_separate_output_dir", False):
compile_orig = self.compiler.compile

def compile_new(*args, **kwargs):
return compile_orig(*args, **{
**kwargs,
"output_dir": os.path.join(
kwargs["output_dir"],
self.thread_ext_name_map[threading.current_thread().ident]),
})
self.compiler.compile = compile_new
self.compiler._compile_separate_output_dir = True
self.thread_ext_name_map[threading.current_thread().ident] = ext.name
objects = super().build_extension(ext)
return objects


setup(
name="apex",
version="0.1",
Expand All @@ -868,6 +902,6 @@ def check_cudnn_version_and_warn(global_option: str, required_cudnn_version: int
install_requires=["packaging>20.6"],
description="PyTorch Extensions written by NVIDIA",
ext_modules=ext_modules,
cmdclass={"build_ext": BuildExtension} if ext_modules else {},
cmdclass={"build_ext": BuildExtensionSeparateDir.with_options(parallel=parallel)} if ext_modules else {},
extras_require=extras,
)