129 lines
3.9 KiB
Python
129 lines
3.9 KiB
Python
# Copyright (c) OpenMMLab. All rights reserved.
|
|
import os
|
|
import sys
|
|
import time
|
|
from pathlib import Path
|
|
|
|
from ubuntu_utils import (cmd_result, cu_version_name, ensure_base_env,
|
|
get_job, pytorch_version)
|
|
|
|
g_jobs = 2
|
|
|
|
|
|
def install_libtorch(dep_dir):
|
|
print('-' * 10 + 'install libtorch' + '-' * 10)
|
|
time.sleep(2)
|
|
|
|
os.chdir(dep_dir)
|
|
unzipped_name = 'libtorch'
|
|
if os.path.exists(unzipped_name):
|
|
return os.path.join(dep_dir, unzipped_name)
|
|
|
|
torch_version = pytorch_version()
|
|
if torch_version is None:
|
|
print('torch version is None, try 1.11.0')
|
|
torch_version = '1.11.0'
|
|
|
|
version_name = None
|
|
|
|
# first check `nvcc` version, if failed, use `nvidia-smi`
|
|
cuda = cmd_result(
|
|
" nvcc --version | grep release | awk '{print $5}' | awk -F , '{print $1}' " # noqa: E501
|
|
)
|
|
if cuda is None or len(cuda) < 1:
|
|
cuda = cmd_result(" nvidia-smi | grep CUDA | awk '{print $9}' ")
|
|
|
|
if cuda is not None and len(cuda) > 0:
|
|
version_name = cu_version_name(cuda)
|
|
else:
|
|
version_name = 'cpu'
|
|
|
|
filename = 'libtorch-shared-with-deps-{}%2B{}.zip'.format(
|
|
torch_version, version_name)
|
|
url = 'https://download.pytorch.org/libtorch/{}/{}'.format(
|
|
version_name, filename)
|
|
os.system('wget -q --show-progress {} -O libtorch.zip'.format(url))
|
|
os.system('unzip libtorch.zip')
|
|
if not os.path.exists(unzipped_name):
|
|
print(
|
|
'download or unzip libtorch from {} failed, please check https://pytorch.org/get-started/locally/' # noqa: E501
|
|
.format(url))
|
|
return None
|
|
return os.path.join(dep_dir, unzipped_name)
|
|
|
|
|
|
def install_mmdeploy(work_dir, libtorch_dir):
|
|
print('-' * 10 + 'build and install mmdeploy' + '-' * 10)
|
|
time.sleep(3)
|
|
|
|
os.chdir(work_dir)
|
|
os.system('git submodule init')
|
|
os.system('git submodule update')
|
|
|
|
if not os.path.exists('build'):
|
|
os.system('mkdir build')
|
|
|
|
os.system('rm -rf build/CMakeCache.txt')
|
|
|
|
cmd = 'cd build && Torch_DIR={} cmake ..'.format(libtorch_dir)
|
|
cmd += ' -DMMDEPLOY_BUILD_SDK=ON '
|
|
cmd += ' -DMMDEPLOY_BUILD_EXAMPLES=ON '
|
|
cmd += ' -DMMDEPLOY_BUILD_SDK_PYTHON_API=ON '
|
|
cmd += ' -DMMDEPLOY_TARGET_DEVICES=cpu '
|
|
cmd += ' -DMMDEPLOY_TARGET_BACKENDS=torchscript '
|
|
cmd += ' -DTORCHSCRIPT_DIR={} '.format(libtorch_dir)
|
|
os.system(cmd)
|
|
|
|
os.system('cd build && make -j {} && make install'.format(g_jobs))
|
|
os.system('python3 -m pip install -e .')
|
|
try:
|
|
import mmcv
|
|
print(mmcv.__version__)
|
|
os.system('python3 tools/check_env.py')
|
|
except Exception:
|
|
print('Please install torch & mmcv later.. ≥▽≤')
|
|
return 0
|
|
|
|
|
|
def main():
|
|
"""Auto install mmdeploy with ort. To verify this script:
|
|
|
|
1) use `sudo docker run -v /path/to/mmdeploy:/root/mmdeploy -v /path/to/Miniconda3-latest-Linux-x86_64.sh:/root/miniconda.sh -it ubuntu:18.04 /bin/bash` # noqa: E501
|
|
2) install conda and setup python environment
|
|
3) run `python3 tools/scripts/build_ubuntu_x64_torchscript.py`
|
|
|
|
Returns:
|
|
_type_: _description_
|
|
"""
|
|
global g_jobs
|
|
g_jobs = get_job(sys.argv)
|
|
print('g_jobs {}'.format(g_jobs))
|
|
|
|
work_dir = os.path.abspath(os.path.join(__file__, '..', '..', '..'))
|
|
dep_dir = os.path.abspath(os.path.join(work_dir, '..', 'mmdeploy-dep'))
|
|
if not os.path.exists(dep_dir):
|
|
if os.path.isfile(dep_dir):
|
|
print('{} already exists and it is a file, exit.'.format(work_dir))
|
|
return -1
|
|
os.mkdir(dep_dir)
|
|
|
|
success = ensure_base_env(work_dir, dep_dir)
|
|
if success != 0:
|
|
return -1
|
|
|
|
libtorch_dir = install_libtorch(dep_dir)
|
|
|
|
if libtorch_dir is None:
|
|
return -1
|
|
|
|
if install_mmdeploy(work_dir, libtorch_dir) != 0:
|
|
return -1
|
|
|
|
if os.path.exists(Path('~/mmdeploy.env').expanduser()):
|
|
print('Please source ~/mmdeploy.env to setup your env !')
|
|
os.system('cat ~/mmdeploy.env')
|
|
|
|
|
|
if __name__ == '__main__':
|
|
main()
|