Files
pytorch/tools/build_pytorch_libs.py
Hong Xu b159e0ce08 Significantly simplify the spawning of pytorch libs building process. (#21105)
Summary:
Instead of attempting to hardcode calls to "ninja" or "make", we should always let cmake do it. This better integrates build configurations (DEBUG or REL_WITH_DEB_INFO) and better handles the case in which the native build tool is not in PATH (cmake has some capacity to find them and has options for users to specify their locations).
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21105

Differential Revision: D15602883

Pulled By: soumith

fbshipit-source-id: 32ac46d438af00e791defde6ae5ac21c437d0bb0
2019-06-03 08:28:19 -07:00

331 lines
12 KiB
Python

from .setup_helpers.env import (IS_64BIT, IS_DARWIN, IS_WINDOWS,
DEBUG, REL_WITH_DEB_INFO, USE_MKLDNN,
check_env_flag, check_negative_env_flag)
import os
import sys
import distutils
import distutils.sysconfig
from subprocess import check_call, check_output
from distutils.version import LooseVersion
from .setup_helpers.cuda import USE_CUDA, CUDA_HOME
from .setup_helpers.dist_check import USE_DISTRIBUTED, USE_GLOO_IBVERBS
from .setup_helpers.nccl import USE_SYSTEM_NCCL, NCCL_INCLUDE_DIR, NCCL_ROOT_DIR, NCCL_SYSTEM_LIB, USE_NCCL
from .setup_helpers.rocm import USE_ROCM
from .setup_helpers.nnpack import USE_NNPACK
from .setup_helpers.qnnpack import USE_QNNPACK
from .setup_helpers.cudnn import CUDNN_INCLUDE_DIR, CUDNN_LIBRARY, USE_CUDNN
from pprint import pprint
from glob import glob
import multiprocessing
import shutil
def which(thefile):
path = os.environ.get("PATH", os.defpath).split(os.pathsep)
for dir in path:
fname = os.path.join(dir, thefile)
fnames = [fname]
if IS_WINDOWS:
exts = os.environ.get('PATHEXT', '').split(os.pathsep)
fnames += [fname + ext for ext in exts]
for name in fnames:
if (os.path.exists(name) and os.access(name, os.F_OK | os.X_OK)
and not os.path.isdir(name)):
return name
return None
def cmake_version(cmd):
for line in check_output([cmd, '--version']).decode('utf-8').split('\n'):
if 'version' in line:
return LooseVersion(line.strip().split(' ')[2])
raise Exception('no version found')
def get_cmake_command():
cmake_command = 'cmake'
if IS_WINDOWS:
return cmake_command
cmake3 = which('cmake3')
if cmake3 is not None:
cmake = which('cmake')
if cmake is not None:
bare_version = cmake_version(cmake)
if bare_version < LooseVersion("3.5.0") and cmake_version(cmake3) > bare_version:
cmake_command = 'cmake3'
return cmake_command
def cmake_defines(lst, **kwargs):
for key in sorted(kwargs.keys()):
value = kwargs[key]
if value is not None:
lst.append('-D{}={}'.format(key, value))
# Ninja
# Use ninja if it is on the PATH. Previous version of PyTorch required the
# ninja python package, but we no longer use it, so we do not have to import it
USE_NINJA = not check_negative_env_flag('USE_NINJA') and (which('ninja') is not None)
base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
install_dir = base_dir + "/torch"
build_type = "Release"
if DEBUG:
build_type = "Debug"
elif REL_WITH_DEB_INFO:
build_type = "RelWithDebInfo"
def overlay_windows_vcvars(env):
if sys.version_info >= (3, 5):
from distutils._msvccompiler import _get_vc_env
vc_arch = 'x64' if IS_64BIT else 'x86'
vc_env = _get_vc_env(vc_arch)
# Keys in `_get_vc_env` are always lowercase.
# We turn them into uppercase before overlaying vcvars
# because OS environ keys are always uppercase on Windows.
# https://stackoverflow.com/a/7797329
vc_env = {k.upper(): v for k, v in vc_env.items()}
for k, v in env.items():
uk = k.upper()
if uk not in vc_env:
vc_env[uk] = v
return vc_env
else:
return env
def mkdir_p(dir):
try:
os.makedirs(dir)
except OSError:
pass
def create_build_env():
# XXX - our cmake file sometimes looks at the system environment
# and not cmake flags!
# you should NEVER add something to this list. It is bad practice to
# have cmake read the environment
my_env = os.environ.copy()
if USE_CUDNN:
my_env['CUDNN_LIBRARY'] = escape_path(CUDNN_LIBRARY)
my_env['CUDNN_INCLUDE_DIR'] = escape_path(CUDNN_INCLUDE_DIR)
if USE_CUDA:
my_env['CUDA_BIN_PATH'] = escape_path(CUDA_HOME)
if IS_WINDOWS and USE_NINJA:
# When using Ninja under Windows, the gcc toolchain will be chosen as
# default. But it should be set to MSVC as the user's first choice.
my_env = overlay_windows_vcvars(my_env)
my_env.setdefault('CC', 'cl')
my_env.setdefault('CXX', 'cl')
return my_env
def run_cmake(version,
cmake_python_library,
build_python,
build_test,
build_dir,
my_env):
cmake_args = [
get_cmake_command()
]
if USE_NINJA:
cmake_args.append('-GNinja')
elif IS_WINDOWS:
cmake_args.append('-GVisual Studio 15 2017')
if IS_64BIT:
cmake_args.append('-Ax64')
cmake_args.append('-Thost=x64')
try:
import numpy as np
except ImportError:
USE_NUMPY = False
NUMPY_INCLUDE_DIR = None
else:
NUMPY_INCLUDE_DIR = np.get_include()
USE_NUMPY = True
cflags = os.getenv('CFLAGS', "") + " " + os.getenv('CPPFLAGS', "")
ldflags = os.getenv('LDFLAGS', "")
if IS_WINDOWS:
cmake_defines(cmake_args, MSVC_Z7_OVERRIDE=os.getenv('MSVC_Z7_OVERRIDE', "ON"))
cflags += " /EHa"
mkdir_p(install_dir)
mkdir_p(build_dir)
cmake_defines(
cmake_args,
PYTHON_EXECUTABLE=escape_path(sys.executable),
PYTHON_LIBRARY=escape_path(cmake_python_library),
PYTHON_INCLUDE_DIR=escape_path(distutils.sysconfig.get_python_inc()),
BUILDING_WITH_TORCH_LIBS=os.getenv("BUILDING_WITH_TORCH_LIBS", "ON"),
TORCH_BUILD_VERSION=version,
CMAKE_BUILD_TYPE=build_type,
BUILD_PYTHON=build_python,
BUILD_SHARED_LIBS=os.getenv("BUILD_SHARED_LIBS", "ON"),
BUILD_BINARY=check_env_flag('BUILD_BINARY'),
BUILD_TEST=build_test,
INSTALL_TEST=build_test,
BUILD_CAFFE2_OPS=not check_negative_env_flag('BUILD_CAFFE2_OPS'),
ONNX_NAMESPACE=os.getenv("ONNX_NAMESPACE", "onnx_torch"),
ONNX_ML=os.getenv("ONNX_ML", False),
USE_CUDA=USE_CUDA,
USE_DISTRIBUTED=USE_DISTRIBUTED,
USE_FBGEMM=not (check_env_flag('NO_FBGEMM') or check_negative_env_flag('USE_FBGEMM')),
NAMEDTENSOR_ENABLED=(check_env_flag('USE_NAMEDTENSOR') or check_negative_env_flag('NO_NAMEDTENSOR')),
USE_NUMPY=USE_NUMPY,
NUMPY_INCLUDE_DIR=escape_path(NUMPY_INCLUDE_DIR),
USE_SYSTEM_NCCL=USE_SYSTEM_NCCL,
NCCL_INCLUDE_DIR=NCCL_INCLUDE_DIR,
NCCL_ROOT_DIR=NCCL_ROOT_DIR,
NCCL_SYSTEM_LIB=NCCL_SYSTEM_LIB,
CAFFE2_STATIC_LINK_CUDA=check_env_flag('USE_CUDA_STATIC_LINK'),
USE_ROCM=USE_ROCM,
USE_NNPACK=USE_NNPACK,
USE_LEVELDB=check_env_flag('USE_LEVELDB'),
USE_LMDB=check_env_flag('USE_LMDB'),
USE_OPENCV=check_env_flag('USE_OPENCV'),
USE_QNNPACK=USE_QNNPACK,
USE_TENSORRT=check_env_flag('USE_TENSORRT'),
USE_FFMPEG=check_env_flag('USE_FFMPEG'),
USE_SYSTEM_EIGEN_INSTALL="OFF",
USE_MKLDNN=USE_MKLDNN,
USE_NCCL=USE_NCCL,
NCCL_EXTERNAL=USE_NCCL,
CMAKE_INSTALL_PREFIX=install_dir,
CMAKE_C_FLAGS=cflags,
CMAKE_CXX_FLAGS=cflags,
CMAKE_EXE_LINKER_FLAGS=ldflags,
CMAKE_SHARED_LINKER_FLAGS=ldflags,
THD_SO_VERSION="1",
CMAKE_PREFIX_PATH=os.getenv('CMAKE_PREFIX_PATH') or distutils.sysconfig.get_python_lib(),
BLAS=os.getenv('BLAS'),
CUDA_NVCC_EXECUTABLE=escape_path(os.getenv('CUDA_NVCC_EXECUTABLE')),
USE_REDIS=os.getenv('USE_REDIS'),
USE_GLOG=os.getenv('USE_GLOG'),
USE_GFLAGS=os.getenv('USE_GFLAGS'),
USE_ASAN=check_env_flag('USE_ASAN'),
WERROR=os.getenv('WERROR'))
if os.getenv('_GLIBCXX_USE_CXX11_ABI'):
cmake_defines(cmake_args, GLIBCXX_USE_CXX11_ABI=os.getenv('_GLIBCXX_USE_CXX11_ABI'))
if os.getenv('USE_OPENMP'):
cmake_defines(cmake_args, USE_OPENMP=check_env_flag('USE_OPENMP'))
if os.getenv('USE_TBB'):
cmake_defines(cmake_args, USE_TBB=check_env_flag('USE_TBB'))
if os.getenv('MKL_SEQ'):
cmake_defines(cmake_args, INTEL_MKL_SEQUENTIAL=check_env_flag('MKL_SEQ'))
if os.getenv('MKL_TBB'):
cmake_defines(cmake_args, INTEL_MKL_TBB=check_env_flag('MKL_TBB'))
mkldnn_threading = os.getenv('MKLDNN_THREADING')
if mkldnn_threading:
cmake_defines(cmake_args, MKLDNN_THREADING=mkldnn_threading)
parallel_backend = os.getenv('PARALLEL_BACKEND')
if parallel_backend:
cmake_defines(cmake_args, PARALLEL_BACKEND=parallel_backend)
if USE_GLOO_IBVERBS:
cmake_defines(cmake_args, USE_IBVERBS="1", USE_GLOO_IBVERBS="1")
if USE_MKLDNN:
cmake_defines(cmake_args, MKLDNN_ENABLE_CONCURRENT_EXEC="ON")
expected_wrapper = '/usr/local/opt/ccache/libexec'
if IS_DARWIN and os.path.exists(expected_wrapper):
cmake_defines(cmake_args,
CMAKE_C_COMPILER="{}/gcc".format(expected_wrapper),
CMAKE_CXX_COMPILER="{}/g++".format(expected_wrapper))
for env_var_name in my_env:
if env_var_name.startswith('gh'):
# github env vars use utf-8, on windows, non-ascii code may
# cause problem, so encode first
try:
my_env[env_var_name] = str(my_env[env_var_name].encode("utf-8"))
except UnicodeDecodeError as e:
shex = ':'.join('{:02x}'.format(ord(c)) for c in my_env[env_var_name])
sys.stderr.write('Invalid ENV[{}] = {}\n'.format(env_var_name, shex))
# According to the CMake manual, we should pass the arguments first,
# and put the directory as the last element. Otherwise, these flags
# may not be passed correctly.
# Reference:
# 1. https://cmake.org/cmake/help/latest/manual/cmake.1.html#synopsis
# 2. https://stackoverflow.com/a/27169347
cmake_args.append(base_dir)
pprint(cmake_args)
check_call(cmake_args, cwd=build_dir, env=my_env)
def build_caffe2(version,
cmake_python_library,
build_python,
rerun_cmake,
cmake_only,
build_dir):
my_env = create_build_env()
build_test = not check_negative_env_flag('BUILD_TEST')
max_jobs = os.getenv('MAX_JOBS', str(multiprocessing.cpu_count()))
cmake_cache_file = 'build/CMakeCache.txt'
if rerun_cmake and os.path.isfile(cmake_cache_file):
os.remove(cmake_cache_file)
if not os.path.exists(cmake_cache_file) or (USE_NINJA and not os.path.exists('build/build.ninja')):
run_cmake(version,
cmake_python_library,
build_python,
build_test,
build_dir,
my_env)
if cmake_only:
return
build_cmd = [get_cmake_command(),
'--build', '.', '--target', 'install', '--config', build_type]
# This ``if-else'' clause would be unnecessary when cmake 3.12 becomes
# minimum, which provides a '-j' option: build_cmd += ['-j', max_jobs]
# would be sufficient by then.
if IS_WINDOWS and not USE_NINJA: # We are likely using msbuild here
build_cmd += ['--', '/maxcpucount:{}'.format(max_jobs)]
else:
build_cmd += ['--', '-j', max_jobs]
check_call(build_cmd, cwd=build_dir, env=my_env)
# in cmake, .cu compilation involves generating certain intermediates
# such as .cu.o and .cu.depend, and these intermediates finally get compiled
# into the final .so.
# Ninja updates build.ninja's timestamp after all dependent files have been built,
# and re-kicks cmake on incremental builds if any of the dependent files
# have a timestamp newer than build.ninja's timestamp.
# There is a cmake bug with the Ninja backend, where the .cu.depend files
# are still compiling by the time the build.ninja timestamp is updated,
# so the .cu.depend file's newer timestamp is screwing with ninja's incremental
# build detector.
# This line works around that bug by manually updating the build.ninja timestamp
# after the entire build is finished.
if os.path.exists('build/build.ninja'):
os.utime('build/build.ninja', None)
if build_python:
for proto_file in glob('build/caffe2/proto/*.py'):
if os.path.sep != '/':
proto_file = proto_file.replace(os.path.sep, '/')
if proto_file != 'build/caffe2/proto/__init__.py':
shutil.copyfile(proto_file, "caffe2/proto/" + os.path.basename(proto_file))
def escape_path(path):
if os.path.sep != '/' and path is not None:
return path.replace(os.path.sep, '/')
return path