ColossalAI/setup.py

236 lines
9.9 KiB
Python
Raw Normal View History

2021-10-28 16:21:23 +00:00
import os
import subprocess
import re
2022-02-14 09:09:30 +00:00
from setuptools import find_packages, setup
2021-10-28 16:21:23 +00:00
# ninja build does not work unless include_dirs are abs path
this_dir = os.path.dirname(os.path.abspath(__file__))
2022-02-14 09:09:30 +00:00
build_cuda_ext = True
ext_modules = []
if int(os.environ.get('NO_CUDA_EXT', '0')) == 1:
2022-02-14 09:09:30 +00:00
build_cuda_ext = False
2021-10-28 16:21:23 +00:00
def get_cuda_bare_metal_version(cuda_dir):
2022-01-13 08:47:17 +00:00
raw_output = subprocess.check_output([cuda_dir + "/bin/nvcc", "-V"], universal_newlines=True)
2021-10-28 16:21:23 +00:00
output = raw_output.split()
release_idx = output.index("release") + 1
release = output[release_idx].split(".")
bare_metal_major = release[0]
bare_metal_minor = release[1][0]
return raw_output, bare_metal_major, bare_metal_minor
def check_cuda_torch_binary_vs_bare_metal(cuda_dir):
2022-01-13 08:47:17 +00:00
raw_output, bare_metal_major, bare_metal_minor = get_cuda_bare_metal_version(cuda_dir)
torch_binary_major = torch.version.cuda.split(".")[0]
torch_binary_minor = torch.version.cuda.split(".")[1]
print("\nCompiling cuda extensions with")
print(raw_output + "from " + cuda_dir + "/bin\n")
2022-02-14 09:09:30 +00:00
if bare_metal_major != torch_binary_major:
2022-03-09 07:11:35 +00:00
print(f'The detected CUDA version ({raw_output}) mismatches the version that was used to compile PyTorch '
f'({torch.version.cuda}). CUDA extension will not be installed.')
2022-02-14 09:09:30 +00:00
return False
if bare_metal_minor != torch_binary_minor:
print("\nWarning: Cuda extensions are being compiled with a version of Cuda that does "
2022-03-09 07:11:35 +00:00
"not match the version used to compile Pytorch binaries. "
f"Pytorch binaries were compiled with Cuda {torch.version.cuda}.\n"
"In some cases, a minor-version mismatch will not cause later errors: "
"https://github.com/NVIDIA/apex/pull/323#discussion_r287021798. ")
2022-02-14 09:09:30 +00:00
return True
def check_cuda_availability(cuda_dir):
if not torch.cuda.is_available():
# https://github.com/NVIDIA/apex/issues/486
2022-03-09 07:11:35 +00:00
# Extension builds after https://github.com/pytorch/pytorch/pull/23408 attempt to query
# torch.cuda.get_device_capability(), which will fail if you are compiling in an environment
# without visible GPUs (e.g. during an nvidia-docker build command).
print(
'\nWarning: Torch did not find available GPUs on this system.\n',
'If your intention is to cross-compile, this is not an error.\n'
'By default, Colossal-AI will cross-compile for Pascal (compute capabilities 6.0, 6.1, 6.2),\n'
'Volta (compute capability 7.0), Turing (compute capability 7.5),\n'
'and, if the CUDA version is >= 11.0, Ampere (compute capability 8.0).\n'
'If you wish to cross-compile for a single specific architecture,\n'
'export TORCH_CUDA_ARCH_LIST="compute capability" before running setup.py.\n')
2022-02-14 09:09:30 +00:00
if os.environ.get("TORCH_CUDA_ARCH_LIST", None) is None:
_, bare_metal_major, _ = get_cuda_bare_metal_version(cuda_dir)
if int(bare_metal_major) == 11:
os.environ["TORCH_CUDA_ARCH_LIST"] = "6.0;6.1;6.2;7.0;7.5;8.0"
else:
os.environ["TORCH_CUDA_ARCH_LIST"] = "6.0;6.1;6.2;7.0;7.5"
return False
if cuda_dir is None:
2022-03-09 07:11:35 +00:00
print("nvcc was not found. CUDA extension will not be installed. If you're installing within a container from "
"https://hub.docker.com/r/pytorch/pytorch, only images whose names contain 'devel' will provide nvcc.")
2022-02-14 09:09:30 +00:00
return False
return True
2022-01-13 08:47:17 +00:00
def append_nvcc_threads(nvcc_extra_args):
_, bare_metal_major, bare_metal_minor = get_cuda_bare_metal_version(CUDA_HOME)
if int(bare_metal_major) >= 11 and int(bare_metal_minor) >= 2:
return nvcc_extra_args + ["--threads", "4"]
return nvcc_extra_args
def fetch_requirements(path):
with open(path, 'r') as fd:
return [r.strip() for r in fd.readlines()]
2022-02-15 07:15:03 +00:00
def fetch_readme():
with open('README.md', encoding='utf-8') as f:
return f.read()
def get_version():
with open('version.txt') as f:
2022-04-27 07:26:12 +00:00
version = f.read().strip()
if build_cuda_ext:
2022-05-05 06:59:12 +00:00
torch_version = '.'.join(torch.__version__.split('.')[:2])
cuda_version = '.'.join(get_cuda_bare_metal_version(CUDA_HOME)[1:])
2022-04-27 07:26:12 +00:00
version += f'+torch{torch_version}cu{cuda_version}'
return version
2022-02-15 07:15:03 +00:00
2022-02-14 09:09:30 +00:00
if build_cuda_ext:
try:
import torch
2022-03-09 07:11:35 +00:00
from torch.utils.cpp_extension import (CUDA_HOME, BuildExtension, CUDAExtension)
2022-02-14 09:09:30 +00:00
print("\n\ntorch.__version__ = {}\n\n".format(torch.__version__))
TORCH_MAJOR = int(torch.__version__.split('.')[0])
TORCH_MINOR = int(torch.__version__.split('.')[1])
if TORCH_MAJOR < 1 or (TORCH_MAJOR == 1 and TORCH_MINOR < 8):
raise RuntimeError("Colossal-AI requires Pytorch 1.8 or newer.\n"
2022-03-09 07:11:35 +00:00
"The latest stable release can be obtained from https://pytorch.org/")
2022-02-14 09:09:30 +00:00
except ImportError:
print('torch is not found. CUDA extension will not be installed')
build_cuda_ext = False
if build_cuda_ext:
build_cuda_ext = check_cuda_availability(CUDA_HOME) and check_cuda_torch_binary_vs_bare_metal(CUDA_HOME)
if build_cuda_ext:
# Set up macros for forward/backward compatibility hack around
# https://github.com/pytorch/pytorch/commit/4404762d7dd955383acee92e6f06b48144a0742e
# and
# https://github.com/NVIDIA/apex/issues/456
# https://github.com/pytorch/pytorch/commit/eb7b39e02f7d75c26d8a795ea8c7fd911334da7e#diff-4632522f237f1e4e728cb824300403ac
version_dependent_macros = ['-DVERSION_GE_1_1', '-DVERSION_GE_1_3', '-DVERSION_GE_1_5']
def cuda_ext_helper(name, sources, extra_cuda_flags, extra_cxx_flags=[]):
2022-03-09 07:11:35 +00:00
return CUDAExtension(
name=name,
sources=[os.path.join('colossalai/kernel/cuda_native/csrc', path) for path in sources],
include_dirs=[os.path.join(this_dir, 'colossalai/kernel/cuda_native/csrc/kernels/include')],
extra_compile_args={
'cxx': ['-O3'] + version_dependent_macros + extra_cxx_flags,
'nvcc': append_nvcc_threads(['-O3', '--use_fast_math'] + version_dependent_macros + extra_cuda_flags)
})
cc_flag = []
for arch in torch.cuda.get_arch_list():
res = re.search(r'sm_(\d+)', arch)
if res:
arch_cap = res[1]
if int(arch_cap) >= 60:
cc_flag.extend(['-gencode', f'arch=compute_{arch_cap},code={arch}'])
2022-02-14 09:09:30 +00:00
2022-05-07 10:01:04 +00:00
extra_cuda_flags = ['-lineinfo']
ext_modules.append(
cuda_ext_helper('colossal_C', [
'colossal_C_frontend.cpp', 'multi_tensor_sgd_kernel.cu', 'multi_tensor_scale_kernel.cu',
'multi_tensor_adam.cu', 'multi_tensor_l2norm_kernel.cu', 'multi_tensor_lamb.cu'
], extra_cuda_flags + cc_flag))
2022-03-09 07:11:35 +00:00
extra_cuda_flags = [
'-U__CUDA_NO_HALF_OPERATORS__', '-U__CUDA_NO_HALF_CONVERSIONS__', '--expt-relaxed-constexpr',
'--expt-extended-lambda'
]
2022-02-14 09:09:30 +00:00
2022-03-09 07:11:35 +00:00
ext_modules.append(
cuda_ext_helper('colossal_scaled_upper_triang_masked_softmax',
['scaled_upper_triang_masked_softmax.cpp', 'scaled_upper_triang_masked_softmax_cuda.cu'],
extra_cuda_flags + cc_flag))
2022-02-14 09:09:30 +00:00
2022-03-09 07:11:35 +00:00
ext_modules.append(
cuda_ext_helper('colossal_scaled_masked_softmax',
['scaled_masked_softmax.cpp', 'scaled_masked_softmax_cuda.cu'], extra_cuda_flags + cc_flag))
2022-02-14 09:09:30 +00:00
2022-03-09 07:11:35 +00:00
ext_modules.append(
cuda_ext_helper('colossal_moe_cuda', ['moe_cuda.cpp', 'moe_cuda_kernel.cu'], extra_cuda_flags + cc_flag))
2022-02-14 09:09:30 +00:00
extra_cuda_flags = ['-maxrregcount=50']
2022-03-09 07:11:35 +00:00
ext_modules.append(
cuda_ext_helper('colossal_layer_norm_cuda', ['layer_norm_cuda.cpp', 'layer_norm_cuda_kernel.cu'],
extra_cuda_flags + cc_flag))
extra_cuda_flags = [
'-std=c++14', '-U__CUDA_NO_HALF_OPERATORS__', '-U__CUDA_NO_HALF_CONVERSIONS__', '-U__CUDA_NO_HALF2_OPERATORS__',
'-DTHRUST_IGNORE_CUB_VERSION_CHECK'
]
ext_modules.append(
cuda_ext_helper('colossal_multihead_attention', [
'multihead_attention_1d.cpp', 'kernels/cublas_wrappers.cu', 'kernels/transform_kernels.cu',
'kernels/dropout_kernels.cu', 'kernels/normalize_kernels.cu', 'kernels/softmax_kernels.cu',
'kernels/general_kernels.cu', 'kernels/cuda_util.cu'
], extra_cuda_flags + cc_flag))
extra_cxx_flags = ['-std=c++14', '-lcudart', '-lcublas', '-g', '-Wno-reorder', '-fopenmp', '-march=native']
2022-03-09 07:11:35 +00:00
ext_modules.append(cuda_ext_helper('cpu_adam', ['cpu_adam.cpp'], extra_cuda_flags, extra_cxx_flags))
2021-10-28 16:21:23 +00:00
setup(
name='colossalai',
2022-02-15 07:15:03 +00:00
version=get_version(),
2022-03-09 07:11:35 +00:00
packages=find_packages(exclude=(
'benchmark',
'docker',
'tests',
'docs',
'examples',
'tests',
'scripts',
'requirements',
'*.egg-info',
)),
description='An integrated large-scale model training system with efficient parallelization techniques',
2022-02-15 07:15:03 +00:00
long_description=fetch_readme(),
long_description_content_type='text/markdown',
license='Apache Software License 2.0',
2022-02-15 07:15:03 +00:00
url='https://www.colossalai.org',
project_urls={
'Forum': 'https://github.com/hpcaitech/ColossalAI/discussions',
'Bug Tracker': 'https://github.com/hpcaitech/ColossalAI/issues',
'Examples': 'https://github.com/hpcaitech/ColossalAI-Examples',
'Documentation': 'http://colossalai.readthedocs.io',
'Github': 'https://github.com/hpcaitech/ColossalAI',
},
2021-10-28 16:21:23 +00:00
ext_modules=ext_modules,
cmdclass={'build_ext': BuildExtension} if ext_modules else {},
2022-02-14 09:09:30 +00:00
install_requires=fetch_requirements('requirements/requirements.txt'),
entry_points='''
[console_scripts]
colossalai=colossalai.cli:cli
''',
python_requires='>=3.6',
2022-02-15 07:15:03 +00:00
classifiers=[
'Programming Language :: Python :: 3',
'License :: OSI Approved :: Apache Software License',
'Environment :: GPU :: NVIDIA CUDA',
'Topic :: Scientific/Engineering :: Artificial Intelligence',
'Topic :: System :: Distributed Computing',
],
2022-01-13 08:47:17 +00:00
)