xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-14 00:45:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

Build CUDA11.8 wheels for release (#1596)

Browse Source
This commit is contained in:
Woosuk Kwon 2023-11-09 03:52:29 -08:00 committed by GitHub
parent 8efe23f150
commit fd58b73a40
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 14 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
.github/workflows/publish.yml vendored
View File

@ -50,7 +50,7 @@ jobs:
os: ['ubuntu-20.04'] os: ['ubuntu-20.04']
python-version: ['3.8', '3.9', '3.10', '3.11'] python-version: ['3.8', '3.9', '3.10', '3.11']
pytorch-version: ['2.1.0'] pytorch-version: ['2.1.0']
cuda-version: ['12.1'] cuda-version: ['11.8', '12.1']
steps: steps:
- name: Checkout - name: Checkout

15
setup.py
View File

@ -12,6 +12,8 @@ from torch.utils.cpp_extension import BuildExtension, CUDAExtension, CUDA_HOME
ROOT_DIR = os.path.dirname(__file__) ROOT_DIR = os.path.dirname(__file__)
MAIN_CUDA_VERSION = "12.1"
# Supported NVIDIA GPU architectures. # Supported NVIDIA GPU architectures.
SUPPORTED_ARCHS = {"7.0", "7.5", "8.0", "8.6", "8.9", "9.0"} SUPPORTED_ARCHS = {"7.0", "7.5", "8.0", "8.6", "8.9", "9.0"}
@ -225,7 +227,7 @@ def get_path(*filepath) -> str:
return os.path.join(ROOT_DIR, *filepath) return os.path.join(ROOT_DIR, *filepath)
def find_version(filepath: str): def find_version(filepath: str) -> str:
"""Extract version information from the given filepath. """Extract version information from the given filepath.
Adapted from https://github.com/ray-project/ray/blob/0b190ee1160eeca9796bc091e07eaebf4c85b511/python/setup.py Adapted from https://github.com/ray-project/ray/blob/0b190ee1160eeca9796bc091e07eaebf4c85b511/python/setup.py
@ -238,6 +240,15 @@ def find_version(filepath: str):
raise RuntimeError("Unable to find version string.") raise RuntimeError("Unable to find version string.")
def get_vllm_version() -> str:
version = find_version(get_path("vllm", "__init__.py"))
cuda_version = str(nvcc_cuda_version)
if cuda_version != MAIN_CUDA_VERSION:
cuda_version_str = cuda_version.replace(".", "")[:3]
version += f"+cu{cuda_version_str}"
return version
def read_readme() -> str: def read_readme() -> str:
"""Read the README file if present.""" """Read the README file if present."""
p = get_path("README.md") p = get_path("README.md")
@ -256,7 +267,7 @@ def get_requirements() -> List[str]:
setuptools.setup( setuptools.setup(
name="vllm", name="vllm",
version=find_version(get_path("vllm", "__init__.py")), version=get_vllm_version(),
author="vLLM Team", author="vLLM Team",
license="Apache 2.0", license="Apache 2.0",
description=("A high-throughput and memory-efficient inference and " description=("A high-throughput and memory-efficient inference and "