GitHub - SystemPanic/vllm-windows: A high-throughput and memory-efficient inference and serving engine for LLMs (Windows build & kernels) (original) (raw)

vLLM

Easy, fast, and cheap LLM serving for everyone

| Documentation | Blog | Paper | Twitter/X | User Forum | Developer Slack |

vLLM for Windows

vLLM for Windows build & kernels. This repository will be updated when new versions of vLLM are released.

Don't open a new Issue to request a specific commit build. Wait for a new stable release.

Don't open Issues for general vLLM questions or non Windows related problems. Only Windows specific issues. Any Issue opened that is not Windows specific will be closed automatically.

Don't request a wheel for your specific environment. Currently, the only wheels I will publish are for Python 3.12 + CUDA 12.4 + torch 2.6.0. If you have another versions, build your own wheel from source by following the instructions below.

Windows instructions:

Installing an existing release wheel:

  1. Ensure that you have the correct Python and CUDA version of the wheel. The Python and CUDA version of the wheel is specified in the release version
  2. Download the wheel from the release version of your preference
  3. Install it with pip install DOWNLOADED_WHEEL_PATH

Building from source:

A Visual Studio 2019 or newer is required to launch the compiler x64 environment. The installation path is referred in the instructions as VISUAL_STUDIO_INSTALL_PATH.

CUDA path will be found automatically if you have the bin folder in your PATH, or have the CUDA installation path settled on well-known environment vars like CUDA_ROOT, CUDA_HOME or CUDA_PATH.

If none of these are present, make sure to set the environment variable before starting the build: set CUDA_ROOT=CUDA_INSTALLATION_PATH

  1. Open a Command Line (cmd.exe)
  2. Clone the vLLM repository: cd C:\ & git clone https://github.com/SystemPanic/vllm-windows.git
  3. Execute (in cmd) VISUAL_STUDIO_INSTALL_PATH\VC\Auxiliary\Build\vcvarsall.bat x64
  4. Change the working directory to the cloned repository path, for example: cd C:\vllm-windows
  5. Set the following environment variables:
set DISTUTILS_USE_SDK=1
set VLLM_TARGET_DEVICE=cuda
#(replace 10 with your desired cpu threads to use in parallel to speed up compilation)
set MAX_JOBS=10

#Optional variables:

#To include cuDSS (only if you have cuDSS installed)
set USE_CUDSS=1
set CUDSS_LIBRARY_PATH=PATH_TO_CUDSS_INSTALL_DIR\lib\12
set CUDSS_INCLUDE_PATH=PATH_TO_CUDSS_INSTALL_DIR\include

#To include cuSPARSELt (only if you have cuSPARSELt installed)
set USE_CUSPARSELT=1
set CUSPARSELT_INCLUDE_PATH=PATH_TO_CUSPARSELT_INSTALL_DIR\include
set CUSPARSELT_LIBRARY_PATH=PATH_TO_CUSPARSELT_INSTALL_DIR\lib

#To include cuDNN:
set USE_CUDNN=1
set CUDNN_LIBRARY_PATH=PATH_TO_CUDNN_INSTALL_DIR\lib\CUDNN_CUDA_VERSION\x64
set CUDNN_INCLUDE_PATH=PATH_TO_CUDNN_INSTALL_DIR\include\CUDNN_CUDA_VERSION

#Flash Attention v3 build has been disabled inside WSL2 and Windows due to compiler being killed on WSL2, and extremely long compiling times on Windows. Hopper is not available on Windows, so FA3 has no sense anyway. 
#Build can be forcefully enabled using the following environment var:
set VLLM_FORCE_FA3_WINDOWS_BUILD=1
  1. Build & install:
#With torch 2.6.0 cuda 12.4 (change cu124 with your installed CUDA version)
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124

#With your already installed torch cuda version (make sure you have torch cuda installed if you use a virtual environment)
python use_existing_torch.py

pip install -r requirements/build.txt
pip install -r requirements/windows.txt
pip install . --no-build-isolation

Latest News 🔥

About

vLLM is a fast and easy-to-use library for LLM inference and serving.

Originally developed in the Sky Computing Lab at UC Berkeley, vLLM has evolved into a community-driven project with contributions from both academia and industry.

vLLM is fast with:

Performance benchmark: We include a performance benchmark at the end of our blog post. It compares the performance of vLLM against other LLM serving engines (TensorRT-LLM, SGLang and LMDeploy). The implementation is under nightly-benchmarks folder and you can reproduce this benchmark using our one-click runnable script.

vLLM is flexible and easy to use with:

vLLM seamlessly supports most popular open-source models on HuggingFace, including:

Find the full list of supported models here.

Getting Started

Install vLLM with pip or from source:

Visit our documentation to learn more.

Contributing

We welcome and value any contributions and collaborations. Please check out Contributing to vLLM for how to get involved.

Sponsors

vLLM is a community project. Our compute resources for development and testing are supported by the following organizations. Thank you for your support!

Cash Donations:

Compute Resources:

Slack Sponsor: Anyscale

We also have an official fundraising venue through OpenCollective. We plan to use the fund to support the development, maintenance, and adoption of vLLM.

Citation

If you use vLLM for your research, please cite our paper:

@inproceedings{kwon2023efficient, title={Efficient Memory Management for Large Language Model Serving with PagedAttention}, author={Woosuk Kwon and Zhuohan Li and Siyuan Zhuang and Ying Sheng and Lianmin Zheng and Cody Hao Yu and Joseph E. Gonzalez and Hao Zhang and Ion Stoica}, booktitle={Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles}, year={2023} }

Contact Us

Media Kit