Tensor Fusion is a AI infra solution focusing on maximizing GPU usage with pooling and intelligent scheduling. It's based on a cutting-edge API-remoting GPU virtualization.
Less GPUs, More AI Apps.
Table of Contents
::warn:: This project is currently in heavy development.
Onboard your GPU cluster to Tensor Fusion in minutes.
- NVIDIA GPU pooling
- Pytorch support
- Seamless CUDA Context migration
- QoS levels
- Autoscaling policies
- TensorFlow and other framework support
- Compression, multiplexing & RDMA
- Support multiple GPU/NPU vendors based on ZLUDA
- Metrics & tracing of CUDA calls
- Advanced scheduling, super burst mode
- Hybrid scheduling, auto move communication intensive models schedule to local GPU
- Management dashboard
- Run on VM/BareMetal
- Run on Kubernetes clusters
- Run on ARM chips
See the open issues for a full list of proposed features (and known issues).
Contributions are what make the open source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated.
If you have a suggestion that would make this better, please fork the repo and create a pull request. You can also simply open an issue with the tag "enhancement". Don't forget to give the project a star! Thanks again!
- Fork the Project
- Create your Feature Branch (
git checkout -b feature/AmazingFeature) - Commit your Changes (
git commit -m 'Add some AmazingFeature') - Push to the Branch (
git push origin feature/AmazingFeature) - Open a Pull Request
TensorFusion is not open sourced as of now. It will be open after GA with BSL license.
- Discord channel: https://discord.gg/2bybv9yQNk
- Email address: @[email protected]
- Project link: https://github.com/NexsusGPU/TensorFusion