xdit-project/xDiT
A scalable inference engine for running diffusion transformer models (DiTs) with parallel execution across multiple GPUs and machines.
Not currently ranked — collecting fresh signals.
star history
xDiT is a distributed inference system designed for Diffusion Transformers, enabling high-throughput image and video generation by leveraging parallelism strategies including PipeFusion, sequence parallel, and hybrid parallel methods. It addresses the quadratic computational complexity of attention mechanisms in DiTs by distributing workloads across multiple GPUs and machines to meet real-time inference requirements. The project integrates with ComfyUI and provides compilation and cache acceleration for single GPU execution.
Frequently asked
- What is xdit-project/xDiT?
- A scalable inference engine for running diffusion transformer models (DiTs) with parallel execution across multiple GPUs and machines.
- Is xDiT open source?
- Yes — xdit-project/xDiT is open source, released under the Apache-2.0 license.
- What language is xDiT written in?
- xdit-project/xDiT is primarily written in Python.
- How popular is xDiT?
- xdit-project/xDiT has 2.7k stars on GitHub.
- Where can I find xDiT?
- xdit-project/xDiT is on GitHub at https://github.com/xdit-project/xDiT.