Search papers, labs, and topics across Lattice.
Odin is a novel training system for distributed point-based neural rendering (PBNR) that enhances training efficiency by eliminating unnecessary synchronization barriers. By leveraging lightweight visibility metadata and a two-phase design, Odin synchronizes only when overlapping visible parameters are updated, significantly reducing communication overhead. The system achieves a remarkable 1.89脳 throughput speedup on 64 GPUs while maintaining baseline accuracy, demonstrating its effectiveness in optimizing distributed rendering workflows.
Achieving a 1.89脳 speedup in distributed neural rendering by cleverly minimizing synchronization barriers could redefine efficiency in large-scale training.
We present Odin, a training system for distributed point-based neural rendering (PBNR) that accelerates training by removing unnecessary global barriers. Odin observes that two training steps need synchronization only when they update overlapping visible parameters. It uses lightweight visibility metadata and a two-phase design with ahead-of-time scheduling plus runtime refinement to handle irregular and dynamic behavior. Odin hides 82% of exposed communication and achieves up to 1.89 脳 throughput speedup on 64 GPUs while preserving baseline accuracy.
Tsinghua AI