# Wave Forcing: Towards Speed-of-Light Streaming Video Generation

DevFeed: [Wave Forcing: Towards Speed-of-Light Streaming Video Generation](<https://devfeed.tech/articles/wave-forcing-towards-speed-of-light-streaming-video-generation-113782.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.00859>)

Author: Yijie Jin, Yuxuan Liu, Jiajun Xu, Ye Huang, Chuzhi Wang, Jiayang Li, Ruiyang Chen, Chenling Hong, Xu Wang, Zhijie Deng

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Streaming](<https://devfeed.tech/topics/streaming.md>), [GPU optimization](<https://devfeed.tech/topics/gpu-optimization.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [attention](<https://devfeed.tech/tags/attention.md>), [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [cache](<https://devfeed.tech/tags/cache.md>), [camera](<https://devfeed.tech/tags/camera.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [denoising](<https://devfeed.tech/tags/denoising.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [execution](<https://devfeed.tech/tags/execution.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [faster](<https://devfeed.tech/tags/faster.md>), [fps](<https://devfeed.tech/tags/fps.md>), [generation](<https://devfeed.tech/tags/generation.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [kernels](<https://devfeed.tech/tags/kernels.md>), [motion](<https://devfeed.tech/tags/motion.md>), [streaming](<https://devfeed.tech/tags/streaming.md>), [streaming-video](<https://devfeed.tech/tags/streaming-video.md>), [video](<https://devfeed.tech/tags/video.md>), [video-generation](<https://devfeed.tech/tags/video-generation.md>)

## AI overview

Wave Forcing (WF) pipelines continuous video generation with multi-step denoising: chunk-wise causal attention can use earlier, cleaner chunks before they finish denoising, while staggered denoising and cache refresh across full-model replicas yield one output chunk per concurrent forward round after pipeline fill. Its three-stage training adds mixed-noise denoising, motion supervision, then causal adaptation using self-generated history. Wave Runtime combines early K/V delivery and communication-computation overlap with optimized kernels and pipelined VAE decoding. On Wan2.1-T2V-1.3B, 5-step WF had overall quality comparable to Rolling Forcing and higher Dynamic Degree on short and long videos. On eight H200 GPUs, 4-step WF reached 134.8 end-to-end FPS, reported as 7.07x single-GPU end-to-end throughput. In matched DiT-only tests across six model scales, Wave Parallelism reached up to 2.53x the throughput of the faster Ulysses or Ring baseline.

## Source excerpt

arXiv:2611.00859v1 Announce Type: cross Abstract: Real-time long-video generation requires continuous output with consistent appearance and natural motion. Existing chunk-wise autoregressive methods face challenges in both generation efficiency and long-video quality: serialized denoising and history-cache refresh require two forward passes even with one-step denoising, and longer rollouts can degrade quality and cause camera drift. We propose Wave Forcing (WF), which co-designs generation, training, and execution for continuous video output with multi-step denoising. WF uses chunk-wise causal attention to access earlier, cleaner chunks before their denoising is complete, and distributes staggered denoising and cache-refresh stages across full-model replicas to achieve 1F1C: one output chunk per concurrent forward round after pipeline fill. To adapt the model to this continuous generation process, we develop a three-stage training procedure that successively introduces mixed-noise denoising, motion supervision, and causal adaptation using self-generated history. To execute this process efficiently, Wave Runtime (WaveRT) uses Wave Parallelism (WP) for early K/V delivery and communication-computation overlap, together with optimized kernels and pipelined VAE decoding. Experiments on Wan2.1-T2V-1.3B show that 5-step WF maintains overall quality comparable to Rolling Forcing while achieving higher Dynamic Degree on both short and long videos. Our system efficiency experiments show that 4-step WF achieves 147.6 DiT FPS and 134.8 end-to-end FPS on eight H200 GPUs, yielding 7.07x the single-GPU end-to-end throughput. In DiT-only tests with matched SageAttention+W8A8 backends, denoising steps, and GPU counts, WP delivers up to 2.53x the throughput of the faster of the Ulysses and Ring context-parallel baselines across six model scales (1.3B-14B).