Nunchux speeds video diffusion with value smoothing and FP8 softmax casting
Nunchux AI's Muyang Li and collaborators report up to 1.70x faster end-to-end generation without retraining, with MiniMax H3 among four models tested.
By RuntimeWire Staff · Published · Updated
Primary source: MiniMax
Why it matters
Video generation costs increasingly depend on inference engineering beneath the model. VC-Attention shows how a young startup can turn academic kernel work into a credible platform wedge.
Muyang Li (@lmxyy1999) and his collaborators at Nunchux AI have published a low-bit attention method that attacks two stubborn costs inside video diffusion models: quantizing value activations and running softmax at higher precision. The September 14th VC-Attention paper reports attention-kernel speedups of 1.46x to 1.59x on NVIDIA's H200 and B200 data-center GPUs, with larger gains on workstation cards.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.