TL;DR: Video generation with DiTs is painfully slow – HunyuanVideo takes 16 minutes to generate just a 5-second video on an H100 with FlashAttention3. Our sliding tile attention (STA) slashes this to 5 minutes with zero quality loss, no extra training required. Specifically, STA accelerates attention alone by 2.8–17x over FlashAttention-2 and 1.6–10x over FlashAttention-3. With STA and other optimizations, our solution boosts end-to-end generation speed by 2.98× compared to the FA3 full attention baseline, without quality loss or the need for training. Enabling finetuning unlocks even greater speedups!
👉Try out kernel in our FastVideo project project and we’d love to hear what you think!