Rapid-H3 renders MiniMax-H3 video+audio 3-5x faster on one DGX Spark
vchauhan_ · reddit · 2026-10-30
Rapid-H3 is a ComfyUI node pack that renders MiniMax-H3 (text/image → video with stereo audio) 3-5x faster on a single NVIDIA DGX Spark (GB10) than stock ComfyUI, with no human-noticeable loss in picture, motion, voice or lip sync.
- tsh2 default preset: 117s for a 768×1344, 141-frame (5.9s) first-frame clip with audio
- tss6 preset (camera moves, close-up faces): 201s, matching stock on all scored criteria
- Stock ComfyUI bf16 at 10-step reference: 590-630s
- First render on a fresh install is slower while CUDA kernels compile (one-time)
More from Infra
- Ollama installed ≠ model downloaded: pitfalls when running it on cloud GPUs — NoConflict2854 · 2026-10-30
- NVIDIA launches CUDA Rust: two tracks for writing GPU kernels in Rust — AxSaucedo · 2026-10-30
- KAIST's LOWN Cuts Diffusion VLM Decoding FLOPs to 35% While Keeping 99.6% Accuracy — kaist-ai · 2026-10-30
- ByteDance Seed: Effective Learning Rate Trajectories, Not Normalization, Drive WN Training Gains — ByteDance-Seed · 2026-10-30
- Dell workstation with 96GB NVIDIA GPU may cost less than the card alone — chemist_slime · 2026-10-30
- Midwit meme on Databricks vs Snowflake: just put it in S3 — cto_junior · 2026-10-30