Speed Any mode 6 steps
Parasyte Turbo 6 NFE (arena i2v #1)
Top bracket of the H3 Acceleration Arena (top-10 in both slices, highest vote count); 6 NFE ≈ 4.1× timed speedup — the current sweet spot.
How to run it
- 1 Download the LoRA and load it per the repo docs (check the license)
- 2 Run 6–7 steps; validate quality on a short clip first
- 3 Stacks orthogonally with Sage/Kitchen attention
We haven't run this LoRA ourselves — read the repo docs and license first.
Representative GPUs
864 × 480 · 8s · 32GB RAM / fast-disk; click a GPU for its full matrix.
| GPU | VRAM | Status | Per clip | VRAM peak |
|---|---|---|---|---|
| RTX 4060 8GB | 8GB | Slow | 6 min 19 s | 7G |
| RTX 3060 12GB | 12GB | OK | 6 min | 11G |
| RTX 4070 12GB | 12GB | OK | 3 min 30 s | 11G |
| RTX 4070 Ti Super 16GB | 16GB | OK | 2 min 34 s | 15G |
| RTX 4080 Super 16GB | 16GB | OK | 2 min 18 s | 15G |
| RTX 4090 24GB | 24GB | OK | 1 min 40 s | 23G |
| RTX 5090 32GB | 32GB | OK | 1 min 14 s | 26G |
| RTX PRO 6000 Blackwell 96GB | 96GB | OK | 1 min 9 s | 47G |
1344×768 · 5s: 48 of 49 listed GPUs workable, from about 6GB VRAM.
FAQ
What's the minimum GPU for this recipe?
About 6GB+ of VRAM for 1344×768 · 5s (48 of 49 listed GPUs workable). Short 736×416 clips run on less.
How does this recipe combine with other acceleration?
Step distillation picks one option; attention backends (Sage/Kitchen/SLA/VSA) and caches stack on top. See the conflicts table for exact rules.