20-asset 3D game kit

Concept art straight into 3D meshes, the pipeline that only runs on serious VRAM.

The pipeline

  1. Generate 20 concepts, stable-diffusion-xl
  2. Convert each into a 3D asset, trellis-image-to-3d

Fastest card where every stage is a real measurement: NVIDIA H100 80GB HBM3, 2.2 min for the whole job.

Every GPU, slowest job to fastest

GPUTotal timeComputeEnergyBasis
NVIDIA H100 80GB HBM32.2 min2 min12.69 Whall 2 stages measured
NVIDIA H2002.4 min2 min13.15 Whall 2 stages measured
NVIDIA L40S2.9 min2.7 min13.29 Whall 2 stages measured
NVIDIA A100 80GB SXM44 min3.7 min16.75 Whall 2 stages measured
NVIDIA A10G7.6 min7.2 min21.16 Whall 2 stages measured
NVIDIA L48.3 min8.1 min9.43 Whall 2 stages measured

Cards that can't run this pipeline

Every one of these fails on the same kind of wall, a stage that will not fit in VRAM.

How these numbers are built

Each stage time is the quantity of work divided by that card's measured throughput for that model, from our own bench. The pipeline is assumed to run batched, every image, then every clip, so each model loads once. Model load time is added where we recorded it; our text-generation runs don't carry a load measurement yet, so pipelines with a language-model stage are slightly optimistic. Nothing here is a single timed run of the whole pipeline, and we don't present it as one.