ProductBenchmarksDevelopers EnterprisePricingDownload ResearchFAQStatus GitHub Trust & Legal 한국어로 보기
Evidence record

VZK-BENCH-WAN22-S2V-14B-3090

Wan-AI/Wan2.2-S2V-14B · speech-to-video (Korean talking-head)

VEIZIK-MEASURED · SINGLE RUN 0 external reproductions 2026-07-22
Conditions

What was run

Condition · Value
GPURTX 3090 24GB
Power cap (W)300
Resolution832x480
Frames49
Steps30
Seed42
Precision profilefp8 offload, persist=13e9 resident
Audio encoderwav2vec2-korean-swap (Korean)
Measured

Results

Metric · Value
Wall time (s)884.9
Per-step time (s)28.9
Peak VRAM (GB)17.97
Peak host RAM (GB)46.5
Load time (s)11.1
Repeat count1
Render stabilitySTABLE (sharp_ratio 1.004, motion 1.906)
Notes & caveats

Full context

Wan2.2-S2V-14B renders a Korean audio-driven talking-head (832x480/49f/30 steps) on a single 24GB RTX 3090 in 884.9s at GPU1=300W: peak VRAM 17.97GB (accurate torch.cuda.max_memory_allocated), 46.5GB host RAM, quality STABLE. Korean wav2vec2-korean-swap encoder. First model with an accurate in-process VRAM number (nvidia-smi VRAM is unreliable on WSL2 GPU-PV). Fix applied: librosa.load->soundfile (numba/NumPy-2.2 incompat). Video shown in veizik dashboard gallery + [internal] drive.

Reproduce

Run it yourself

Run the same fixed configuration on your card and open a result PR — see Run it yourself. Raw run manifests are attached to the internal evidence record and are released alongside external reproduction tooling.

← Back to all benchmarks