🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the NVIDIA H100 SXM5 80GB run Qwen-Audio-Chat?
Yes — here's how fast and up to what context, with real measured numbers.
Qwen-Audio-Chat on the NVIDIA H100 SXM5 80GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 421k | 144.5-172.3 t/s | ~1.7-2.1 (slow) |
| Q8 | yes | 486k | 289.1-344.5 t/s | ~3.5-4.1 (slow) |
| Q4 | yes | 514k | 516.2-615.2 t/s | ~6.2-7.3 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to system RAM offload (depends on your system RAM) and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA H100 SXM5 80GB run Qwen-Audio-Chat?▶
Yes. At Q4 it generates ~516.2-615.2 tok/s and fits up to 514k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~6.2-7.3 tok/s.
How much context fits?▶
At Q4, up to about 514k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.