🔌
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 Instinct MI455X run Llama Nemotron Ultra 253B?
Yes — here's how fast and up to what context, with real measured numbers.
Llama Nemotron Ultra 253B on the Instinct MI455X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | no | — | — | — |
| Q8 | yes | 273k | 48.8-59.7 t/s | ~0.1-0.1 (slow) |
| Q4 | yes | 485k | 87.2-106.5 t/s | ~0.2-0.2 (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 Instinct MI455X run Llama Nemotron Ultra 253B?▶
Yes. At Q4 it generates ~87.2-106.5 tok/s and fits up to 485k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~0.2-0.2 tok/s.
How much context fits?▶
At Q4, up to about 485k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.