🔌
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 MI430X run ERNIE-4.5-VL-28B-A3B?
Yes — here's how fast and up to what context, with real measured numbers.
ERNIE-4.5-VL-28B-A3B on the Instinct MI430X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1297k | 262.1-320.4 t/s | ~0.5-0.6 (slow) |
| Q8 | yes | 1404k | 524.3-640.8 t/s | ~0.9-1.1 (slow) |
| Q4 | yes | 1451k | 936.2-1144.2 t/s | ~1.6-2 (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 MI430X run ERNIE-4.5-VL-28B-A3B?▶
Yes. At Q4 it generates ~936.2-1144.2 tok/s and fits up to 1451k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~1.6-2 tok/s.
How much context fits?▶
At Q4, up to about 1451k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.