🔌
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 XVERSE-MoE-A4.2B?
Yes — here's how fast and up to what context, with real measured numbers.
XVERSE-MoE-A4.2B on the Instinct MI430X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 5913k | 1747.5-2135.8 t/s | ~3-3.7 (slow) |
| Q8 | yes | 5977k | 3495-4271.7 t/s | ~6-7.3 (slow) |
| Q4 | yes | 6006k | 6241.1-7628 t/s | ~10.7-13.1 (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 XVERSE-MoE-A4.2B?▶
Yes. At Q4 it generates ~6241.1-7628 tok/s and fits up to 6006k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~10.7-13.1 tok/s.
How much context fits?▶
At Q4, up to about 6006k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.