🔌
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 Apple M2 Max 96GB run MoEFold2D?
Yes — here's how fast and up to what context, with real measured numbers.
MoEFold2D on the Apple M2 Max 96GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1295k | 141666.7-170833.3 t/s | ~3541.7-4270.8 (slow) |
| Q8 | yes | 1295k | 283333.3-341666.7 t/s | ~7083.3-8541.7 (slow) |
| Q4 | yes | 1295k | 505952.4-610119 t/s | ~12648.8-15253 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
FAQ
Can the Apple M2 Max 96GB run MoEFold2D?▶
Yes. At Q4 it generates ~505952.4-610119 tok/s and fits up to 1295k context; beyond that it spills to SSD swap and slows to ~12648.8-15253 tok/s.
How much context fits?▶
At Q4, up to about 1295k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.