New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the Intel Gaudi 3 run Adaptive Inputs + LayerDrop?
Yes — here's how fast and up to what context, with real measured numbers.
Adaptive Inputs + LayerDrop on the Intel Gaudi 3
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1761k | 2536.6-3148.9 t/s | ~27.4-34 (slow) |
| Q8 | yes | 1768k | 5073.3-6297.9 t/s | ~54.8-68.1 (slow) |
| Q4 | yes | 1770k | 9059.4-11246.2 t/s | ~97.9-121.6 (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 Intel Gaudi 3 run Adaptive Inputs + LayerDrop?▶
Yes. At Q4 it generates ~9059.4-11246.2 tok/s and fits up to 1770k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~97.9-121.6 tok/s.
How much context fits?▶
At Q4, up to about 1770k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.