Skip to content

Can Tesla P40 24G run Ternary Bonsai 2 27B?

Comfortable: GGUF Q2_0 needs 8.2 GB at 4K, batch 1; headroom 15.8 GB.

Text input, FP16 KV cache, one card. Unified-memory capacity uses the GPU guide’s usable limit; negative headroom may require host offload. Speed source settings may differ from this 4K estimate.

Measured runs

No measured run for this pair. Sizing is an estimate, not a runtime test.

Quant memory at 4K

QuantTotal GBVerdictWeight sizing
GGUF Q2_08.2ComfortableVerified file

Context capacity · GGUF Q2_0

ContextTotal GBVerdict
4K8.2Comfortable
16K9.1Comfortable
32K10.2Comfortable
128K16.8Comfortable

Generation speed

GGUF Q2_0: 27–36 tok/s · Derived range

Other model / GPU pairs