Features

KV Cache Offloading

• UMBPStoreConnector See Quantized ReplaySSM checkpoints for Mamba2 shared and separate checkpoint modes.

Compatibility Matrix

The tables below show mutually exclusive features and the support on some hardware.

The symbols used have the following meanings:

• ✅ = Full compatibility
• 🟠 = Partial compatibility
• ❌ = No compatibility
• ❔ = Unknown or TBD

Note

Check the ❌ or 🟠 with links to see tracking issue for unsupported feature/hardware combination.

Feature x Feature

Feature	CP	APC	LoRA	SD	CUDA graph	pooling	enc-dec	logP	prmpt logP	async output	multi-step	mm	best-of	beam-search	prompt-embeds
CP	✅
APC	✅	✅
LoRA	✅	✅	✅
SD	✅	✅	❌	✅
CUDA graph	✅	✅	✅	✅	✅
pooling	🟠*	🟠*	✅	❌	✅	✅
enc-dec	❌	❌	❌	❌	✅	✅	✅
logP	✅	✅	✅	✅	✅	❌	✅	✅
prmpt logP	✅	✅	✅	✅	✅	❌	✅	✅	✅
async output	✅	✅	✅	❌	✅	❌	❌	✅	✅	✅
multi-step	❌	✅	❌	❌	✅	❌	❌	✅	✅	✅	✅
mm	✅	✅	🟠^	❔	✅	✅	✅	✅	✅	✅	❔	✅
best-of	✅	✅	✅	❌	✅	❌	✅	✅	✅	❔	❌	✅	✅
beam-search	✅	✅	✅	❌	✅	❌	✅	✅	✅	❔	❌	❔	✅	✅
prompt-embeds	✅	✅	✅	❌	✅	❌	❌	✅	❌	❔	❔	✅	❔	❔	✅

* Chunked prefill and prefix caching are only applicable to last-token or all pooling with causal attention.
^{^} LoRA is only applicable to the language backbone of multimodal models.

Feature x Hardware

Feature	Volta	Turing	Ampere	Ada	Hopper	CPU	AMD	Intel GPU
CP	❌	✅	✅	✅	✅	✅	✅	✅
APC	❌	✅	✅	✅	✅	✅	✅	✅
LoRA	✅	✅	✅	✅	✅	✅	✅	✅
SD	✅	✅	✅	✅	✅	✅	✅	✅
CUDA graph	✅	✅	✅	✅	✅	❌	✅	❌
pooling	✅	✅	✅	✅	✅	✅	✅	✅
enc-dec	✅	✅	✅	✅	✅	✅	❌	✅
mm	✅	✅	✅	✅	✅	✅	✅	✅
prompt-embeds	✅	✅	✅	✅	✅	✅	❔	✅
logP	✅	✅	✅	✅	✅	✅	✅	✅
prmpt logP	✅	✅	✅	✅	✅	✅	✅	✅
async output	✅	✅	✅	✅	✅	❌	❌	✅
multi-step	✅	✅	✅	✅	✅	❌	✅	✅
best-of	✅	✅	✅	✅	✅	✅	✅	✅
beam-search	✅	✅	✅	✅	✅	✅	✅	✅

Note

For information on feature support on Google TPU, please refer to the TPU-Inference Feature Matrix documentation.