[TRTLLM-11875][feat] Support fine-grained context chunk management#16427

Open

VALLIS-NERIA wants to merge 12 commits into

NVIDIA:mainfrom

VALLIS-NERIA:agent/mamba-scheduling-disagg-lifecycle

Commits on Jul 16, 2026

[None][fix] Harden scheduling and disagg lifecycle (By Agent)
VALLIS-NERIA
committed
[None][chore] Apply resource manager test formatting (By Agent)
VALLIS-NERIA
committed
[None][fix] Address scheduling and transfer review (By Agent)
VALLIS-NERIA
committed
[None][fix] Align hybrid Mamba C++ disagg routing (By Agent)
VALLIS-NERIA
committed
[None][test] Wait for hybrid transfer cancellation (By Agent)
VALLIS-NERIA
committed
[None][test] Cover mixed Mamba Python transfer (By Agent)
VALLIS-NERIA
committed
[None][test] Cover hybrid runtime auto fallback (By Agent)
VALLIS-NERIA
committed
[None][fix] Address PR1 review feedback (By Agent)
VALLIS-NERIA
committed
[None][fix] Address CodeRabbit review feedback (By Agent)
VALLIS-NERIA
committed
[None][fix] Use request snapshot points for Mamba chunking (By Agent)
VALLIS-NERIA
committed
[None][refactor] Consolidate Mamba snapshot point preparation (By Agent)
VALLIS-NERIA
committed

Commits on Jul 17, 2026

Merge branch 'main' into agent/mamba-scheduling-disagg-lifecycle
VALLIS-NERIA
authored