Demo-131 Cuda graph with optimized paged attention#205

Merged

wooway777 merged 19 commits intomainfrom

Feb 13, 2026

Commits on Feb 10, 2026

issue/219: support vllm bench
Ceng23333
committed
issue/143 feat: static and paged graph compilers

PanZezhong1725
authored and
wooway777
committed
issue/143 init input to support warmup

PanZezhong1725
authored and
wooway777
committed
issue/143 add barrier for compilers

PanZezhong1725
authored and
wooway777
committed
issue/143 fix add compile after model init

PanZezhong1725
authored and
wooway777
committed
issue/143 use add_rmsnorm, nt flash attn, nt kv caching

PanZezhong1725
authored and
wooway777
committed
issue/204 - support graph in server scripts
wooway777
committed
issue/143 fix bench script, worker cleanup, compiler initial input

PanZezhong1725
authored and
wooway777
committed
issue/991 optimize input preparation

PanZezhong1725
authored and
wooway777
committed
issue/208 - adapt to ali ppu
wooway777
committed
issue/214 - update attn and caching logics
wooway777
committed

Commits on Feb 11, 2026