[Qwen3-next] Add PD disaggregation support for mamba with extra_buffer (#15180)

Signed-off-by: Shangming Cai <csmthu@gmail.com> Co-authored-by: ybyang <10629930+whybeyoung@users.noreply.github.com>
2025-12-16 14:36:00 +08:00
parent 6292d97135
commit 36fcf71fff
5 changed files with 79 additions and 7 deletions
@@ -1382,9 +1382,6 @@ class ServerArgs:
                assert (
                    is_cuda()
                ), "Mamba extra_buffer is only supported on CUDA devices with FLA backend"
-                assert (
-                    self.disaggregation_mode == "null"
-                ), "Mamba extra_buffer is not compatible with disaggregation mode yet."
                if self.speculative_num_draft_tokens is not None:
                    assert (
                        self.mamba_track_interval >= self.speculative_num_draft_tokens