[Optimize] Support and robust for tpN for PD (#4595)

* [Optimize] Support and robust for tpN for PD

* fix

* fix

* support dpM tpN for cache messager

* fix

* fix token counter

* fix bug for merge develop

* fix bug

* robust cache messager for v0
This commit is contained in:
chenjian
2025-11-03 15:38:31 +08:00
committed by GitHub
parent 7b35488779
commit 25498efcf3
9 changed files with 452 additions and 197 deletions
@@ -275,6 +275,7 @@ class SplitwiseConnector:
decode_diagg = task.disaggregate_info["cache_info"]
task.disaggregate_info["cache_info"] = self.cfg.disaggregate_info["cache_info"]
task.disaggregate_info["cache_info"]["rdma"]["current_id"] = current_id
task.disaggregate_info["role"] = "decode"
self._send_message(addr, "prefill", [task])
task.disaggregate_info["cache_info"] = decode_diagg
task.disaggregate_info["role"] = "prefill"