Merge branch 'main' into fix_hang

[Bugfix] Fix hang with DP+EP on B200
2025-10-20 14:53:52 +08:00 · 2025-09-27 10:44:26 -04:00 · 2025-09-26 19:11:45 -04:00 · 2025-09-26 15:25:58 -07:00
1 changed files with 1 additions and 1 deletions
--- a/vllm/v1/worker/gpu_worker.py
+++ b/vllm/v1/worker/gpu_worker.py
@ -488,7 +488,7 @@ class Worker(WorkerBase):
                    sort_by="self_cuda_time_total"))
    def execute_dummy_batch(self) -> None:
-        self.model_runner._dummy_run(1, uniform_decode=True)
+        self.model_runner._dummy_run(16, uniform_decode=True)
    def add_lora(self, lora_request: LoRARequest) -> bool:
        return self.model_runner.add_lora(lora_request)
Author	SHA1	Message	Date
Robert Shaw	562107efb1	Merge branch 'main' into fix_hang	2025-09-27 10:44:26 -04:00
Robert Shaw	2a548e8ef1	Merge branch 'main' into fix_hang	2025-09-26 19:11:45 -04:00
Alexander Matveev	a0256414e9	[Bugfix] Fix hang with DP+EP on B200 Signed-off-by: Alexander Matveev <amatveev@redhat.com>	2025-09-26 15:25:58 -07:00