From f1115cf58db5124d16ea4e4d948e7f93fca584dd Mon Sep 17 00:00:00 2001 From: Baizhou Zhang Date: Sun, 30 Nov 2025 13:49:46 -0700 Subject: [PATCH] Revert "[Minor]Raise Error when deepep num dispatch token per rank is smaller than cuda graph bs" (#14171) --- .../sglang/srt/layers/moe/token_dispatcher/deepep.py | 10 ---------- 1 file changed, 10 deletions(-) diff --git a/python/sglang/srt/layers/moe/token_dispatcher/deepep.py b/python/sglang/srt/layers/moe/token_dispatcher/deepep.py index d36e68087..17ca310ab 100644 --- a/python/sglang/srt/layers/moe/token_dispatcher/deepep.py +++ b/python/sglang/srt/layers/moe/token_dispatcher/deepep.py @@ -23,7 +23,6 @@ from sglang.srt.layers.moe.utils import ( get_moe_runner_backend, is_tbo_enabled, ) -from sglang.srt.server_args import get_global_server_args from sglang.srt.utils import ( get_bool_env_var, get_int_env_var, @@ -312,15 +311,6 @@ class _DeepEPDispatcherImplBase: self.num_max_dispatch_tokens_per_rank = get_int_env_var( "SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK", 128 ) - if ( - not get_global_server_args().disable_cuda_graph - and get_global_server_args().cuda_graph_max_bs - > self.num_max_dispatch_tokens_per_rank - ): - raise ValueError( - f"Maximum captured batch size {get_global_server_args().cuda_graph_max_bs} is larger than num_max_dispatch_tokens_per_rank for deepep {self.num_max_dispatch_tokens_per_rank}. Please set a smaller value for --cuda-graph-max-bs server argument, or increase SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK." - ) - # DeepEP internode_ll dispatch uses FINISHED_SUM_TAG=1024 # and the logic requires num-tokens-sent-from-one-rank-to-another-rank less than it assert self.num_max_dispatch_tokens_per_rank <= 1024