[b200] support trt-llm allreduce fuse rms_norm_add kernel (#7621)

2025-07-03 10:36:20 +08:00
parent 82f021e22e
commit 8e64140e35
5 changed files with 253 additions and 2 deletions
--- a/python/sglang/srt/managers/schedule_batch.py
+++ b/python/sglang/srt/managers/schedule_batch.py
@@ -85,6 +85,7 @@ GLOBAL_SERVER_ARGS_KEYS = [
    "deepep_mode",
    "enable_ep_moe",
    "enable_flashinfer_moe",
+    "enable_flashinfer_allreduce_fusion",
    "moe_dense_tp_size",
    "ep_dispatch_algorithm",
    "deepep_config",