Overlap shared experts with deepep dispatch for single batch overlap on Blackwell (#17289)

This commit is contained in:
Baizhou Zhang
2026-01-21 02:56:55 +08:00
committed by GitHub
parent 16802fb6b2
commit 6ea491e439
4 changed files with 53 additions and 1 deletions
+1
View File
@@ -341,6 +341,7 @@ class Envs:
SGLANG_DEEPEP_BF16_DISPATCH = EnvBool(False)
SGLANG_DEEPEP_NUM_MAX_DISPATCH_TOKENS_PER_RANK = EnvInt(128)
SGLANG_DEEPEP_LL_COMBINE_SEND_NUM_SMS = EnvInt(32)
SGLANG_BLACKWELL_OVERLAP_SHARED_EXPERTS_OUTSIDE_SBO = EnvBool(False)
# NSA Backend
SGLANG_NSA_FUSE_TOPK = EnvBool(True)