Support MHA with chunked prefix cache for flashinfer/flashmla backend, support page size > 1 for MHA chunked prefix (#8616)

Co-authored-by: xuyongfei.xyf <xuyongfei.xyf@antgroup.com>
2025-08-22 09:19:44 +08:00
parent 704ced1b2e
commit 9708d353b7
6 changed files with 184 additions and 75 deletions
--- a/python/sglang/srt/managers/schedule_batch.py
+++ b/python/sglang/srt/managers/schedule_batch.py
@@ -106,6 +106,7 @@ GLOBAL_SERVER_ARGS_KEYS = [
    "enable_symm_mem",
    "quantization",
    "enable_custom_logit_processor",
+    "disaggregation_mode",
 ]

 # Put some global args for easy access