EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Support (1 <= dp < tp) in the dp attention in DeepEP (#4770)

Browse Source 
Co-authored-by: Cheng Wan <cwan39@gatech.edu>
This commit is contained in:
tarinkk
2025-03-27 20:09:35 -04:00
committed by GitHub
parent 98a2cfa9b2
commit 7f19e083c1
10 changed files with 238 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
python/sglang/srt/server_args.py
View File

@@ -290,12 +290,17 @@ class ServerArgs:
logger.warning(
f"DP attention is enabled. The chunked prefill size is adjusted to {self.chunked_prefill_size} to avoid MoE kernel issues. "
)
# DeepEP MoE
if self.enable_deepep_moe:
self.ep_size = self.dp_size
logger.info(
f"DeepEP MoE is enabled. The expert parallel size is adjusted to be the same as the data parallel size[{self.dp_size}]."
)
self.enable_sp_layernorm = False
# DeepEP MoE
if self.enable_deepep_moe:
self.ep_size = self.tp_size
self.enable_sp_layernorm = (
self.dp_size < self.tp_size if self.enable_dp_attention else True
)
logger.info(
f"DeepEP MoE is enabled. The expert parallel size is adjusted to be the same as the tensor parallel size[{self.tp_size}]."
)
# Speculative Decoding
if self.speculative_algorithm == "NEXTN":