Revert "[server_args] compute mem_fraction_static after dp chunked-prefill division" (#28991)

2026-06-22 20:48:16 -07:00
parent fc0d8fa262
commit 4740f23e1f
1 changed files with 1 additions and 4 deletions
@@ -3266,12 +3266,9 @@ class ServerArgs:
        if self.mem_fraction_static is None:
            # Constant meta data (e.g., from attention backend)
            reserved_mem = 512
-            effective_chunked_prefill_size = self.chunked_prefill_size // (
-                self.dp_size if self.enable_dp_attention and self.dp_size > 1 else 1
-            )
            # For activation during large prefill
            if self.chunked_prefill_size > 0:
-                reserved_mem += max(effective_chunked_prefill_size, 2048) * 1.5
+                reserved_mem += max(self.chunked_prefill_size, 2048) * 1.5
            else:
                reserved_mem += max(self.max_prefill_tokens, 2048) * 1.5
            # For cuda graphs