Ascend attention backend(PA&MLA) (#7722)

Co-authored-by: Maksim <makcum888e@mail.ru>
Co-authored-by: VDV1985 <vladdv85@mail.ru>

python/sglang/srt/server_args.py

@@ -380,6 +380,12 @@ class ServerArgs:
             )
             self.disable_cuda_graph = True
 
+        if self.attention_backend == "ascend":
+            logger.warning(
+                "At this moment Ascend attention backend only supports a page_size of 128, change page_size to 128."
+            )
+            self.page_size = 128
+
         # Choose grammar backend
         if self.grammar_backend is None:
             self.grammar_backend = "xgrammar"
@@ -1113,6 +1119,7 @@ class ServerArgs:
                 "flashmla",
                 "intel_amx",
                 "torch_native",
+                "ascend",
                 "triton",
             ],
             default=ServerArgs.attention_backend,