Fix GPU OOM (#6564)

Co-authored-by: michael <michael.zhang@amd.com>

python/sglang/srt/layers/attention/aiter_backend.py

@@ -506,6 +506,7 @@ class AiterIndicesUpdaterPrefill:
                 spec_info.generate_attn_arg_prefill(
                     req_pool_indices,
                     paged_kernel_lens,
+                    None,
                     self.req_to_token,
                 )
             )

python/sglang/srt/model_executor/model_runner.py

@@ -412,6 +412,10 @@ class ModelRunner:
         if not server_args.disable_chunked_prefix_cache:
             logger.info("Chunked prefix cache is turned on.")
 
+        if server_args.attention_backend == "aiter":
+            if self.model_config.context_len > 8192:
+                self.mem_fraction_static *= 0.85
+
     def init_torch_distributed(self):
         logger.info("Init torch distributed begin.")