diff --git a/profiler/compare_tools/compare_backend/generator/detail_performance_generator.py b/profiler/compare_tools/compare_backend/generator/detail_performance_generator.py
index 4ae6ebcec26749bc4e1dfa6c0a28acb550b44154..677b0d18c2cea7e72ccdc78b2a5d2b750276e0ed 100644
--- a/profiler/compare_tools/compare_backend/generator/detail_performance_generator.py
+++ b/profiler/compare_tools/compare_backend/generator/detail_performance_generator.py
@@ -108,6 +108,8 @@ class DetailPerformanceGenerator(BaseGenerator):
             self._profiling_data_dict.get(Constant.COMPARISON_DATA)).build_module_tree()
         for index, base_node in enumerate(base_root_node):
             comparison_node = comparison_root_node[index] if index < len(comparison_root_node) else None
+            if not base_node or not comparison_node:
+                continue
             module_compare_result.extend(self._matching_all_modules(base_node, comparison_node))
         return module_compare_result
 
diff --git a/profiler/compare_tools/compare_backend/profiling_parser/gpu_profiling_parser.py b/profiler/compare_tools/compare_backend/profiling_parser/gpu_profiling_parser.py
index 77785d0261f6b6cbc15dfd55cbee3110037ac287..923854bdf73ad4d60f8b48b7eb3c83d526113b82 100644
--- a/profiler/compare_tools/compare_backend/profiling_parser/gpu_profiling_parser.py
+++ b/profiler/compare_tools/compare_backend/profiling_parser/gpu_profiling_parser.py
@@ -9,7 +9,7 @@ from compare_backend.utils.constant import Constant
 
 class GPUProfilingParser(BaseProfilingParser):
     CUBE_MARK = 'gemm'
-    FA_MARK_LIST = [['fmha', 'kernel'], ['flash', 'kernel']]
+    FA_MARK_LIST = [['fmha', 'kernel'], ['flash', 'kernel'], ['attention', 'kernel']]
     SDMA_MARK_LIST = ['htod', 'dtod', 'dtoh', 'memset (device)']
     FLOW_CAT = ("async_gpu", "async_cpu_to_gpu", "ac2g", "async")
     TORCH_OP_CAT = ("cpu_op", "user_annotation", "cuda_runtime", "operator")
@@ -138,7 +138,7 @@ class GPUProfilingParser(BaseProfilingParser):
             func_set.add(self._picking_kernel_event)
         if self._enable_operator_compare:
             func_set.add(self._picking_python_function_event)
-            func_set .add(self._picking_fwdbwd_flow_event)
+            func_set.add(self._picking_fwdbwd_flow_event)
         if self._enable_operator_compare or self._args.max_kernel_num:
             func_set.add(self._picking_kernel_event)
             func_set.add(self._picking_flow_event)