diff --git a/python/paddle/fluid/transpiler/memory_optimization_transpiler.py b/python/paddle/fluid/transpiler/memory_optimization_transpiler.py
index 52c1aea288fa2bb7478ad14186367900c05f64e7..31d8e44cececd39cf88c6149fd89d1d1d0e82c1d 100755
--- a/python/paddle/fluid/transpiler/memory_optimization_transpiler.py
+++ b/python/paddle/fluid/transpiler/memory_optimization_transpiler.py
@@ -355,6 +355,10 @@ class ControlFlowGraph(object):
                                                  is_forward).dtype()
                         cache_dtype = self._find_var(block_desc, cache_var,
                                                      is_forward).dtype()
+                        if x_dtype != cache_dtype:
+                            if PRINT_LOG:
+                                print("x_dtype and cache_dtype are different!")
+                            continue
 
                         if not compare_shape(x_shape, cache_shape, level):
                             continue