diff --git a/paddle/phi/kernels/reduce_all_kernel.cc b/paddle/phi/kernels/reduce_all_kernel.cc
index 9b4515ee2909f73f4c8a21463ca5d17effc7e281..5b8d2cbecca5f5c3b85de2f9bac96bc51a0e319b 100644
--- a/paddle/phi/kernels/reduce_all_kernel.cc
+++ b/paddle/phi/kernels/reduce_all_kernel.cc
@@ -26,6 +26,9 @@ void AllKernel(const Context& dev_ctx,
                bool keep_dim,
                DenseTensor* out) {
   bool reduce_all = false;
+  if (dims.size() == 0 || static_cast<int>(dims.size()) == x.dims().size()) {
+    reduce_all = true;
+  }
   AllRawKernel<T>(dev_ctx, x, dims, keep_dim, reduce_all, out);
 }
 
diff --git a/python/paddle/fluid/layers/nn.py b/python/paddle/fluid/layers/nn.py
index 53994eed80f107b0a874e695bb67e9b6d13db1e9..b5615ee1bb7423c18143e441a3de825e249e38a3 100755
--- a/python/paddle/fluid/layers/nn.py
+++ b/python/paddle/fluid/layers/nn.py
@@ -5032,6 +5032,10 @@ def reduce_all(input, dim=None, keep_dim=False, name=None):
     """
     if dim is not None and not isinstance(dim, list):
         dim = [dim]
+
+    if in_dygraph_mode():
+        return _C_ops.all(input, dim if dim != None else [], keep_dim)
+
     check_variable_and_dtype(input, 'input', ('bool'), 'reduce_all')
     helper = LayerHelper('reduce_all', **locals())
     out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())