diff --git a/include/ck_tile/ops/transpose/block_transpose.hpp b/include/ck_tile/ops/transpose/block_transpose.hpp
index e19fda8ede..fca17a6135 100644
--- a/include/ck_tile/ops/transpose/block_transpose.hpp
+++ b/include/ck_tile/ops/transpose/block_transpose.hpp
@@ -149,25 +149,40 @@ struct BlockTranspose
 
         auto y = load_tile_transpose(load_from_lds_window);
 
-        constexpr auto lds_load_distr = Policy::template MakeLdsLoadTileDistribution<Problem>();
-        // constexpr auto glb_store_distr = Policy::template MakeOutputDistribution<Problem>();
-
-        constexpr auto y_in_desc = lds_load_distr.get_ys_to_d_descriptor();
-        // constexpr auto y_out_desc = glb_store_distr{}.get_ys_to_d_descriptor();
-
-        constexpr index_t NDimY = lds_load_distr.get_num_of_dimension_y();
-
-        constexpr auto y_in_lengths   = to_sequence(y_in_desc.get_lengths());
-        constexpr index_t vecLoadSize = y_in_lengths[NDimY - 1];
-        // constexpr auto lds_distr_y_indx_zeros = uniform_sequence_gen_t<decltype(Policy::template
-        // MakeLdsLoadTileDistribution<Problem>())::NDimY, 0>{};
         auto out_tensor = make_static_distributed_tensor<DataType>(
             Policy::template MakeOutputDistribution<Problem>());
-        using InVec  = array<DataType, vecLoadSize>;
-        using OutVec = array<DataType, vecLoadSize>;
 
-        out_tensor.get_thread_buffer().template set_as<OutVec>(
-            number<0>{}, y.get_thread_buffer().template get_as<InVec>(number<0>{}));
+        constexpr auto lds_load_distr  = Policy::template MakeLdsLoadTileDistribution<Problem>();
+        constexpr auto glb_store_distr = Policy::template MakeOutputDistribution<Problem>();
+
+        constexpr auto y_in_desc  = lds_load_distr.get_ys_to_d_descriptor();
+        constexpr auto y_out_desc = glb_store_distr.get_ys_to_d_descriptor();
+
+        constexpr index_t NDimYIn  = lds_load_distr.get_num_of_dimension_y();
+        constexpr index_t NDimYOut = glb_store_distr.get_num_of_dimension_y();
+
+        constexpr auto y_in_lengths  = to_sequence(y_in_desc.get_lengths());
+        constexpr auto y_out_lengths = to_sequence(y_out_desc.get_lengths());
+
+        constexpr auto y_in_element_space_size  = y_in_desc.get_element_space_size();
+        constexpr auto y_out_element_space_size = y_out_desc.get_element_space_size();
+        static_assert(y_in_element_space_size == y_out_element_space_size,
+                      "the element space size is not the same!");
+        static_assert(y_in_lengths[NDimYIn - 1] == y_out_lengths[NDimYOut - 1],
+                      "the vector length is not the same!");
+        constexpr index_t vecLoadSize = y_in_lengths[NDimYIn - 1];
+        constexpr index_t num_of_access =
+            reduce_on_sequence(y_in_lengths, multiplies{}, number<1>{}) / vecLoadSize;
+
+        // constexpr auto lds_distr_y_indx_zeros = uniform_sequence_gen_t<decltype(Policy::template
+        // MakeLdsLoadTileDistribution<Problem>())::NDimY, 0>{};
+
+        using DataVec = array<DataType, vecLoadSize>;
+        static_for<0, num_of_access, 1>{}([&](auto iAccess) {
+            out_tensor.get_thread_buffer().template set_as<DataVec>(
+                number<iAccess>{},
+                y.get_thread_buffer().template get_as<DataVec>(number<iAccess>{}));
+        });
 
         store_tile(output_tile_window, out_tensor);
     }