diff --git a/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v3.hpp b/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v3.hpp
index d6ee099930..5823d81491 100644
--- a/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v3.hpp
+++ b/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v3.hpp
@@ -71,7 +71,7 @@ __launch_bounds__(CK_MAX_THREAD_PER_BLOCK, MinimumOccupancy)
     const long_index_t e_batch_offset =
         amd_wave_read_first_lane(compute_ptr_offset_of_batch.GetEPtrOffset(g_idx));
 
-    __shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte(get_device_arch())];
+    __shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte()];
 
     index_t left     = 0;
     index_t right    = gemms_count;
diff --git a/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_conv_v3.hpp b/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_conv_v3.hpp
index c92b562222..4afd5e3633 100644
--- a/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_conv_v3.hpp
+++ b/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_conv_v3.hpp
@@ -705,12 +705,12 @@ struct GridwiseGemm_xdl_cshuffle_conv_v3
                                    BDataType,
                                    ComputeTypeA,
                                    AccDataType,
-                                   decltype(GetABlockDescriptor_AK0PerBlock_MPerBlock_AK1(get_device_arch())),
-                                   decltype(GetBBlockDescriptor_BK0PerBlock_NPerBlock_BK1(get_device_arch())),
+                                   decltype(GetABlockDescriptor_AK0PerBlock_MPerBlock_AK1()),
+                                   decltype(GetBBlockDescriptor_BK0PerBlock_NPerBlock_BK1()),
                                    decltype(MakeAMmaTileDescriptor_M0_M1_M2_K(
-                     GetABlockDescriptor_AK0PerBlock_MPerBlock_AK1(get_device_arch()))),
+                     GetABlockDescriptor_AK0PerBlock_MPerBlock_AK1())),
                                    decltype(MakeBMmaTileDescriptor_N0_N1_N2_K(
-                     GetBBlockDescriptor_BK0PerBlock_NPerBlock_BK1(get_device_arch()))),
+                     GetBBlockDescriptor_BK0PerBlock_NPerBlock_BK1())),
                                    ABlockTransferSrcScalarPerVector,
                                    BBlockTransferSrcScalarPerVector,
                                    MPerBlock,