diff --git a/include/ck/tensor_operation/gpu/device/impl/device_moe_mx_gemm_bpreshuffle.hpp b/include/ck/tensor_operation/gpu/device/impl/device_moe_mx_gemm_bpreshuffle.hpp
index 6b2228f439..e64970145b 100644
--- a/include/ck/tensor_operation/gpu/device/impl/device_moe_mx_gemm_bpreshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/impl/device_moe_mx_gemm_bpreshuffle.hpp
@@ -122,7 +122,7 @@ struct DeviceMoeGemmMXBPreShuffle : public DeviceMoEGemmMXBPreShuffle<ALayout,
         MPerXDL,
         NPerXDL,
         MXdlPerWave,
-        math::max(2, NXdlPerWave_),
+        math::max(CShuffleNXdlPerWavePerShuffle, NXdlPerWave_),
         ABlockTransferThreadClusterLengths_AK0_M_AK1,
         ABlockTransferThreadClusterArrangeOrder,
         ABlockTransferSrcAccessOrder,