refactor

composable_kernel/include/tensor_operation/blockwise_gemm_v2.hpp

@@ -229,10 +229,10 @@ template <index_t BlockSize,
           index_t M1PerThread,
           index_t N1PerThread,
           index_t KPerThread,
-          index_t M1N1ThreadClusterM11,
-          index_t M1N1ThreadClusterN11,
           index_t M1N1ThreadClusterM10,
           index_t M1N1ThreadClusterN10,
+          index_t M1N1ThreadClusterM11,
+          index_t M1N1ThreadClusterN11,
           index_t AThreadCopyScalarPerVector_M1,
           index_t BThreadCopyScalarPerVector_N1,
           typename std::enable_if<ABlockDesc::IsKnownAtCompileTime() &&

driver/include/device_dynamic_convolution_forward_implicit_gemm_v4r4_nchw_kcyx_nkhw.hpp

@@ -506,10 +506,10 @@ void device_dynamic_convolution_forward_implicit_gemm_v4r4_nchw_kcyx_nkhw(
             GemmMPerThread,
             GemmNPerThread,
             GemmKPerThread,
-            GemmMLevel0Cluster,
-            GemmNLevel0Cluster,
             GemmMLevel1Cluster,
             GemmNLevel1Cluster,
+            GemmMLevel0Cluster,
+            GemmNLevel0Cluster,
             GemmABlockTransferThreadSliceLengths_GemmK_GemmM,
             GemmABlockTransferThreadClusterLengths_GemmK_GemmM,
             Sequence<1, 0>,