refactor

include/ck/tensor_operation/gpu/block/thread_group_tensor_slice_transfer_v7.hpp

@@ -13,20 +13,20 @@ namespace ck {
 // 2. ThreadwiseTensorSliceTransfer_v3 does not keep reference to tensor descriptor
 // 3. ThreadwiseTensorSliceTransfer_v3::Run() does not construct new tensor coordinate
 template <typename ThreadGroup,
-          typename ElementwiseOperation,
-          typename SliceLengths,
-          typename ThreadClusterLengths,
-          typename ThreadClusterArrangeOrder,
           typename SrcDatas,
           typename DstDatas,
           typename SrcDescs,
           typename DstDescs,
+          typename ElementwiseOperation,
+          typename DstInMemOps, // Sequence<InMemoryDataOperationEnum ...>
+          typename SliceLengths,
+          typename ThreadClusterLengths,
+          typename ThreadClusterArrangeOrder,
           typename DimAccessOrder,
           index_t VectorDim,
           index_t ScalarPerVector,
           typename ThreadTransferSrcResetCoordinateAfterRunFlags,
-          typename ThreadTransferDstResetCoordinateAfterRunFlags,
-          InMemoryDataOperationEnum... DstInMemOps>
+          typename ThreadTransferDstResetCoordinateAfterRunFlags>
 struct ThreadGroupTensorSliceTransfer_v7
 {
     static constexpr index_t nDim =
@@ -147,13 +147,13 @@ struct ThreadGroupTensorSliceTransfer_v7
                                          SrcDescs,
                                          DstDescs,
                                          ElementwiseOperation,
+                                         DstInMemOps,
                                          decltype(thread_slice_lengths),
                                          DimAccessOrder,
                                          VectorDim,
                                          ScalarPerVector,
                                          ThreadTransferSrcResetCoordinateAfterRunFlags,
-                                         ThreadTransferDstResetCoordinateAfterRunFlags,
-                                         DstInMemOps...>;
+                                         ThreadTransferDstResetCoordinateAfterRunFlags>;
 
     ThreadwiseTransfer threadwise_transfer_;
 };

include/ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp

@@ -548,26 +548,27 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
                                         ds_grid_desc_mblock_mperblock_nblock_nperblock[I1]);
 
             auto cde_block_copy_lds_and_global = ThreadGroupTensorSliceTransfer_v7<
-                ThisThreadBlock,         // ThreadGroup
-                CDEElementwiseOperation, // ElementwiseOperation,
-                Sequence<1,
-                         CShuffleMXdlPerWavePerShuffle * MWave * MPerXdl,
-                         1,
-                         CShuffleNXdlPerWavePerShuffle * NWave * NPerXdl>, // BlockSliceLengths,
-                CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
-                Sequence<0, 1, 2, 3>, // typename ThreadClusterArrangeOrder,
+                ThisThreadBlock, // ThreadGroup
                 Tuple<FloatCShuffle,
                       remove_cvref_t<tuple_element_t<0, DsDataType>>,
                       remove_cvref_t<tuple_element_t<1, DsDataType>>>,
                 Tuple<FloatE>, // typename DstData,
                 decltype(c_ds_descs),
                 decltype(tie(e_grid_desc_mblock_mperblock_nblock_nperblock)),
-                Sequence<0, 1, 2, 3>,                             // typename DimAccessOrder,
-                3,                                                // index_t VectorDim,
+                CDEElementwiseOperation,                                    // ElementwiseOperation,
+                Sequence<static_cast<index_t>(EGlobalMemoryDataOperation)>, // FIXME: make Sequence
+                                                                            // support arbitray type
+                Sequence<1,
+                         CShuffleMXdlPerWavePerShuffle * MWave * MPerXdl,
+                         1,
+                         CShuffleNXdlPerWavePerShuffle * NWave * NPerXdl>, // BlockSliceLengths,
+                CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
+                Sequence<0, 1, 2, 3>, // typename ThreadClusterArrangeOrder,
+                Sequence<0, 1, 2, 3>, // typename DimAccessOrder,
+                3,                    // index_t VectorDim,
                 CDEShuffleBlockTransferScalarPerVector_NPerBlock, // index_t ScalarPerVector,
                 Sequence<true, false, false>, // bool ThreadTransferSrcResetCoordinateAfterRunFlags
-                Sequence<false>,              // bool ThreadTransferDstResetCoordinateAfterRunFlags
-                EGlobalMemoryDataOperation>   // DstInMemOp,
+                Sequence<false>>              // bool ThreadTransferDstResetCoordinateAfterRunFlags
                 {c_ds_descs,
                  make_tuple(make_multi_index(0, 0, 0, 0),
                             make_multi_index(block_work_idx[I0], 0, block_work_idx[I1], 0),

include/ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer_v7.hpp

@@ -24,13 +24,13 @@ template <typename SrcDatas,
           typename SrcDescs,
           typename DstDescs,
           typename ElementwiseOperation,
+          typename DstInMemOps, // Sequence<InMemoryDataOperationEnum ...>
           typename SliceLengths,
           typename DimAccessOrder,
           index_t VectorDim,
           index_t ScalarPerVector,
-          typename SrcResetCoordinateAfterRunFlags, // Sequence<...>
-          typename DstResetCoordinateAfterRunFlags, // Sequence<...>
-          InMemoryDataOperationEnum... DstInMemOps>
+          typename SrcResetCoordinateAfterRunFlags, // Sequence<bool ...>
+          typename DstResetCoordinateAfterRunFlags> // Sequence<bool ...>
 struct ThreadwiseTensorSliceTransfer_v7
 {
     static constexpr auto I0 = Number<0>{};
@@ -165,7 +165,8 @@ struct ThreadwiseTensorSliceTransfer_v7
                     coordinate_has_valid_offset_assuming_visible_index_is_valid(dst_descs[i],
                                                                                 dst_coords_[i]);
 
-                constexpr auto DstInMemOp = make_tuple(DstInMemOps...)[i];
+                constexpr InMemoryDataOperationEnum DstInMemOp =
+                    static_cast<InMemoryDataOperationEnum>(DstInMemOps::At(i.value));
 
                 dst_bufs(i).template Update<DstInMemOp, dst_vector_t>(
                     dst_coords_[i].GetOffset(),