Temporal sampling implementation, still debugging

cpp/CMakeLists.txt

@@ -150,10 +150,13 @@ endif()
 #       which should give us a better parallel schedule.
 
 set(CUGRAPH_SOURCES
+    src/sampling/temporal_sampling_sg_v32_e32.cu
+    src/sampling/temporal_sampling_sg_v64_e64.cu
     src/utilities/shuffle_vertices_mg_v32_fp.cu
     src/utilities/shuffle_vertices_mg_v32_integral.cu
     src/utilities/shuffle_vertices_mg_v64_fp.cu
     src/utilities/shuffle_vertices_mg_v64_integral.cu
+    src/utilities/validation_checks.cu
     src/detail/permute_range_v32.cu
     src/detail/permute_range_v64.cu
     src/utilities/shuffle_vertex_pairs_mg_v32_e32.cu
@@ -262,12 +265,24 @@ set(CUGRAPH_SOURCES
     src/sampling/detail/check_edge_bias_values_sg_v32_e32.cu
     src/sampling/detail/check_edge_bias_values_mg_v64_e64.cu
     src/sampling/detail/check_edge_bias_values_mg_v32_e32.cu
+    src/sampling/detail/temporal_partition_vertices_v64.cu
+    src/sampling/detail/temporal_partition_vertices_v32.cu
     src/sampling/detail/sample_edges_sg_v64_e64.cu
     src/sampling/detail/sample_edges_sg_v32_e32.cu
     src/sampling/detail/sample_edges_mg_v64_e64.cu
     src/sampling/detail/sample_edges_mg_v32_e32.cu
-    src/sampling/detail/shuffle_and_organize_output_mg_v64_e64.cu
-    src/sampling/detail/shuffle_and_organize_output_mg_v32_e32.cu
+    src/sampling/detail/temporal_sample_edges_sg_v64_e64.cu
+    src/sampling/detail/temporal_sample_edges_sg_v32_e32.cu
+    src/sampling/detail/temporal_sample_edges_mg_v64_e64.cu
+    src/sampling/detail/temporal_sample_edges_mg_v32_e32.cu
+    src/sampling/detail/shuffle_and_organize_output_mg_v64_e64_t32_w32.cu
+    src/sampling/detail/shuffle_and_organize_output_mg_v32_e32_t32_w32.cu
+    src/sampling/detail/shuffle_and_organize_output_mg_v64_e64_t32_w64.cu
+    src/sampling/detail/shuffle_and_organize_output_mg_v32_e32_t32_w64.cu
+    src/sampling/detail/shuffle_and_organize_output_mg_v64_e64_t64_w32.cu
+    src/sampling/detail/shuffle_and_organize_output_mg_v32_e32_t64_w32.cu
+    src/sampling/detail/shuffle_and_organize_output_mg_v64_e64_t64_w64.cu
+    src/sampling/detail/shuffle_and_organize_output_mg_v32_e32_t64_w64.cu
     src/sampling/neighbor_sampling_mg_v32_e32.cu
     src/sampling/neighbor_sampling_mg_v64_e64.cu
     src/sampling/neighbor_sampling_sg_v32_e32.cu

cpp/include/cugraph/detail/shuffle_wrappers.hpp

@@ -200,6 +200,33 @@ shuffle_int_vertex_value_pairs_to_local_gpu_by_vertex_partitioning(
   rmm::device_uvector<value_t>&& values,
   raft::host_span<vertex_t const> vertex_partition_range_lasts);
 
+/**
+ * @ingroup shuffle_wrappers_cpp
+ * @brief Shuffle vertices using the internal vertex key function which returns the target GPU ID.
+ *
+ * @tparam vertex_t Type of vertex identifiers. Needs to be an integral type.
+ * @tparam value1_t Type of first vertex values. Needs to be an integral type.
+ * @tparam value2_t Type of second vertex values. Needs to be an integral type.
+ *
+ * @param[in] handle RAFT handle object to encapsulate resources (e.g. CUDA stream, communicator,
+ * @param[in] vertices Vertex IDs to shuffle
+ * @param[in] values1 First Vertex Values to shuffle
+ * @param[in] values2 Second Vertex Values to shuffle
+ * @param[in] vertex_partition_range_lasts From graph view, vector of last vertex id for each gpu
+ *
+ * @return tuple containing device vector of shuffled vertices and device vectors of corresponding
+ *         values
+ */
+template <typename vertex_t, typename value1_t, typename value2_t>
+std::
+  tuple<rmm::device_uvector<vertex_t>, rmm::device_uvector<value1_t>, rmm::device_uvector<value2_t>>
+  shuffle_int_vertex_two_value_pairs_to_local_gpu_by_vertex_partitioning(
+    raft::handle_t const& handle,
+    rmm::device_uvector<vertex_t>&& vertices,
+    rmm::device_uvector<value1_t>&& values1,
+    rmm::device_uvector<value2_t>&& values2,
+    raft::host_span<vertex_t const> vertex_partition_range_lasts);
+
 /**
  * @ingroup shuffle_wrappers_cpp
  * @brief Groupby and count edgelist using the key function which returns the target local partition

cpp/include/cugraph/edge_property.hpp

@@ -26,6 +26,7 @@
 #include <thrust/iterator/iterator_traits.h>
 
 #include <optional>
+#include <tuple>
 #include <type_traits>
 
 namespace cugraph {
@@ -202,4 +203,16 @@ auto view_concat(edge_property_view_t<edge_t, Iters, Types> const&... views)
     edge_partition_concat_value_firsts, first_view.edge_counts());
 }
 
+template <typename... Ts, std::size_t... Is>
+auto view_concat_impl(std::tuple<Ts...> const& tuple, std::index_sequence<Is...>)
+{
+  return view_concat(std::get<Is>(tuple)...);
+}
+
+template <typename... Ts>
+auto view_concat(std::tuple<Ts...> const& tuple)
+{
+  return view_concat_impl(tuple, std::index_sequence_for<Ts...>{});
+}
+
 }  // namespace cugraph