Merge branch 'branch-24.12' into branch-24.12_optimize-ktruss

CHANGELOG.md

@@ -1,3 +1,79 @@
+# cugraph 24.12.00 (11 Dec 2024)
+
+## 🚨 Breaking Changes
+
+- Fix SSL Error ([#4825](https://github.com/rapidsai/cugraph/pull/4825)) [@nv-rliu](https://github.com/nv-rliu)
+- [BUG] Set Max PyTorch Version, Skip 11.4 Tests Using WholeGraph ([#4808](https://github.com/rapidsai/cugraph/pull/4808)) [@alexbarghi-nv](https://github.com/alexbarghi-nv)
+- Remove GNN Packages ([#4765](https://github.com/rapidsai/cugraph/pull/4765)) [@alexbarghi-nv](https://github.com/alexbarghi-nv)
+- remove cugraph-dgl and cugraph-pyg building/testing/publishing ([#4752](https://github.com/rapidsai/cugraph/pull/4752)) [@jameslamb](https://github.com/jameslamb)
+- Remove `nx-cugraph` build/test/publish from cugraph ([#4748](https://github.com/rapidsai/cugraph/pull/4748)) [@nv-rliu](https://github.com/nv-rliu)
+- Remove CMake/C++ references to cugraph-ops ([#4744](https://github.com/rapidsai/cugraph/pull/4744)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Delete the deprecated data_type_id_t enum ([#4737](https://github.com/rapidsai/cugraph/pull/4737)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Don't compile int32_t/int64_t vertex_t/edge_t combinations ([#4720](https://github.com/rapidsai/cugraph/pull/4720)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Remove deprecated C API functions for graph creation/graph free ([#4718](https://github.com/rapidsai/cugraph/pull/4718)) [@ChuckHastings](https://github.com/ChuckHastings)
+
+## 🐛 Bug Fixes
+
+- Fix SSL Error ([#4825](https://github.com/rapidsai/cugraph/pull/4825)) [@nv-rliu](https://github.com/nv-rliu)
+- [BUG] Set Max PyTorch Version, Skip 11.4 Tests Using WholeGraph ([#4808](https://github.com/rapidsai/cugraph/pull/4808)) [@alexbarghi-nv](https://github.com/alexbarghi-nv)
+- Fix CCCL 2.7.0-rc2 compile issue by removing reference from values. ([#4799](https://github.com/rapidsai/cugraph/pull/4799)) [@bdice](https://github.com/bdice)
+- Re-enable balanced edge cut test ([#4781](https://github.com/rapidsai/cugraph/pull/4781)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Fix debug build failure ([#4774](https://github.com/rapidsai/cugraph/pull/4774)) [@seunghwak](https://github.com/seunghwak)
+- Add sphinx-lint pre-commit and some docs fixes ([#4771](https://github.com/rapidsai/cugraph/pull/4771)) [@eriknw](https://github.com/eriknw)
+- Fix improper initialization of degree_type ([#4755](https://github.com/rapidsai/cugraph/pull/4755)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Put a ceiling on cuda-python ([#4747](https://github.com/rapidsai/cugraph/pull/4747)) [@bdice](https://github.com/bdice)
+- Fix MG similarity issues ([#4741](https://github.com/rapidsai/cugraph/pull/4741)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Address Leiden clustering generating too many clusters ([#4730](https://github.com/rapidsai/cugraph/pull/4730)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Add support for storing results for all k-values ([#4728](https://github.com/rapidsai/cugraph/pull/4728)) [@nv-rliu](https://github.com/nv-rliu)
+- Remove `--collect-only` Option Left by Accident ([#4727](https://github.com/rapidsai/cugraph/pull/4727)) [@nv-rliu](https://github.com/nv-rliu)
+- Update nx-cugraph to NetworkX 3.4 ([#4717](https://github.com/rapidsai/cugraph/pull/4717)) [@eriknw](https://github.com/eriknw)
+- Symmetrize edgelist when creating a CSR graph ([#4716](https://github.com/rapidsai/cugraph/pull/4716)) [@jnke2016](https://github.com/jnke2016)
+
+## 📖 Documentation
+
+- nx-cugraph: add `bipartite_betweenness_centrality` to supported algorithms ([#4778](https://github.com/rapidsai/cugraph/pull/4778)) [@eriknw](https://github.com/eriknw)
+- Notebook to run Centrality against PatentsView data for Medium Blog ([#4769](https://github.com/rapidsai/cugraph/pull/4769)) [@acostadon](https://github.com/acostadon)
+- Drop support for NetworkX 3.0 and 3.1 for nx-cugraph ([#4766](https://github.com/rapidsai/cugraph/pull/4766)) [@eriknw](https://github.com/eriknw)
+- Update nx-cugraph Docs URLs ([#4761](https://github.com/rapidsai/cugraph/pull/4761)) [@nv-rliu](https://github.com/nv-rliu)
+- Update `nx-cugraph` README with New Env Variable ([#4705](https://github.com/rapidsai/cugraph/pull/4705)) [@nv-rliu](https://github.com/nv-rliu)
+
+## 🚀 New Features
+
+- [FEA] Support Edge ID Lookup in PyLibcuGraph ([#4687](https://github.com/rapidsai/cugraph/pull/4687)) [@alexbarghi-nv](https://github.com/alexbarghi-nv)
+
+## 🛠️ Improvements
+
+- Increase max_iterations in MG HITS TEST ([#4783](https://github.com/rapidsai/cugraph/pull/4783)) [@seunghwak](https://github.com/seunghwak)
+- Updates READMEs, updates `core_number` to properly ignore `degree_type`, minor cleanup ([#4776](https://github.com/rapidsai/cugraph/pull/4776)) [@rlratzel](https://github.com/rlratzel)
+- Remove edge renumber map from the homogeneous sampling API ([#4775](https://github.com/rapidsai/cugraph/pull/4775)) [@jnke2016](https://github.com/jnke2016)
+- Remove GNN Packages ([#4765](https://github.com/rapidsai/cugraph/pull/4765)) [@alexbarghi-nv](https://github.com/alexbarghi-nv)
+- Remove cugraph-equivariance ([#4762](https://github.com/rapidsai/cugraph/pull/4762)) [@BradReesWork](https://github.com/BradReesWork)
+- deprecate NetworkX Graph as input ([#4759](https://github.com/rapidsai/cugraph/pull/4759)) [@BradReesWork](https://github.com/BradReesWork)
+- Remove `nx-cugraph` from this repo ([#4756](https://github.com/rapidsai/cugraph/pull/4756)) [@eriknw](https://github.com/eriknw)
+- enforce wheel size limits, README formatting in CI ([#4754](https://github.com/rapidsai/cugraph/pull/4754)) [@jameslamb](https://github.com/jameslamb)
+- remove cugraph-dgl and cugraph-pyg building/testing/publishing ([#4752](https://github.com/rapidsai/cugraph/pull/4752)) [@jameslamb](https://github.com/jameslamb)
+- Primitives & BFS performance improvements ([#4751](https://github.com/rapidsai/cugraph/pull/4751)) [@seunghwak](https://github.com/seunghwak)
+- extract the edgelist from the graph ([#4750](https://github.com/rapidsai/cugraph/pull/4750)) [@jnke2016](https://github.com/jnke2016)
+- Remove `nx-cugraph` build/test/publish from cugraph ([#4748](https://github.com/rapidsai/cugraph/pull/4748)) [@nv-rliu](https://github.com/nv-rliu)
+- Remove CMake/C++ references to cugraph-ops ([#4744](https://github.com/rapidsai/cugraph/pull/4744)) [@ChuckHastings](https://github.com/ChuckHastings)
+- add telemetry ([#4740](https://github.com/rapidsai/cugraph/pull/4740)) [@msarahan](https://github.com/msarahan)
+- Delete the deprecated data_type_id_t enum ([#4737](https://github.com/rapidsai/cugraph/pull/4737)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Updates README with new dataset, removes mention of script no longer used ([#4736](https://github.com/rapidsai/cugraph/pull/4736)) [@rlratzel](https://github.com/rlratzel)
+- devcontainer: replace `VAULT_HOST` with `AWS_ROLE_ARN` ([#4732](https://github.com/rapidsai/cugraph/pull/4732)) [@jjacobelli](https://github.com/jjacobelli)
+- Adds new dataset for benchmarking in the 100k node 1M edge range, adds additional k-values for BC benchmarks ([#4726](https://github.com/rapidsai/cugraph/pull/4726)) [@rlratzel](https://github.com/rlratzel)
+- re-run all CI when files in .github/workflows change ([#4723](https://github.com/rapidsai/cugraph/pull/4723)) [@jameslamb](https://github.com/jameslamb)
+- Don't compile int32_t/int64_t vertex_t/edge_t combinations ([#4720](https://github.com/rapidsai/cugraph/pull/4720)) [@ChuckHastings](https://github.com/ChuckHastings)
+- print sccache stats in builds ([#4719](https://github.com/rapidsai/cugraph/pull/4719)) [@jameslamb](https://github.com/jameslamb)
+- Remove deprecated C API functions for graph creation/graph free ([#4718](https://github.com/rapidsai/cugraph/pull/4718)) [@ChuckHastings](https://github.com/ChuckHastings)
+- Clean up the use of cuco hash functions ([#4707](https://github.com/rapidsai/cugraph/pull/4707)) [@PointKernel](https://github.com/PointKernel)
+- Relax PyTorch upper bound (allowing 2.4) ([#4703](https://github.com/rapidsai/cugraph/pull/4703)) [@jakirkham](https://github.com/jakirkham)
+- combine pip install calls in wheel-testing scripts ([#4701](https://github.com/rapidsai/cugraph/pull/4701)) [@jameslamb](https://github.com/jameslamb)
+- Merge branch-24.10 into branch-24.12 ([#4697](https://github.com/rapidsai/cugraph/pull/4697)) [@jameslamb](https://github.com/jameslamb)
+- Merge branch-24.10 into branch-24.12 ([#4682](https://github.com/rapidsai/cugraph/pull/4682)) [@jameslamb](https://github.com/jameslamb)
+- Update all rmm imports to use pylibrmm/librmm ([#4671](https://github.com/rapidsai/cugraph/pull/4671)) [@Matt711](https://github.com/Matt711)
+- Prune workflows based on changed files ([#4634](https://github.com/rapidsai/cugraph/pull/4634)) [@KyleFromNVIDIA](https://github.com/KyleFromNVIDIA)
+- support heterogenous fanout type ([#4608](https://github.com/rapidsai/cugraph/pull/4608)) [@jnke2016](https://github.com/jnke2016)
+
 # cugraph 24.10.00 (9 Oct 2024)
 
 ## 🚨 Breaking Changes

conda/environments/all_cuda-118_arch-x86_64.yaml

@@ -10,6 +10,7 @@ dependencies:
 - aiohttp
 - breathe
 - c-compiler
+- certifi
 - cmake>=3.26.4,!=3.30.0
 - cuda-nvtx
 - cuda-version=11.8
@@ -53,7 +54,7 @@ dependencies:
 - pytest-cov
 - pytest-xdist
 - python-louvain
-- pytorch>=2.3
+- pytorch>=2.3,<2.5a0
 - raft-dask==24.12.*,>=0.0.0a0
 - rapids-build-backend>=0.3.1,<0.4.0.dev0
 - rapids-dask-dependency==24.12.*,>=0.0.0a0

conda/environments/all_cuda-125_arch-x86_64.yaml

@@ -10,6 +10,7 @@ dependencies:
 - aiohttp
 - breathe
 - c-compiler
+- certifi
 - cmake>=3.26.4,!=3.30.0
 - cuda-cudart-dev
 - cuda-nvcc
@@ -58,7 +59,7 @@ dependencies:
 - pytest-cov
 - pytest-xdist
 - python-louvain
-- pytorch>=2.3
+- pytorch>=2.3,<2.5a0
 - raft-dask==24.12.*,>=0.0.0a0
 - rapids-build-backend>=0.3.1,<0.4.0.dev0
 - rapids-dask-dependency==24.12.*,>=0.0.0a0

cpp/include/cugraph/sampling_functions.hpp

@@ -306,7 +306,7 @@ struct sampling_flags_t {
  * @param edge_type_view Optional view object holding edge types for @p graph_view.
  * @param starting_vertices Device span of starting vertex IDs for the sampling.
  * In a multi-gpu context the starting vertices should be local to this GPU.
- * @param starting_vertex_label_offsets Optional device span of labels associated with each starting
+ * @param starting_vertex_labels Optional device span of labels associated with each starting
  * vertex for the sampling.
  * @param label_to_output_comm_rank Optional device span identifying which rank should get sampling
  * outputs of each vertex label.  This should be the same on each rank.
@@ -340,7 +340,7 @@ homogeneous_uniform_neighbor_sample(
   std::optional<edge_property_view_t<edge_t, edge_t const*>> edge_id_view,
   std::optional<edge_property_view_t<edge_t, edge_type_t const*>> edge_type_view,
   raft::device_span<vertex_t const> starting_vertices,
-  std::optional<raft::device_span<int32_t const>> starting_vertex_label_offsets,
+  std::optional<raft::device_span<int32_t const>> starting_vertex_labels,
   std::optional<raft::device_span<int32_t const>> label_to_output_comm_rank,
   raft::host_span<int32_t const> fan_out,
   sampling_flags_t sampling_flags,
@@ -385,7 +385,7 @@ homogeneous_uniform_neighbor_sample(
  * corresponding edge can never be selected.
  * @param starting_vertices Device span of starting vertex IDs for the sampling.
  * In a multi-gpu context the starting vertices should be local to this GPU.
- * @param starting_vertex_label_offsets Optional device span of labels associated with each starting
+ * @param starting_vertex_labels Optional device span of labels associated with each starting
  * vertex for the sampling.
  * @param label_to_output_comm_rank Optional device span identifying which rank should get sampling
  * outputs of each vertex label.  This should be the same on each rank.
@@ -421,7 +421,7 @@ homogeneous_biased_neighbor_sample(
   std::optional<edge_property_view_t<edge_t, edge_type_t const*>> edge_type_view,
   edge_property_view_t<edge_t, bias_t const*> edge_bias_view,
   raft::device_span<vertex_t const> starting_vertices,
-  std::optional<raft::device_span<int32_t const>> starting_vertex_label_offsets,
+  std::optional<raft::device_span<int32_t const>> starting_vertex_labels,
   std::optional<raft::device_span<int32_t const>> label_to_output_comm_rank,
   raft::host_span<int32_t const> fan_out,
   sampling_flags_t sampling_flags,
@@ -462,7 +462,7 @@ homogeneous_biased_neighbor_sample(
  * @param edge_type_view Optional view object holding edge types for @p graph_view.
  * @param starting_vertices Device span of starting vertex IDs for the sampling.
  * In a multi-gpu context the starting vertices should be local to this GPU.
- * @param starting_vertex_label_offsets Optional device span of labels associated with each starting
+ * @param starting_vertex_labels Optional device span of labels associated with each starting
  * vertex for the sampling.
  * @param label_to_output_comm_rank Optional device span identifying which rank should get sampling
  * outputs of each vertex label.  This should be the same on each rank.
@@ -498,7 +498,7 @@ heterogeneous_uniform_neighbor_sample(
   std::optional<edge_property_view_t<edge_t, edge_t const*>> edge_id_view,
   std::optional<edge_property_view_t<edge_t, edge_type_t const*>> edge_type_view,
   raft::device_span<vertex_t const> starting_vertices,
-  std::optional<raft::device_span<int32_t const>> starting_vertex_label_offsets,
+  std::optional<raft::device_span<int32_t const>> starting_vertex_labels,
   std::optional<raft::device_span<int32_t const>> label_to_output_comm_rank,
   raft::host_span<int32_t const> fan_out,
   edge_type_t num_edge_types,
@@ -545,7 +545,7 @@ heterogeneous_uniform_neighbor_sample(
  * corresponding edge can never be selected.
  * @param starting_vertices Device span of starting vertex IDs for the sampling.
  * In a multi-gpu context the starting vertices should be local to this GPU.
- * @param starting_vertex_label_offsets Optional device span of labels associated with each starting
+ * @param starting_vertex_labels Optional device span of labels associated with each starting
  * vertex for the sampling.
  * @param label_to_output_comm_rank Optional device span identifying which rank should get sampling
  * outputs of each vertex label.  This should be the same on each rank.
@@ -583,7 +583,7 @@ heterogeneous_biased_neighbor_sample(
   std::optional<edge_property_view_t<edge_t, edge_type_t const*>> edge_type_view,
   edge_property_view_t<edge_t, bias_t const*> edge_bias_view,
   raft::device_span<vertex_t const> starting_vertices,
-  std::optional<raft::device_span<int32_t const>> starting_vertex_label_offsets,
+  std::optional<raft::device_span<int32_t const>> starting_vertex_labels,
   std::optional<raft::device_span<int32_t const>> label_to_output_comm_rank,
   raft::host_span<int32_t const> fan_out,
   edge_type_t num_edge_types,

cpp/include/cugraph_c/sampling_algorithms.h

@@ -551,6 +551,8 @@ cugraph_error_code_t cugraph_homogeneous_biased_neighbor_sample(
  * @param [in]  start_vertices Device array of start vertices for the sampling
  * @param [in]  starting_vertex_label_offsets Device array of the offsets for each label in
  * the seed list. This parameter is only used with the retain_seeds option.
+ * @param [in]  vertex_type_offsets Device array of the offsets for each vertex type in the
+ * graph.
  * @param [in]  fan_out       Host array defining the fan out at each step in the sampling
  * algorithm. We only support fan_out values of type INT32
  * @param [in]  num_edge_types Number of edge types where a value of 1 translates to homogeneous
@@ -570,6 +572,7 @@ cugraph_error_code_t cugraph_heterogeneous_uniform_neighbor_sample(
   cugraph_graph_t* graph,
   const cugraph_type_erased_device_array_view_t* start_vertices,
   const cugraph_type_erased_device_array_view_t* starting_vertex_label_offsets,
+  const cugraph_type_erased_device_array_view_t* vertex_type_offsets,
   const cugraph_type_erased_host_array_view_t* fan_out,
   int num_edge_types,
   const cugraph_sampling_options_t* options,
@@ -598,6 +601,8 @@ cugraph_error_code_t cugraph_heterogeneous_uniform_neighbor_sample(
  * @param [in]  start_vertices Device array of start vertices for the sampling
  * @param [in]  starting_vertex_label_offsets Device array of the offsets for each label in
  * the seed list. This parameter is only used with the retain_seeds option.
+ * @param [in]  vertex_type_offsets Device array of the offsets for each vertex type in the
+ * graph.
  * @param [in]  fan_out       Host array defining the fan out at each step in the sampling
  * algorithm. We only support fan_out values of type INT32
  * @param [in]  num_edge_types Number of edge types where a value of 1 translates to homogeneous
@@ -618,6 +623,7 @@ cugraph_error_code_t cugraph_heterogeneous_biased_neighbor_sample(
   const cugraph_edge_property_view_t* edge_biases,
   const cugraph_type_erased_device_array_view_t* start_vertices,
   const cugraph_type_erased_device_array_view_t* starting_vertex_label_offsets,
+  const cugraph_type_erased_device_array_view_t* vertex_type_offsets,
   const cugraph_type_erased_host_array_view_t* fan_out,
   int num_edge_types,
   const cugraph_sampling_options_t* options,
@@ -735,6 +741,16 @@ cugraph_type_erased_device_array_view_t* cugraph_sample_result_get_hop(
 cugraph_type_erased_device_array_view_t* cugraph_sample_result_get_label_hop_offsets(
   const cugraph_sample_result_t* result);
 
+/**
+ * @ingroup samplingC
+ * @brief     Get the label-type-hop offsets from the sampling algorithm result
+ *
+ * @param [in]   result   The result from a sampling algorithm
+ * @return type erased array pointing to the label-type-hop offsets
+ */
+cugraph_type_erased_device_array_view_t* cugraph_sample_result_get_label_type_hop_offsets(
+  const cugraph_sample_result_t* result);
+
 /**
  * @ingroup samplingC
  * @brief     Get the index from the sampling algorithm result

cpp/libcugraph_etl/src/renumbering.cu

@@ -730,7 +730,7 @@ __global__ static void select_unrenumber_string(str_hash_value* idx_to_col_row,
 }
 
 struct struct_sort_descending {
-  __host__ __device__ bool operator()(str_hash_value& a, str_hash_value& b)
+  __host__ __device__ bool operator()(str_hash_value a, str_hash_value b)
   {
     return (a.count_ > b.count_);
   }