Merge branch 'branch-25.02' into renumber_for_tigergraph

rapidsai · Jan 31, 2025 · 2b9721b · 2b9721b
2 parents 9e19ad8 + dc435cf
commit 2b9721b
Show file tree

Hide file tree

Showing 92 changed files with 1,649 additions and 1,098 deletions.
diff --git a/...ontainer/cuda12.5-conda/devcontainer.json → ...ontainer/cuda12.8-conda/devcontainer.json b/...ontainer/cuda12.5-conda/devcontainer.json → ...ontainer/cuda12.8-conda/devcontainer.json
@@ -3,15 +3,15 @@
     "context": "${localWorkspaceFolder}/.devcontainer",
     "dockerfile": "${localWorkspaceFolder}/.devcontainer/Dockerfile",
     "args": {
-      "CUDA": "12.5",
+      "CUDA": "12.8",
       "PYTHON_PACKAGE_MANAGER": "conda",
       "BASE": "rapidsai/devcontainers:25.02-cpp-mambaforge-ubuntu22.04"
     }
   },
   "runArgs": [
     "--rm",
     "--name",
-    "${localEnv:USER:anon}-rapids-${localWorkspaceFolderBasename}-25.02-cuda12.5-conda"
+    "${localEnv:USER:anon}-rapids-${localWorkspaceFolderBasename}-25.02-cuda12.8-conda"
   ],
   "hostRequirements": {"gpu": "optional"},
   "features": {
@@ -20,7 +20,7 @@
   "overrideFeatureInstallOrder": [
     "ghcr.io/rapidsai/devcontainers/features/rapids-build-utils"
   ],
-  "initializeCommand": ["/bin/bash", "-c", "mkdir -m 0755 -p ${localWorkspaceFolder}/../.{aws,cache,config,conda/pkgs,conda/${localWorkspaceFolderBasename}-cuda12.5-envs}"],
+  "initializeCommand": ["/bin/bash", "-c", "mkdir -m 0755 -p ${localWorkspaceFolder}/../.{aws,cache,config,conda/pkgs,conda/${localWorkspaceFolderBasename}-cuda12.8-envs}"],
   "postAttachCommand": ["/bin/bash", "-c", "if [ ${CODESPACES:-false} = 'true' ]; then . devcontainer-utils-post-attach-command; . rapids-post-attach-command; fi"],
   "workspaceFolder": "/home/coder",
   "workspaceMount": "source=${localWorkspaceFolder},target=/home/coder/cugraph,type=bind,consistency=consistent",
@@ -29,7 +29,7 @@
     "source=${localWorkspaceFolder}/../.cache,target=/home/coder/.cache,type=bind,consistency=consistent",
     "source=${localWorkspaceFolder}/../.config,target=/home/coder/.config,type=bind,consistency=consistent",
     "source=${localWorkspaceFolder}/../.conda/pkgs,target=/home/coder/.conda/pkgs,type=bind,consistency=consistent",
-    "source=${localWorkspaceFolder}/../.conda/${localWorkspaceFolderBasename}-cuda12.5-envs,target=/home/coder/.conda/envs,type=bind,consistency=consistent"
+    "source=${localWorkspaceFolder}/../.conda/${localWorkspaceFolderBasename}-cuda12.8-envs,target=/home/coder/.conda/envs,type=bind,consistency=consistent"
   ],
   "customizations": {
     "vscode": {

diff --git a/.devcontainer/cuda12.5-pip/devcontainer.json → .devcontainer/cuda12.8-pip/devcontainer.json b/.devcontainer/cuda12.5-pip/devcontainer.json → .devcontainer/cuda12.8-pip/devcontainer.json
@@ -3,20 +3,20 @@
     "context": "${localWorkspaceFolder}/.devcontainer",
     "dockerfile": "${localWorkspaceFolder}/.devcontainer/Dockerfile",
     "args": {
-      "CUDA": "12.5",
+      "CUDA": "12.8",
       "PYTHON_PACKAGE_MANAGER": "pip",
-      "BASE": "rapidsai/devcontainers:25.02-cpp-cuda12.5-ucx1.18.0-openmpi-ubuntu22.04"
+      "BASE": "rapidsai/devcontainers:25.02-cpp-cuda12.8-ucx1.18.0-openmpi-ubuntu22.04"
     }
   },
   "runArgs": [
     "--rm",
     "--name",
-    "${localEnv:USER:anon}-rapids-${localWorkspaceFolderBasename}-25.02-cuda12.5-pip"
+    "${localEnv:USER:anon}-rapids-${localWorkspaceFolderBasename}-25.02-cuda12.8-pip"
   ],
   "hostRequirements": {"gpu": "optional"},
   "features": {
     "ghcr.io/rapidsai/devcontainers/features/cuda:25.2": {
-      "version": "12.5",
+      "version": "12.8",
       "installcuBLAS": true,
       "installcuSOLVER": true,
       "installcuRAND": true,
@@ -28,15 +28,15 @@
     "ghcr.io/rapidsai/devcontainers/features/cuda",
     "ghcr.io/rapidsai/devcontainers/features/rapids-build-utils"
   ],
-  "initializeCommand": ["/bin/bash", "-c", "mkdir -m 0755 -p ${localWorkspaceFolder}/../.{aws,cache,config/pip,local/share/${localWorkspaceFolderBasename}-cuda12.5-venvs}"],
+  "initializeCommand": ["/bin/bash", "-c", "mkdir -m 0755 -p ${localWorkspaceFolder}/../.{aws,cache,config/pip,local/share/${localWorkspaceFolderBasename}-cuda12.8-venvs}"],
   "postAttachCommand": ["/bin/bash", "-c", "if [ ${CODESPACES:-false} = 'true' ]; then . devcontainer-utils-post-attach-command; . rapids-post-attach-command; fi"],
   "workspaceFolder": "/home/coder",
   "workspaceMount": "source=${localWorkspaceFolder},target=/home/coder/cugraph,type=bind,consistency=consistent",
   "mounts": [
     "source=${localWorkspaceFolder}/../.aws,target=/home/coder/.aws,type=bind,consistency=consistent",
     "source=${localWorkspaceFolder}/../.cache,target=/home/coder/.cache,type=bind,consistency=consistent",
     "source=${localWorkspaceFolder}/../.config,target=/home/coder/.config,type=bind,consistency=consistent",
-    "source=${localWorkspaceFolder}/../.local/share/${localWorkspaceFolderBasename}-cuda12.5-venvs,target=/home/coder/.local/share/venvs,type=bind,consistency=consistent"
+    "source=${localWorkspaceFolder}/../.local/share/${localWorkspaceFolderBasename}-cuda12.8-venvs,target=/home/coder/.local/share/venvs,type=bind,consistency=consistent"
   ],
   "customizations": {
     "vscode": {

diff --git a/.github/workflows/pr.yaml b/.github/workflows/pr.yaml
@@ -195,7 +195,7 @@ jobs:
     uses: rapidsai/shared-workflows/.github/workflows/[email protected]
     with:
       arch: '["amd64"]'
-      cuda: '["12.5"]'
+      cuda: '["12.8"]'
       node_type: cpu32
       build_command: |
         sccache -z;

diff --git a/...nvironments/all_cuda-125_arch-x86_64.yaml → ...nvironments/all_cuda-128_arch-x86_64.yaml b/...nvironments/all_cuda-125_arch-x86_64.yaml → ...nvironments/all_cuda-128_arch-x86_64.yaml
@@ -17,7 +17,7 @@ dependencies:
 - cuda-nvtx-dev
 - cuda-profiler-api
 - cuda-python>=12.6.2,<13.0a0
-- cuda-version=12.5
+- cuda-version=12.8
 - cudf==25.2.*,>=0.0.0a0
 - cupy>=12.0.0
 - cxx-compiler
@@ -77,4 +77,4 @@ dependencies:
 - torchmetrics
 - ucx-py==0.42.*,>=0.0.0a0
 - wheel
-name: all_cuda-125_arch-x86_64
+name: all_cuda-128_arch-x86_64
diff --git a/cpp/include/cugraph/detail/decompress_edge_partition.cuh b/cpp/include/cugraph/detail/decompress_edge_partition.cuh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020-2024, NVIDIA CORPORATION.
+ * Copyright (c) 2020-2025, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -29,7 +29,6 @@
 #include <thrust/fill.h>
 #include <thrust/for_each.h>
 #include <thrust/iterator/counting_iterator.h>
-#include <thrust/optional.h>
 #include <thrust/sequence.h>
 #include <thrust/tuple.h>
 

diff --git a/cpp/include/cugraph/edge_partition_device_view.cuh b/cpp/include/cugraph/edge_partition_device_view.cuh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020-2024, NVIDIA CORPORATION.
+ * Copyright (c) 2020-2025, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -26,10 +26,10 @@
 #include <rmm/device_uvector.hpp>
 #include <rmm/exec_policy.hpp>
 
+#include <cuda/std/optional>
 #include <thrust/binary_search.h>
 #include <thrust/distance.h>
 #include <thrust/execution_policy.h>
-#include <thrust/optional.h>
 #include <thrust/transform.h>
 #include <thrust/transform_reduce.h>
 #include <thrust/tuple.h>
@@ -43,18 +43,18 @@ namespace cugraph {
 namespace detail {
 
 template <typename vertex_t>
-__device__ thrust::optional<vertex_t> major_hypersparse_idx_from_major_nocheck_impl(
+__device__ cuda::std::optional<vertex_t> major_hypersparse_idx_from_major_nocheck_impl(
   raft::device_span<vertex_t const> dcs_nzd_vertices, vertex_t major)
 {
   // we can avoid binary search (and potentially improve performance) if we add an auxiliary array
   // or cuco::static_map (at the expense of additional memory)
   auto it =
     thrust::lower_bound(thrust::seq, dcs_nzd_vertices.begin(), dcs_nzd_vertices.end(), major);
   return it != dcs_nzd_vertices.end()
-           ? (*it == major ? thrust::optional<vertex_t>{static_cast<vertex_t>(
+           ? (*it == major ? cuda::std::optional<vertex_t>{static_cast<vertex_t>(
                                thrust::distance(dcs_nzd_vertices.begin(), it))}
-                           : thrust::nullopt)
-           : thrust::nullopt;
+                           : cuda::std::nullopt)
+           : cuda::std::nullopt;
 }
 
 template <typename vertex_t, typename edge_t, typename return_type_t, bool multi_gpu, bool use_dcs>
@@ -490,7 +490,7 @@ class edge_partition_device_view_t<vertex_t, edge_t, multi_gpu, std::enable_if_t
     return major_value_start_offset_;
   }
 
-  __host__ __device__ thrust::optional<vertex_t> major_hypersparse_first() const noexcept
+  __host__ __device__ cuda::std::optional<vertex_t> major_hypersparse_first() const noexcept
   {
     return major_hypersparse_first_;
   }
@@ -528,15 +528,16 @@ class edge_partition_device_view_t<vertex_t, edge_t, multi_gpu, std::enable_if_t
     return major_range_first_ + major_offset;
   }
 
-  __device__ thrust::optional<vertex_t> major_idx_from_major_nocheck(vertex_t major) const noexcept
+  __device__ cuda::std::optional<vertex_t> major_idx_from_major_nocheck(
+    vertex_t major) const noexcept
   {
     if (major_hypersparse_first_ && (major >= *major_hypersparse_first_)) {
       auto major_hypersparse_idx =
         detail::major_hypersparse_idx_from_major_nocheck_impl(*dcs_nzd_vertices_, major);
       return major_hypersparse_idx
-               ? thrust::make_optional((*major_hypersparse_first_ - major_range_first_) +
-                                       *major_hypersparse_idx)
-               : thrust::nullopt;
+               ? cuda::std::make_optional((*major_hypersparse_first_ - major_range_first_) +
+                                          *major_hypersparse_idx)
+               : cuda::std::nullopt;
     } else {
       return major - major_range_first_;
     }
@@ -554,60 +555,60 @@ class edge_partition_device_view_t<vertex_t, edge_t, multi_gpu, std::enable_if_t
   }
 
   // major_hypersparse_idx: index within the hypersparse segment
-  __device__ thrust::optional<vertex_t> major_hypersparse_idx_from_major_nocheck(
+  __device__ cuda::std::optional<vertex_t> major_hypersparse_idx_from_major_nocheck(
     vertex_t major) const noexcept
   {
     if (dcs_nzd_vertices_) {
       return detail::major_hypersparse_idx_from_major_nocheck_impl(*dcs_nzd_vertices_, major);
     } else {
-      return thrust::nullopt;
+      return cuda::std::nullopt;
     }
   }
 
   // major_hypersparse_idx: index within the hypersparse segment
-  __device__ thrust::optional<vertex_t> major_from_major_hypersparse_idx_nocheck(
+  __device__ cuda::std::optional<vertex_t> major_from_major_hypersparse_idx_nocheck(
     vertex_t major_hypersparse_idx) const noexcept
   {
     return dcs_nzd_vertices_
-             ? thrust::optional<vertex_t>{(*dcs_nzd_vertices_)[major_hypersparse_idx]}
-             : thrust::nullopt;
+             ? cuda::std::optional<vertex_t>{(*dcs_nzd_vertices_)[major_hypersparse_idx]}
+             : cuda::std::nullopt;
   }
 
   __host__ __device__ vertex_t minor_from_minor_offset_nocheck(vertex_t minor_offset) const noexcept
   {
     return minor_range_first_ + minor_offset;
   }
 
-  // FIxME: better return thrust::optional<raft::device_span<vertex_t const>> for consistency (see
-  // dcs_nzd_range_bitmap())
-  __host__ __device__ thrust::optional<vertex_t const*> dcs_nzd_vertices() const
+  // FIxME: better return cuda::std::optional<raft::device_span<vertex_t const>> for consistency
+  // (see dcs_nzd_range_bitmap())
+  __host__ __device__ cuda::std::optional<vertex_t const*> dcs_nzd_vertices() const
   {
-    return dcs_nzd_vertices_ ? thrust::optional<vertex_t const*>{(*dcs_nzd_vertices_).data()}
-                             : thrust::nullopt;
+    return dcs_nzd_vertices_ ? cuda::std::optional<vertex_t const*>{(*dcs_nzd_vertices_).data()}
+                             : cuda::std::nullopt;
   }
 
-  __host__ __device__ thrust::optional<vertex_t> dcs_nzd_vertex_count() const
+  __host__ __device__ cuda::std::optional<vertex_t> dcs_nzd_vertex_count() const
   {
     return dcs_nzd_vertices_
-             ? thrust::optional<vertex_t>{static_cast<vertex_t>((*dcs_nzd_vertices_).size())}
-             : thrust::nullopt;
+             ? cuda::std::optional<vertex_t>{static_cast<vertex_t>((*dcs_nzd_vertices_).size())}
+             : cuda::std::nullopt;
   }
 
-  __host__ __device__ thrust::optional<raft::device_span<uint32_t const>> dcs_nzd_range_bitmap()
+  __host__ __device__ cuda::std::optional<raft::device_span<uint32_t const>> dcs_nzd_range_bitmap()
     const
   {
     return dcs_nzd_range_bitmap_
-             ? thrust::make_optional<raft::device_span<uint32_t const>>(
+             ? cuda::std::make_optional<raft::device_span<uint32_t const>>(
                  (*dcs_nzd_range_bitmap_).data(), (*dcs_nzd_range_bitmap_).size())
-             : thrust::nullopt;
+             : cuda::std::nullopt;
   }
 
  private:
   // should be trivially copyable to device
 
-  thrust::optional<raft::device_span<vertex_t const>> dcs_nzd_vertices_{thrust::nullopt};
-  thrust::optional<raft::device_span<uint32_t const>> dcs_nzd_range_bitmap_{thrust::nullopt};
-  thrust::optional<vertex_t> major_hypersparse_first_{thrust::nullopt};
+  cuda::std::optional<raft::device_span<vertex_t const>> dcs_nzd_vertices_{cuda::std::nullopt};
+  cuda::std::optional<raft::device_span<uint32_t const>> dcs_nzd_range_bitmap_{cuda::std::nullopt};
+  cuda::std::optional<vertex_t> major_hypersparse_first_{cuda::std::nullopt};
 
   vertex_t major_range_first_{0};
   vertex_t major_range_last_{0};
@@ -790,10 +791,10 @@ class edge_partition_device_view_t<vertex_t, edge_t, multi_gpu, std::enable_if_t
 
   __host__ __device__ vertex_t major_value_start_offset() const { return vertex_t{0}; }
 
-  __host__ __device__ thrust::optional<vertex_t> major_hypersparse_first() const noexcept
+  __host__ __device__ cuda::std::optional<vertex_t> major_hypersparse_first() const noexcept
   {
     assert(false);
-    return thrust::nullopt;
+    return cuda::std::nullopt;
   }
 
   __host__ __device__ constexpr vertex_t major_range_first() const noexcept { return vertex_t{0}; }
@@ -823,7 +824,8 @@ class edge_partition_device_view_t<vertex_t, edge_t, multi_gpu, std::enable_if_t
     return major_offset;
   }
 
-  __device__ thrust::optional<vertex_t> major_idx_from_major_nocheck(vertex_t major) const noexcept
+  __device__ cuda::std::optional<vertex_t> major_idx_from_major_nocheck(
+    vertex_t major) const noexcept
   {
     return major_offset_from_major_nocheck(major);
   }
@@ -834,34 +836,34 @@ class edge_partition_device_view_t<vertex_t, edge_t, multi_gpu, std::enable_if_t
   }
 
   // major_hypersparse_idx: index within the hypersparse segment
-  __device__ thrust::optional<vertex_t> major_hypersparse_idx_from_major_nocheck(
+  __device__ cuda::std::optional<vertex_t> major_hypersparse_idx_from_major_nocheck(
     vertex_t major) const noexcept
   {
     assert(false);
-    return thrust::nullopt;
+    return cuda::std::nullopt;
   }
 
   // major_hypersparse_idx: index within the hypersparse segment
-  __device__ thrust::optional<vertex_t> major_from_major_hypersparse_idx_nocheck(
+  __device__ cuda::std::optional<vertex_t> major_from_major_hypersparse_idx_nocheck(
     vertex_t major_hypersparse_idx) const noexcept
   {
     assert(false);
-    return thrust::nullopt;
+    return cuda::std::nullopt;
   }
 
   __host__ __device__ vertex_t minor_from_minor_offset_nocheck(vertex_t minor_offset) const noexcept
   {
     return minor_offset;
   }
 
-  __host__ __device__ thrust::optional<vertex_t const*> dcs_nzd_vertices() const
+  __host__ __device__ cuda::std::optional<vertex_t const*> dcs_nzd_vertices() const
   {
-    return thrust::nullopt;
+    return cuda::std::nullopt;
   }
 
-  __host__ __device__ thrust::optional<vertex_t> dcs_nzd_vertex_count() const
+  __host__ __device__ cuda::std::optional<vertex_t> dcs_nzd_vertex_count() const
   {
-    return thrust::nullopt;
+    return cuda::std::nullopt;
   }
 
  private:

diff --git a/cpp/include/cugraph/edge_partition_edge_property_device_view.cuh b/cpp/include/cugraph/edge_partition_edge_property_device_view.cuh
@@ -21,8 +21,8 @@
 #include <cugraph/utilities/packed_bool_utils.hpp>
 #include <cugraph/utilities/thrust_tuple_utils.hpp>
 
+#include <cuda/std/optional>
 #include <thrust/iterator/iterator_traits.h>
-#include <thrust/optional.h>
 
 namespace cugraph {
 
@@ -182,7 +182,7 @@ template <typename edge_t>
 class edge_partition_edge_dummy_property_device_view_t {
  public:
   using edge_type  = edge_t;
-  using value_type = thrust::nullopt_t;
+  using value_type = cuda::std::nullopt_t;
 
   static constexpr bool is_packed_bool          = false;
   static constexpr bool has_packed_bool_element = false;
@@ -194,7 +194,7 @@ class edge_partition_edge_dummy_property_device_view_t {
   {
   }
 
-  __device__ auto get(edge_t offset) const { return thrust::nullopt; }
+  __device__ auto get(edge_t offset) const { return cuda::std::nullopt; }
 };
 
 }  // namespace detail