Merge branch 'branch-24.06' into stdlib-migration

rapidsai · May 3, 2024 · 21fb08c · 21fb08c
2 parents 6feac65 + 7d7043e
commit 21fb08c
Show file tree

Hide file tree

Showing 12 changed files with 1,268 additions and 85 deletions.
diff --git a/cpp/src/graph_ops/append_unique_func.cuh b/cpp/src/graph_ops/append_unique_func.cuh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2023, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2024, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -316,7 +316,7 @@ void graph_append_unique_func(void* target_nodes_ptr,
     <<<num_blocks, kAssignThreadBlockSize, 0, stream>>>(value_id, bucket_count_ptr);
   WM_CUDA_CHECK(cudaGetLastError());
   wholememory_ops::wm_thrust_allocator thrust_allocator(p_env_fns);
-  thrust::exclusive_scan(thrust::cuda::par(thrust_allocator).on(stream),
+  thrust::exclusive_scan(thrust::cuda::par_nosync(thrust_allocator).on(stream),
                          bucket_count_ptr,
                          bucket_count_ptr + num_bucket_count,
                          (int*)bucket_prefix_sum_ptr);

diff --git a/cpp/src/wholegraph_ops/unweighted_sample_without_replacement_func.cuh b/cpp/src/wholegraph_ops/unweighted_sample_without_replacement_func.cuh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2023, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2024, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -337,7 +337,7 @@ void wholegraph_csr_unweighted_sample_without_replacement_func(
 
   // prefix sum
   wholememory_ops::wm_thrust_allocator thrust_allocator(p_env_fns);
-  thrust::exclusive_scan(thrust::cuda::par(thrust_allocator).on(stream),
+  thrust::exclusive_scan(thrust::cuda::par_nosync(thrust_allocator).on(stream),
                          tmp_sample_count_mem_pointer,
                          tmp_sample_count_mem_pointer + center_node_count + 1,
                          (int*)output_sample_offset);

diff --git a/cpp/src/wholegraph_ops/weighted_sample_without_replacement_func.cuh b/cpp/src/wholegraph_ops/weighted_sample_without_replacement_func.cuh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2023, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2024, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -462,7 +462,7 @@ void wholegraph_csr_weighted_sample_without_replacement_func(
 
   // prefix sum
   wholememory_ops::wm_thrust_allocator thrust_allocator(p_env_fns);
-  thrust::exclusive_scan(thrust::cuda::par(thrust_allocator).on(stream),
+  thrust::exclusive_scan(thrust::cuda::par_nosync(thrust_allocator).on(stream),
                          tmp_sample_count_mem_pointer,
                          tmp_sample_count_mem_pointer + center_node_count + 1,
                          static_cast<int*>(output_sample_offset));
@@ -500,7 +500,7 @@ void wholegraph_csr_weighted_sample_without_replacement_func(
   raft::random::detail::DeviceState<raft::random::detail::PCGenerator> rngstate(_rngstate);
   if (max_sample_count > sample_count_threshold) {
     wholememory_ops::wm_thrust_allocator tmp_thrust_allocator(p_env_fns);
-    thrust::exclusive_scan(thrust::cuda::par(tmp_thrust_allocator).on(stream),
+    thrust::exclusive_scan(thrust::cuda::par_nosync(tmp_thrust_allocator).on(stream),
                            tmp_neighbor_counts_mem_pointer,
                            tmp_neighbor_counts_mem_pointer + center_node_count + 1,
                            tmp_neighbor_counts_mem_pointer);