Remove default detail mrs: part6 (#12969)

This is the sixth PR in a sequence removing default mr parameters in detail APIs. Contributes to #12944.

Authors:
  - Vyas Ramasubramani (https://github.com/vyasr)

Approvers:
  - David Wendt (https://github.com/davidwendt)
  - Yunsong Wang (https://github.com/PointKernel)
  - Jason Lowe (https://github.com/jlowe)

URL: #12969

cpp/benchmarks/iterator/iterator.cu

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -140,8 +140,8 @@ void BM_iterator(benchmark::State& state)
   cudf::column_view hasnull_F = wrap_hasnull_F;
 
   // Initialize dev_result to false
-  auto dev_result =
-    cudf::detail::make_zeroed_device_uvector_sync<TypeParam>(1, cudf::get_default_stream());
+  auto dev_result = cudf::detail::make_zeroed_device_uvector_sync<TypeParam>(
+    1, cudf::get_default_stream(), rmm::mr::get_current_device_resource());
   for (auto _ : state) {
     cuda_event_timer raii(state, true);  // flush_l2_cache = true, stream = 0
     if (cub_or_thrust) {
@@ -210,7 +210,7 @@ void BM_pair_iterator(benchmark::State& state)
 
   // Initialize dev_result to false
   auto dev_result = cudf::detail::make_zeroed_device_uvector_sync<thrust::pair<T, bool>>(
-    1, cudf::get_default_stream());
+    1, cudf::get_default_stream(), rmm::mr::get_current_device_resource());
   for (auto _ : state) {
     cuda_event_timer raii(state, true);  // flush_l2_cache = true, stream = 0
     if (cub_or_thrust) {

cpp/include/cudf/detail/gather.cuh

@@ -583,10 +583,12 @@ void gather_bitmask(table_view const& source,
   std::transform(target.begin(), target.end(), target_masks.begin(), [](auto const& col) {
     return col->mutable_view().null_mask();
   });
-  auto d_target_masks = make_device_uvector_async(target_masks, stream);
+  auto d_target_masks =
+    make_device_uvector_async(target_masks, stream, rmm::mr::get_current_device_resource());
 
   auto const device_source = table_device_view::create(source, stream);
-  auto d_valid_counts      = make_zeroed_device_uvector_async<size_type>(target.size(), stream);
+  auto d_valid_counts      = make_zeroed_device_uvector_async<size_type>(
+    target.size(), stream, rmm::mr::get_current_device_resource());
 
   // Dispatch operation enum to get implementation
   auto const impl = [op]() {

cpp/include/cudf/detail/null_mask.cuh

@@ -426,7 +426,8 @@ std::vector<size_type> segmented_count_bits(bitmask_type const* bitmask,
 
   // Construct a contiguous host buffer of indices and copy to device.
   auto const h_indices = std::vector<size_type>(indices_begin, indices_end);
-  auto const d_indices = make_device_uvector_async(h_indices, stream);
+  auto const d_indices =
+    make_device_uvector_async(h_indices, stream, rmm::mr::get_current_device_resource());
 
   // Compute the bit counts over each segment.
   auto first_bit_indices_begin = thrust::make_transform_iterator(

cpp/include/cudf/detail/utilities/vector_factories.hpp

@@ -48,10 +48,9 @@ namespace detail {
  * @return A device_uvector containing zeros
  */
 template <typename T>
-rmm::device_uvector<T> make_zeroed_device_uvector_async(
-  std::size_t size,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+rmm::device_uvector<T> make_zeroed_device_uvector_async(std::size_t size,
+                                                        rmm::cuda_stream_view stream,
+                                                        rmm::mr::device_memory_resource* mr)
 {
   rmm::device_uvector<T> ret(size, stream, mr);
   CUDF_CUDA_TRY(cudaMemsetAsync(ret.data(), 0, size * sizeof(T), stream.value()));
@@ -70,10 +69,9 @@ rmm::device_uvector<T> make_zeroed_device_uvector_async(
  * @return A device_uvector containing zeros
  */
 template <typename T>
-rmm::device_uvector<T> make_zeroed_device_uvector_sync(
-  std::size_t size,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+rmm::device_uvector<T> make_zeroed_device_uvector_sync(std::size_t size,
+                                                       rmm::cuda_stream_view stream,
+                                                       rmm::mr::device_memory_resource* mr)
 {
   rmm::device_uvector<T> ret(size, stream, mr);
   CUDF_CUDA_TRY(cudaMemsetAsync(ret.data(), 0, size * sizeof(T), stream.value()));
@@ -94,10 +92,9 @@ rmm::device_uvector<T> make_zeroed_device_uvector_sync(
  * @return A device_uvector containing the copied data
  */
 template <typename T>
-rmm::device_uvector<T> make_device_uvector_async(
-  host_span<T const> source_data,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+rmm::device_uvector<T> make_device_uvector_async(host_span<T const> source_data,
+                                                 rmm::cuda_stream_view stream,
+                                                 rmm::mr::device_memory_resource* mr)
 {
   rmm::device_uvector<T> ret(source_data.size(), stream, mr);
   CUDF_CUDA_TRY(cudaMemcpyAsync(ret.data(),
@@ -126,9 +123,7 @@ template <
   std::enable_if_t<
     std::is_convertible_v<Container, host_span<typename Container::value_type const>>>* = nullptr>
 rmm::device_uvector<typename Container::value_type> make_device_uvector_async(
-  Container const& c,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+  Container const& c, rmm::cuda_stream_view stream, rmm::mr::device_memory_resource* mr)
 {
   return make_device_uvector_async(host_span<typename Container::value_type const>{c}, stream, mr);
 }
@@ -146,10 +141,9 @@ rmm::device_uvector<typename Container::value_type> make_device_uvector_async(
  * @return A device_uvector containing the copied data
  */
 template <typename T>
-rmm::device_uvector<T> make_device_uvector_async(
-  device_span<T const> source_data,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+rmm::device_uvector<T> make_device_uvector_async(device_span<T const> source_data,
+                                                 rmm::cuda_stream_view stream,
+                                                 rmm::mr::device_memory_resource* mr)
 {
   rmm::device_uvector<T> ret(source_data.size(), stream, mr);
   CUDF_CUDA_TRY(cudaMemcpyAsync(ret.data(),
@@ -178,9 +172,7 @@ template <
   std::enable_if_t<
     std::is_convertible_v<Container, device_span<typename Container::value_type const>>>* = nullptr>
 rmm::device_uvector<typename Container::value_type> make_device_uvector_async(
-  Container const& c,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+  Container const& c, rmm::cuda_stream_view stream, rmm::mr::device_memory_resource* mr)
 {
   return make_device_uvector_async(
     device_span<typename Container::value_type const>{c}, stream, mr);
@@ -199,10 +191,9 @@ rmm::device_uvector<typename Container::value_type> make_device_uvector_async(
  * @return A device_uvector containing the copied data
  */
 template <typename T>
-rmm::device_uvector<T> make_device_uvector_sync(
-  host_span<T const> source_data,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+rmm::device_uvector<T> make_device_uvector_sync(host_span<T const> source_data,
+                                                rmm::cuda_stream_view stream,
+                                                rmm::mr::device_memory_resource* mr)
 {
   auto ret = make_device_uvector_async(source_data, stream, mr);
   stream.synchronize();
@@ -227,9 +218,7 @@ template <
   std::enable_if_t<
     std::is_convertible_v<Container, host_span<typename Container::value_type const>>>* = nullptr>
 rmm::device_uvector<typename Container::value_type> make_device_uvector_sync(
-  Container const& c,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+  Container const& c, rmm::cuda_stream_view stream, rmm::mr::device_memory_resource* mr)
 {
   return make_device_uvector_sync(host_span<typename Container::value_type const>{c}, stream, mr);
 }
@@ -247,10 +236,9 @@ rmm::device_uvector<typename Container::value_type> make_device_uvector_sync(
  * @return A device_uvector containing the copied data
  */
 template <typename T>
-rmm::device_uvector<T> make_device_uvector_sync(
-  device_span<T const> source_data,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+rmm::device_uvector<T> make_device_uvector_sync(device_span<T const> source_data,
+                                                rmm::cuda_stream_view stream,
+                                                rmm::mr::device_memory_resource* mr)
 {
   auto ret = make_device_uvector_async(source_data, stream, mr);
   stream.synchronize();
@@ -275,9 +263,7 @@ template <
   std::enable_if_t<
     std::is_convertible_v<Container, device_span<typename Container::value_type const>>>* = nullptr>
 rmm::device_uvector<typename Container::value_type> make_device_uvector_sync(
-  Container const& c,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource())
+  Container const& c, rmm::cuda_stream_view stream, rmm::mr::device_memory_resource* mr)
 {
   return make_device_uvector_sync(device_span<typename Container::value_type const>{c}, stream, mr);
 }

cpp/include/cudf/lists/lists_column_factories.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021-2022, NVIDIA CORPORATION.
+ * Copyright (c) 2021-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -35,11 +35,10 @@ namespace detail {
  * @param[in] stream CUDA stream used for device memory operations and kernel launches.
  * @param[in] mr Device memory resource used to allocate the returned column's device memory.
  */
-std::unique_ptr<cudf::column> make_lists_column_from_scalar(
-  list_scalar const& value,
-  size_type size,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<cudf::column> make_lists_column_from_scalar(list_scalar const& value,
+                                                            size_type size,
+                                                            rmm::cuda_stream_view stream,
+                                                            rmm::mr::device_memory_resource* mr);
 
 }  // namespace detail
 }  // namespace lists

cpp/include/cudf/structs/detail/concatenate.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020-2021, NVIDIA CORPORATION.
+ * Copyright (c) 2020-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -48,10 +48,9 @@ namespace detail {
  * @param mr      Device memory resource used to allocate the returned column's device memory.
  * @return        New column with concatenated results.
  */
-std::unique_ptr<column> concatenate(
-  host_span<column_view const> columns,
-  rmm::cuda_stream_view stream,
-  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+std::unique_ptr<column> concatenate(host_span<column_view const> columns,
+                                    rmm::cuda_stream_view stream,
+                                    rmm::mr::device_memory_resource* mr);
 
 }  // namespace detail
 }  // namespace structs

cpp/include/cudf_test/column_wrapper.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -732,9 +732,11 @@ class strings_column_wrapper : public detail::column_wrapper {
   {
     auto all_valid        = thrust::make_constant_iterator(true);
     auto [chars, offsets] = detail::make_chars_and_offsets(begin, end, all_valid);
-    auto d_chars   = cudf::detail::make_device_uvector_sync(chars, cudf::get_default_stream());
-    auto d_offsets = cudf::detail::make_device_uvector_sync(offsets, cudf::get_default_stream());
-    wrapped        = cudf::make_strings_column(d_chars, d_offsets);
+    auto d_chars          = cudf::detail::make_device_uvector_sync(
+      chars, cudf::get_default_stream(), rmm::mr::get_current_device_resource());
+    auto d_offsets = cudf::detail::make_device_uvector_sync(
+      offsets, cudf::get_default_stream(), rmm::mr::get_current_device_resource());
+    wrapped = cudf::make_strings_column(d_chars, d_offsets);
   }
 
   /**
@@ -772,10 +774,13 @@ class strings_column_wrapper : public detail::column_wrapper {
     size_type num_strings = std::distance(begin, end);
     auto [chars, offsets] = detail::make_chars_and_offsets(begin, end, v);
     auto null_mask        = detail::make_null_mask_vector(v, v + num_strings);
-    auto d_chars   = cudf::detail::make_device_uvector_sync(chars, cudf::get_default_stream());
-    auto d_offsets = cudf::detail::make_device_uvector_sync(offsets, cudf::get_default_stream());
-    auto d_bitmask = cudf::detail::make_device_uvector_sync(null_mask, cudf::get_default_stream());
-    wrapped        = cudf::make_strings_column(d_chars, d_offsets, d_bitmask);
+    auto d_chars          = cudf::detail::make_device_uvector_sync(
+      chars, cudf::get_default_stream(), rmm::mr::get_current_device_resource());
+    auto d_offsets = cudf::detail::make_device_uvector_sync(
+      offsets, cudf::get_default_stream(), rmm::mr::get_current_device_resource());
+    auto d_bitmask = cudf::detail::make_device_uvector_sync(
+      null_mask, cudf::get_default_stream(), rmm::mr::get_current_device_resource());
+    wrapped = cudf::make_strings_column(d_chars, d_offsets, d_bitmask);
   }
 
   /**

cpp/include/cudf_test/tdigest_utilities.cuh

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2022, NVIDIA CORPORATION.
+ * Copyright (c) 2022-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -168,7 +168,8 @@ void tdigest_minmax_compare(cudf::tdigest::tdigest_column_view const& tdv,
   // verify min/max
   thrust::host_vector<device_span<T const>> h_spans;
   h_spans.push_back({input_values.begin<T>(), static_cast<size_t>(input_values.size())});
-  auto spans = cudf::detail::make_device_uvector_async(h_spans, cudf::get_default_stream());
+  auto spans = cudf::detail::make_device_uvector_async(
+    h_spans, cudf::get_default_stream(), rmm::mr::get_current_device_resource());
 
   auto expected_min = cudf::make_fixed_width_column(
     data_type{type_id::FLOAT64}, spans.size(), mask_state::UNALLOCATED);

cpp/src/copying/concatenate.cu

@@ -76,7 +76,8 @@ auto create_device_views(host_span<column_view const> views, rmm::cuda_stream_vi
                  std::back_inserter(device_views),
                  [](auto const& col) { return *col; });
 
-  auto d_views = make_device_uvector_async(device_views, stream);
+  auto d_views =
+    make_device_uvector_async(device_views, stream, rmm::mr::get_current_device_resource());
 
   // Compute the partition offsets
   auto offsets = thrust::host_vector<size_t>(views.size() + 1);
@@ -87,7 +88,8 @@ auto create_device_views(host_span<column_view const> views, rmm::cuda_stream_vi
     std::next(offsets.begin()),
     [](auto const& col) { return col.size(); },
     thrust::plus{});
-  auto d_offsets         = make_device_uvector_async(offsets, stream);
+  auto d_offsets =
+    make_device_uvector_async(offsets, stream, rmm::mr::get_current_device_resource());
   auto const output_size = offsets.back();
 
   return std::make_tuple(

cpp/src/dictionary/detail/concatenate.cu

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020-2022, NVIDIA CORPORATION.
+ * Copyright (c) 2020-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -114,7 +114,8 @@ struct compute_children_offsets_fn {
       [](auto lhs, auto rhs) {
         return offsets_pair{lhs.first + rhs.first, lhs.second + rhs.second};
       });
-    return cudf::detail::make_device_uvector_sync(offsets, stream);
+    return cudf::detail::make_device_uvector_sync(
+      offsets, stream, rmm::mr::get_current_device_resource());
   }
 
  private:

cpp/src/groupby/hash/groupby.cu

@@ -481,7 +481,8 @@ void compute_single_pass_aggs(table_view const& keys,
   // prepare to launch kernel to do the actual aggregation
   auto d_sparse_table = mutable_table_device_view::create(sparse_table, stream);
   auto d_values       = table_device_view::create(flattened_values, stream);
-  auto const d_aggs   = cudf::detail::make_device_uvector_async(agg_kinds, stream);
+  auto const d_aggs   = cudf::detail::make_device_uvector_async(
+    agg_kinds, stream, rmm::mr::get_current_device_resource());
   auto const skip_key_rows_with_nulls =
     keys_have_nulls and include_null_keys == null_policy::EXCLUDE;
 

cpp/src/groupby/sort/group_quantiles.cu

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -156,7 +156,8 @@ std::unique_ptr<column> group_quantiles(column_view const& values,
                                         rmm::cuda_stream_view stream,
                                         rmm::mr::device_memory_resource* mr)
 {
-  auto dv_quantiles = cudf::detail::make_device_uvector_async(quantiles, stream);
+  auto dv_quantiles = cudf::detail::make_device_uvector_async(
+    quantiles, stream, rmm::mr::get_current_device_resource());
 
   auto values_type = cudf::is_dictionary(values.type())
                        ? dictionary_column_view(values).keys().type()

cpp/src/hash/unordered_multiset.cuh

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2019-2023, NVIDIA CORPORATION.  All rights reserved.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -84,10 +84,10 @@ class unordered_multiset {
     auto d_column = column_device_view::create(col, stream);
     auto d_col    = *d_column;
 
-    auto hash_bins_start =
-      cudf::detail::make_zeroed_device_uvector_async<size_type>(2 * d_col.size() + 1, stream);
-    auto hash_bins_end =
-      cudf::detail::make_zeroed_device_uvector_async<size_type>(2 * d_col.size() + 1, stream);
+    auto hash_bins_start = cudf::detail::make_zeroed_device_uvector_async<size_type>(
+      2 * d_col.size() + 1, stream, rmm::mr::get_current_device_resource());
+    auto hash_bins_end = cudf::detail::make_zeroed_device_uvector_async<size_type>(
+      2 * d_col.size() + 1, stream, rmm::mr::get_current_device_resource());
     auto hash_data = rmm::device_uvector<Element>(d_col.size(), stream);
 
     Hasher hasher;

cpp/src/io/avro/reader_impl.cu

@@ -444,7 +444,8 @@ std::vector<column_buffer> decode_data(metadata& meta,
     }
   }
 
-  auto block_list = cudf::detail::make_device_uvector_async(meta.block_list, stream);
+  auto block_list = cudf::detail::make_device_uvector_async(
+    meta.block_list, stream, rmm::mr::get_current_device_resource());
 
   schema_desc.host_to_device(stream);
 
@@ -574,8 +575,10 @@ table_with_metadata read_avro(std::unique_ptr<cudf::io::datasource>&& source,
           }
         }
 
-        d_global_dict      = cudf::detail::make_device_uvector_async(h_global_dict, stream);
-        d_global_dict_data = cudf::detail::make_device_uvector_async(h_global_dict_data, stream);
+        d_global_dict = cudf::detail::make_device_uvector_async(
+          h_global_dict, stream, rmm::mr::get_current_device_resource());
+        d_global_dict_data = cudf::detail::make_device_uvector_async(
+          h_global_dict_data, stream, rmm::mr::get_current_device_resource());
 
         stream.synchronize();
       }