Merge branch 'branch-23.12' into stream-nvtext-replace

ci/test_python_other.sh

@@ -27,7 +27,7 @@ pytest \
   --cov=dask_cudf \
   --cov-report=xml:"${RAPIDS_COVERAGE_DIR}/dask-cudf-coverage.xml" \
   --cov-report=term \
-  tests
+  .
 popd
 
 rapids-logger "pytest custreamz"

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

@@ -15,6 +15,7 @@ dependencies:
 - c-compiler
 - cachetools
 - cmake>=3.26.4
+- cramjam
 - cubinlinker
 - cuda-nvtx=11.8
 - cuda-python>=11.7.1,<12.0a0

conda/environments/all_cuda-120_arch-x86_64.yaml

@@ -15,6 +15,7 @@ dependencies:
 - c-compiler
 - cachetools
 - cmake>=3.26.4
+- cramjam
 - cuda-cudart-dev
 - cuda-gdb
 - cuda-nvcc

cpp/benchmarks/CMakeLists.txt

@@ -230,6 +230,7 @@ ConfigureNVBench(HASHING_NVBENCH hashing/hash.cpp)
 # ##################################################################################################
 # * merge benchmark -------------------------------------------------------------------------------
 ConfigureBench(MERGE_BENCH merge/merge.cpp)
+ConfigureNVBench(MERGE_NVBENCH merge/merge_structs.cpp merge/merge_lists.cpp)
 
 # ##################################################################################################
 # * null_mask benchmark ---------------------------------------------------------------------------

cpp/benchmarks/sort/nested_types_common.hpp → cpp/benchmarks/common/generate_nested_types.hpp

@@ -16,7 +16,7 @@
 
 #pragma once
 
-#include <benchmarks/common/generate_input.hpp>
+#include "generate_input.hpp"
 
 #include <cudf_test/column_wrapper.hpp>
 

cpp/benchmarks/merge/merge_lists.cpp

@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) 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.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <benchmarks/common/generate_nested_types.hpp>
+
+#include <cudf/detail/merge.hpp>
+#include <cudf/detail/sorting.hpp>
+
+#include <nvbench/nvbench.cuh>
+
+void nvbench_merge_list(nvbench::state& state)
+{
+  rmm::cuda_stream_view stream;
+
+  auto const input1 = create_lists_data(state);
+  auto const sorted_input1 =
+    cudf::detail::sort(*input1, {}, {}, stream, rmm::mr::get_current_device_resource());
+
+  auto const input2 = create_lists_data(state);
+  auto const sorted_input2 =
+    cudf::detail::sort(*input2, {}, {}, stream, rmm::mr::get_current_device_resource());
+
+  stream.synchronize();
+
+  state.exec(nvbench::exec_tag::sync, [&](nvbench::launch& launch) {
+    rmm::cuda_stream_view stream_view{launch.get_stream()};
+
+    cudf::detail::merge({*sorted_input1, *sorted_input2},
+                        {0},
+                        {cudf::order::ASCENDING},
+                        {},
+                        stream_view,
+                        rmm::mr::get_current_device_resource());
+  });
+}
+
+NVBENCH_BENCH(nvbench_merge_list)
+  .set_name("merge_lists")
+  .add_int64_power_of_two_axis("size_bytes", {10, 18, 24, 28})
+  .add_int64_axis("depth", {1, 4})
+  .add_float64_axis("null_frequency", {0, 0.2});

cpp/benchmarks/merge/merge_structs.cpp

@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) 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.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <benchmarks/common/generate_nested_types.hpp>
+
+#include <cudf/detail/merge.hpp>
+#include <cudf/detail/sorting.hpp>
+
+#include <nvbench/nvbench.cuh>
+
+void nvbench_merge_struct(nvbench::state& state)
+{
+  rmm::cuda_stream_view stream;
+
+  auto const input1 = create_structs_data(state);
+  auto const sorted_input1 =
+    cudf::detail::sort(*input1, {}, {}, stream, rmm::mr::get_current_device_resource());
+
+  auto const input2 = create_structs_data(state);
+  auto const sorted_input2 =
+    cudf::detail::sort(*input2, {}, {}, stream, rmm::mr::get_current_device_resource());
+
+  stream.synchronize();
+
+  state.exec(nvbench::exec_tag::sync, [&](nvbench::launch& launch) {
+    rmm::cuda_stream_view stream_view{launch.get_stream()};
+
+    cudf::detail::merge({*sorted_input1, *sorted_input2},
+                        {0},
+                        {cudf::order::ASCENDING},
+                        {},
+                        stream_view,
+                        rmm::mr::get_current_device_resource());
+  });
+}
+
+NVBENCH_BENCH(nvbench_merge_struct)
+  .set_name("merge_struct")
+  .add_int64_power_of_two_axis("NumRows", {10, 18, 26})
+  .add_int64_axis("Depth", {0, 1, 8})
+  .add_int64_axis("Nulls", {0, 1});

cpp/benchmarks/sort/rank_lists.cpp

@@ -14,9 +14,10 @@
  * limitations under the License.
  */
 
-#include "nested_types_common.hpp"
 #include "rank_types_common.hpp"
 
+#include <benchmarks/common/generate_nested_types.hpp>
+
 #include <cudf/sorting.hpp>
 
 #include <cudf_test/column_utilities.hpp>

cpp/benchmarks/sort/rank_structs.cpp

@@ -14,8 +14,8 @@
  * limitations under the License.
  */
 
-#include "nested_types_common.hpp"
 #include "rank_types_common.hpp"
+#include <benchmarks/common/generate_nested_types.hpp>
 
 #include <cudf/sorting.hpp>
 

cpp/benchmarks/sort/sort_lists.cpp

@@ -14,7 +14,7 @@
  * limitations under the License.
  */
 
-#include "nested_types_common.hpp"
+#include <benchmarks/common/generate_nested_types.hpp>
 
 #include <cudf/detail/sorting.hpp>
 

cpp/benchmarks/sort/sort_structs.cpp

@@ -14,7 +14,7 @@
  * limitations under the License.
  */
 
-#include "nested_types_common.hpp"
+#include <benchmarks/common/generate_nested_types.hpp>
 
 #include <cudf/detail/sorting.hpp>
 

cpp/include/cudf/detail/interop.hpp

@@ -194,5 +194,18 @@ std::unique_ptr<table> from_arrow(arrow::Table const& input_table,
 std::unique_ptr<cudf::scalar> from_arrow(arrow::Scalar const& input,
                                          rmm::cuda_stream_view stream,
                                          rmm::mr::device_memory_resource* mr);
+
+/**
+ * @brief Return a maximum precision for a given type.
+ *
+ * @tparam T the type to get the maximum precision for
+ */
+template <typename T>
+constexpr std::size_t max_precision()
+{
+  auto constexpr num_bits = sizeof(T) * 8;
+  return std::floor(num_bits * std::log(2) / std::log(10));
+}
+
 }  // namespace detail
 }  // namespace cudf