Merge branch 'branch-25.02' into enh-json_pylibcudf_all_features

ci/cudf_pandas_scripts/third-party-integration/test.sh

@@ -26,6 +26,8 @@ main() {
     LIBS=${LIBS#[}
     LIBS=${LIBS%]}
 
+    ANY_FAILURES=0
+
     for lib in ${LIBS//,/ }; do
         lib=$(echo "$lib" | tr -d '""')
         echo "Running tests for library $lib"
@@ -56,10 +58,6 @@ main() {
         rapids-logger "Check GPU usage"
         nvidia-smi
 
-        EXITCODE=0
-        trap "EXITCODE=1" ERR
-        set +e
-
         rapids-logger "pytest ${lib}"
 
         NUM_PROCESSES=8
@@ -72,12 +70,20 @@ main() {
             fi
         done
 
+        EXITCODE=0
+        trap "EXITCODE=1" ERR
+        set +e
+
         TEST_DIR=${TEST_DIR} NUM_PROCESSES=${NUM_PROCESSES} ci/cudf_pandas_scripts/third-party-integration/run-library-tests.sh ${lib}
 
+        set -e
         rapids-logger "Test script exiting with value: ${EXITCODE}"
+        if [[ ${EXITCODE} != 0 ]]; then
+            ANY_FAILURES=1
+        fi
     done
 
-    exit ${EXITCODE}
+    exit ${ANY_FAILURES}
 }
 
 main "$@"

cpp/CMakeLists.txt

@@ -1105,7 +1105,7 @@ if(CUDF_BUILD_STREAMS_TEST_UTIL)
       ${_tgt} PRIVATE "$<BUILD_INTERFACE:$<$<COMPILE_LANGUAGE:CXX>:${CUDF_CXX_FLAGS}>>"
     )
     target_include_directories(${_tgt} PRIVATE "$<BUILD_INTERFACE:${CUDF_SOURCE_DIR}/include>")
-    target_link_libraries(${_tgt} PUBLIC CUDA::cudart rmm::rmm)
+    target_link_libraries(${_tgt} PUBLIC CUDA::cudart rmm::rmm rmm::rmm_logger rmm::rmm_logger_impl)
     if(CUDF_BUILD_STACKTRACE_DEBUG)
       target_link_libraries(${_tgt} PRIVATE cudf_backtrace)
     endif()

cpp/benchmarks/CMakeLists.txt

@@ -140,8 +140,9 @@ function(ConfigureNVBench CMAKE_BENCH_NAME)
 endfunction()
 
 # ##################################################################################################
-# * column benchmarks -----------------------------------------------------------------------------
-ConfigureBench(COLUMN_CONCAT_BENCH column/concatenate.cpp)
+# * copying benchmarks
+# -----------------------------------------------------------------------------
+ConfigureNVBench(COPYING_NVBENCH copying/concatenate.cpp)
 
 # ##################################################################################################
 # * gather benchmark ------------------------------------------------------------------------------
@@ -351,11 +352,18 @@ ConfigureNVBench(BINARYOP_NVBENCH binaryop/binaryop.cpp binaryop/compiled_binary
 
 # ##################################################################################################
 # * nvtext benchmark -------------------------------------------------------------------
-ConfigureBench(TEXT_BENCH text/subword.cpp)
-
 ConfigureNVBench(
-  TEXT_NVBENCH text/edit_distance.cpp text/hash_ngrams.cpp text/jaccard.cpp text/minhash.cpp
-  text/ngrams.cpp text/normalize.cpp text/replace.cpp text/tokenize.cpp text/vocab.cpp
+  TEXT_NVBENCH
+  text/edit_distance.cpp
+  text/hash_ngrams.cpp
+  text/jaccard.cpp
+  text/minhash.cpp
+  text/ngrams.cpp
+  text/normalize.cpp
+  text/replace.cpp
+  text/subword.cpp
+  text/tokenize.cpp
+  text/vocab.cpp
 )
 
 # ##################################################################################################

cpp/benchmarks/copying/concatenate.cpp

@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) 2020-2024, NVIDIA CORPORATION.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * 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_input.hpp>
+
+#include <cudf/column/column_view.hpp>
+#include <cudf/concatenate.hpp>
+#include <cudf/strings/strings_column_view.hpp>
+#include <cudf/utilities/default_stream.hpp>
+
+#include <nvbench/nvbench.cuh>
+
+#include <vector>
+
+static void bench_concatenate(nvbench::state& state)
+{
+  auto const num_rows = static_cast<cudf::size_type>(state.get_int64("num_rows"));
+  auto const num_cols = static_cast<cudf::size_type>(state.get_int64("num_cols"));
+  auto const nulls    = static_cast<cudf::size_type>(state.get_float64("nulls"));
+
+  auto input = create_sequence_table(
+    cycle_dtypes({cudf::type_to_id<int64_t>()}, num_cols), row_count{num_rows}, nulls);
+  auto input_columns = input->view();
+  auto column_views  = std::vector<cudf::column_view>(input_columns.begin(), input_columns.end());
+
+  auto stream = cudf::get_default_stream();
+  state.set_cuda_stream(nvbench::make_cuda_stream_view(stream.value()));
+  state.add_global_memory_reads<int64_t>(num_rows * num_cols);
+  state.add_global_memory_writes<int64_t>(num_rows * num_cols);
+
+  state.exec(nvbench::exec_tag::sync,
+             [&](nvbench::launch&) { auto result = cudf::concatenate(column_views); });
+}
+
+NVBENCH_BENCH(bench_concatenate)
+  .set_name("concatenate")
+  .add_int64_axis("num_rows", {64, 512, 4096, 32768, 262144})
+  .add_int64_axis("num_cols", {2, 8, 64, 512, 1024})
+  .add_float64_axis("nulls", {0.0, 0.3});
+
+static void bench_concatenate_strings(nvbench::state& state)
+{
+  auto const num_rows  = static_cast<cudf::size_type>(state.get_int64("num_rows"));
+  auto const num_cols  = static_cast<cudf::size_type>(state.get_int64("num_cols"));
+  auto const row_width = static_cast<cudf::size_type>(state.get_int64("row_width"));
+  auto const nulls     = static_cast<cudf::size_type>(state.get_float64("nulls"));
+
+  data_profile const profile =
+    data_profile_builder()
+      .distribution(cudf::type_id::STRING, distribution_id::NORMAL, 0, row_width)
+      .null_probability(nulls);
+  auto const column = create_random_column(cudf::type_id::STRING, row_count{num_rows}, profile);
+  auto const input  = column->view();
+
+  auto column_views = std::vector<cudf::column_view>(num_cols, input);
+
+  auto stream = cudf::get_default_stream();
+  state.set_cuda_stream(nvbench::make_cuda_stream_view(stream.value()));
+  auto const sv = cudf::strings_column_view(input);
+  state.add_global_memory_reads<int8_t>(sv.chars_size(stream) * num_cols);
+  state.add_global_memory_writes<int64_t>(sv.chars_size(stream) * num_cols);
+
+  state.exec(nvbench::exec_tag::sync,
+             [&](nvbench::launch&) { auto result = cudf::concatenate(column_views); });
+}
+
+NVBENCH_BENCH(bench_concatenate_strings)
+  .set_name("concatenate_strings")
+  .add_int64_axis("num_rows", {256, 512, 4096, 16384})
+  .add_int64_axis("num_cols", {2, 8, 64, 256})
+  .add_int64_axis("row_width", {32, 128})
+  .add_float64_axis("nulls", {0.0, 0.3});

cpp/benchmarks/string/case.cpp

@@ -24,18 +24,14 @@
 
 void bench_case(nvbench::state& state)
 {
-  auto const n_rows    = static_cast<cudf::size_type>(state.get_int64("num_rows"));
-  auto const max_width = static_cast<int32_t>(state.get_int64("row_width"));
+  auto const num_rows  = static_cast<cudf::size_type>(state.get_int64("num_rows"));
+  auto const min_width = static_cast<cudf::size_type>(state.get_int64("min_width"));
+  auto const max_width = static_cast<cudf::size_type>(state.get_int64("max_width"));
   auto const encoding  = state.get_string("encoding");
 
-  if (static_cast<std::size_t>(n_rows) * static_cast<std::size_t>(max_width) >=
-      static_cast<std::size_t>(std::numeric_limits<cudf::size_type>::max())) {
-    state.skip("Skip benchmarks greater than size_type limit");
-  }
-
   data_profile const profile = data_profile_builder().distribution(
-    cudf::type_id::STRING, distribution_id::NORMAL, 0, max_width);
-  auto const column = create_random_column(cudf::type_id::STRING, row_count{n_rows}, profile);
+    cudf::type_id::STRING, distribution_id::NORMAL, min_width, max_width);
+  auto const column = create_random_column(cudf::type_id::STRING, row_count{num_rows}, profile);
 
   auto col_view = column->view();
 
@@ -74,6 +70,7 @@ void bench_case(nvbench::state& state)
 
 NVBENCH_BENCH(bench_case)
   .set_name("case")
-  .add_int64_axis("row_width", {32, 64, 128, 256, 512, 1024, 2048})
-  .add_int64_axis("num_rows", {32768, 262144, 2097152, 16777216})
+  .add_int64_axis("min_width", {0})
+  .add_int64_axis("max_width", {32, 64, 128, 256})
+  .add_int64_axis("num_rows", {32768, 262144, 2097152})
   .add_string_axis("encoding", {"ascii", "utf8"});