Merge branch 'branch-25.02' into pylibcudf-aggregation-serialization

.github/workflows/pr.yaml

@@ -13,6 +13,7 @@ jobs:
   # Please keep pr-builder as the top job here
   pr-builder:
     needs:
+      - check-nightly-ci
       - changed-files
       - checks
       - conda-cpp-build
@@ -54,6 +55,18 @@ jobs:
       - name: Telemetry setup
         if: ${{ vars.TELEMETRY_ENABLED == 'true' }}
         uses: rapidsai/shared-actions/telemetry-dispatch-stash-base-env-vars@main
+  check-nightly-ci:
+    # Switch to ubuntu-latest once it defaults to a version of Ubuntu that
+    # provides at least Python 3.11 (see
+    # https://docs.python.org/3/library/datetime.html#datetime.date.fromisoformat)
+    runs-on: ubuntu-24.04
+    env:
+      RAPIDS_GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+    steps:
+      - name: Check if nightly CI is passing
+        uses: rapidsai/shared-actions/check_nightly_success/dispatch@main
+        with:
+          repo: cudf
   changed-files:
     secrets: inherit
     needs: telemetry-setup
@@ -328,16 +341,11 @@ jobs:
         run_script: "ci/cudf_pandas_scripts/pandas-tests/diff.sh"
 
   telemetry-summarize:
-    runs-on: ubuntu-latest
+    # This job must use a self-hosted runner to record telemetry traces.
+    runs-on: linux-amd64-cpu4
     needs: pr-builder
     if: ${{ vars.TELEMETRY_ENABLED == 'true' && !cancelled() }}
     continue-on-error: true
     steps:
-      - name: Load stashed telemetry env vars
-        uses: rapidsai/shared-actions/telemetry-dispatch-load-base-env-vars@main
-        with:
-            load_service_name: true
       - name: Telemetry summarize
-        uses: rapidsai/shared-actions/telemetry-dispatch-write-summary@main
-        with:
-          cert_concat: "${{ secrets.OTEL_EXPORTER_OTLP_CA_CERTIFICATE }};${{ secrets.OTEL_EXPORTER_OTLP_CLIENT_CERTIFICATE }};${{ secrets.OTEL_EXPORTER_OTLP_CLIENT_KEY }}"
+        uses: rapidsai/shared-actions/telemetry-dispatch-summarize@main

ci/build_docs.sh

@@ -35,6 +35,10 @@ rapids-mamba-retry install \
 
 export RAPIDS_DOCS_DIR="$(mktemp -d)"
 
+EXITCODE=0
+trap "EXITCODE=1" ERR
+set +e
+
 rapids-logger "Build CPP docs"
 pushd cpp/doxygen
 aws s3 cp s3://rapidsai-docs/librmm/html/${RAPIDS_VERSION_MAJOR_MINOR}/rmm.tag . || echo "Failed to download rmm Doxygen tag"
@@ -58,3 +62,5 @@ mv build/dirhtml/* "${RAPIDS_DOCS_DIR}/dask-cudf/html"
 popd
 
 RAPIDS_VERSION_NUMBER="${RAPIDS_VERSION_MAJOR_MINOR}" rapids-upload-docs
+
+exit ${EXITCODE}

cpp/CMakeLists.txt

@@ -276,7 +276,7 @@ rapids_cpm_init()
 
 # Not using rapids-cmake since we never want to find, always download.
 CPMAddPackage(
-  NAME rapids_logger GITHUB_REPOSITORY rapidsai/rapids-logger GIT_SHALLOW TRUE GIT_TAG
+  NAME rapids_logger GITHUB_REPOSITORY rapidsai/rapids-logger GIT_SHALLOW FALSE GIT_TAG
   c510947ae9d3a67530cfe3e5eaccb5a3b8ea0e55 VERSION c510947ae9d3a67530cfe3e5eaccb5a3b8ea0e55
 )
 rapids_make_logger(cudf EXPORT_SET cudf-exports)
@@ -916,7 +916,9 @@ if(CUDF_LARGE_STRINGS_DISABLED)
 endif()
 
 # Define logging level
-target_compile_definitions(cudf PRIVATE "CUDF_LOG_ACTIVE_LEVEL=${LIBCUDF_LOGGING_LEVEL}")
+target_compile_definitions(
+  cudf PRIVATE "CUDF_LOG_ACTIVE_LEVEL=CUDF_LOG_LEVEL_${LIBCUDF_LOGGING_LEVEL}"
+)
 
 # Enable remote IO through KvikIO
 target_compile_definitions(cudf PRIVATE $<$<BOOL:${CUDF_KVIKIO_REMOTE_IO}>:CUDF_KVIKIO_REMOTE_IO>)
@@ -1105,7 +1107,7 @@ 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});