Merge branch 'branch-22.04' of github.com:rapidsai/cudf into improvem…

…ent/rewrite_sample

build.sh

@@ -18,7 +18,7 @@ ARGS=$*
 REPODIR=$(cd $(dirname $0); pwd)
 
 VALIDARGS="clean libcudf cudf dask_cudf benchmarks tests libcudf_kafka cudf_kafka custreamz -v -g -n -l --allgpuarch --disable_nvtx --show_depr_warn --ptds -h --build_metrics --incl_cache_stats"
-HELP="$0 [clean] [libcudf] [cudf] [dask_cudf] [benchmarks] [tests] [libcudf_kafka] [cudf_kafka] [custreamz] [-v] [-g] [-n] [-h] [-l] [--cmake-args=\\\"<args>\\\"]
+HELP="$0 [clean] [libcudf] [cudf] [dask_cudf] [benchmarks] [tests] [libcudf_kafka] [cudf_kafka] [custreamz] [-v] [-g] [-n] [-h] [--cmake-args=\\\"<args>\\\"]
    clean                         - remove all existing build artifacts and configuration (start
                                    over)
    libcudf                       - build the cudf C++ code only
@@ -32,7 +32,6 @@ HELP="$0 [clean] [libcudf] [cudf] [dask_cudf] [benchmarks] [tests] [libcudf_kafk
    -v                            - verbose build mode
    -g                            - build for debug
    -n                            - no install step
-   -l                            - build legacy tests
    --allgpuarch                  - build for all supported GPU architectures
    --disable_nvtx                - disable inserting NVTX profiling ranges
    --show_depr_warn              - show cmake deprecation warnings

ci/benchmark/build.sh

@@ -1,5 +1,5 @@
 #!/bin/bash
-# Copyright (c) 2020, NVIDIA CORPORATION.
+# Copyright (c) 2020-2022, NVIDIA CORPORATION.
 #########################################
 # cuDF GPU build and test script for CI #
 #########################################
@@ -98,11 +98,7 @@ conda list --show-channel-urls
 ################################################################################
 
 logger "Build libcudf..."
-if [[ "${BUILD_MODE}" == "pull-request" ]]; then
-    "$WORKSPACE/build.sh" clean libcudf cudf dask_cudf benchmarks tests --ptds
-else
-    "$WORKSPACE/build.sh" clean libcudf cudf dask_cudf benchmarks tests -l --ptds
-fi
+"$WORKSPACE/build.sh" clean libcudf cudf dask_cudf benchmarks tests --ptds
 
 ################################################################################
 # BENCHMARK - Run and parse libcudf and cuDF benchmarks

ci/cpu/upload.sh

@@ -1,4 +1,5 @@
 #!/bin/bash
+# Copyright (c) 2020-2022, NVIDIA CORPORATION.
 #
 # Adopted from https://github.com/tmcdonell/travis-scripts/blob/dfaac280ac2082cd6bcaba3217428347899f2975/update-accelerate-buildbot.sh
 
@@ -23,56 +24,49 @@ if [ -z "$MY_UPLOAD_KEY" ]; then
   return 0
 fi
 
-################################################################################
-# SETUP - Get conda file output locations
-################################################################################
-
-gpuci_logger "Get conda file output locations"
-
-export LIBCUDF_FILE=`conda build --no-build-id --croot "$WORKSPACE/.conda-bld" conda/recipes/libcudf --output`
-export LIBCUDF_KAFKA_FILE=`conda build --no-build-id --croot "$WORKSPACE/.conda-bld" conda/recipes/libcudf_kafka --output`
-export CUDF_FILE=`conda build --croot ${CONDA_BLD_DIR} conda/recipes/cudf --python=$PYTHON --output`
-export DASK_CUDF_FILE=`conda build --croot ${CONDA_BLD_DIR} conda/recipes/dask-cudf --python=$PYTHON --output`
-export CUDF_KAFKA_FILE=`conda build --croot ${CONDA_BLD_DIR} conda/recipes/cudf_kafka --python=$PYTHON --output`
-export CUSTREAMZ_FILE=`conda build --croot ${CONDA_BLD_DIR} conda/recipes/custreamz --python=$PYTHON --output`
-
 ################################################################################
 # UPLOAD - Conda packages
 ################################################################################
 
 gpuci_logger "Starting conda uploads"
 if [[ "$BUILD_LIBCUDF" == "1" && "$UPLOAD_LIBCUDF" == "1" ]]; then
+  export LIBCUDF_FILE=$(conda build --no-build-id --croot "${CONDA_BLD_DIR}" conda/recipes/libcudf --output)
   test -e ${LIBCUDF_FILE}
   echo "Upload libcudf"
   echo ${LIBCUDF_FILE}
   gpuci_retry anaconda -t ${MY_UPLOAD_KEY} upload -u ${CONDA_USERNAME:-rapidsai} ${LABEL_OPTION} --skip-existing ${LIBCUDF_FILE} --no-progress
 fi
 
 if [[ "$BUILD_CUDF" == "1" && "$UPLOAD_CUDF" == "1" ]]; then
+  export CUDF_FILE=$(conda build --croot "${CONDA_BLD_DIR}" conda/recipes/cudf --python=$PYTHON --output)
   test -e ${CUDF_FILE}
   echo "Upload cudf"
   echo ${CUDF_FILE}
   gpuci_retry anaconda -t ${MY_UPLOAD_KEY} upload -u ${CONDA_USERNAME:-rapidsai} ${LABEL_OPTION} --skip-existing ${CUDF_FILE} --no-progress
 
+  export DASK_CUDF_FILE=$(conda build --croot "${CONDA_BLD_DIR}" conda/recipes/dask-cudf --python=$PYTHON --output)
   test -e ${DASK_CUDF_FILE}
   echo "Upload dask-cudf"
   echo ${DASK_CUDF_FILE}
   gpuci_retry anaconda -t ${MY_UPLOAD_KEY} upload -u ${CONDA_USERNAME:-rapidsai} ${LABEL_OPTION} --skip-existing ${DASK_CUDF_FILE} --no-progress
 
+  export CUSTREAMZ_FILE=$(conda build --croot "${CONDA_BLD_DIR}" conda/recipes/custreamz --python=$PYTHON --output)
   test -e ${CUSTREAMZ_FILE}
   echo "Upload custreamz"
   echo ${CUSTREAMZ_FILE}
   gpuci_retry anaconda -t ${MY_UPLOAD_KEY} upload -u ${CONDA_USERNAME:-rapidsai} ${LABEL_OPTION} --skip-existing ${CUSTREAMZ_FILE} --no-progress
 fi
 
 if [[ "$BUILD_LIBCUDF" == "1" && "$UPLOAD_LIBCUDF_KAFKA" == "1" ]]; then
+  export LIBCUDF_KAFKA_FILE=$(conda build --no-build-id --croot "${CONDA_BLD_DIR}" conda/recipes/libcudf_kafka --output)
   test -e ${LIBCUDF_KAFKA_FILE}
   echo "Upload libcudf_kafka"
   echo ${LIBCUDF_KAFKA_FILE}
   gpuci_retry anaconda -t ${MY_UPLOAD_KEY} upload -u ${CONDA_USERNAME:-rapidsai} ${LABEL_OPTION} --skip-existing ${LIBCUDF_KAFKA_FILE} --no-progress
 fi
 
 if [[ "$BUILD_CUDF" == "1" && "$UPLOAD_CUDF_KAFKA" == "1" ]]; then
+  export CUDF_KAFKA_FILE=$(conda build --croot "${CONDA_BLD_DIR}" conda/recipes/cudf_kafka --python=$PYTHON --output)
   test -e ${CUDF_KAFKA_FILE}
   echo "Upload cudf_kafka"
   echo ${CUDF_KAFKA_FILE}

ci/gpu/build.sh

@@ -1,5 +1,5 @@
 #!/bin/bash
-# Copyright (c) 2018-2021, NVIDIA CORPORATION.
+# Copyright (c) 2018-2022, NVIDIA CORPORATION.
 ##############################################
 # cuDF GPU build and test script for CI      #
 ##############################################
@@ -128,11 +128,7 @@ if [[ -z "$PROJECT_FLASH" || "$PROJECT_FLASH" == "0" ]]; then
     ################################################################################
 
     gpuci_logger "Build from source"
-    if [[ "${BUILD_MODE}" == "pull-request" ]]; then
-        "$WORKSPACE/build.sh" clean libcudf cudf dask_cudf libcudf_kafka cudf_kafka benchmarks tests --ptds
-    else
-        "$WORKSPACE/build.sh" clean libcudf cudf dask_cudf libcudf_kafka cudf_kafka benchmarks tests -l --ptds
-    fi
+    "$WORKSPACE/build.sh" clean libcudf cudf dask_cudf libcudf_kafka cudf_kafka benchmarks tests --ptds
 
     ################################################################################
     # TEST - Run GoogleTest
@@ -226,11 +222,7 @@ else
     install_dask
 
     gpuci_logger "Build python libs from source"
-    if [[ "${BUILD_MODE}" == "pull-request" ]]; then
-        "$WORKSPACE/build.sh" cudf dask_cudf cudf_kafka --ptds
-    else
-        "$WORKSPACE/build.sh" cudf dask_cudf cudf_kafka -l --ptds
-    fi
+    "$WORKSPACE/build.sh" cudf dask_cudf cudf_kafka --ptds
 
 fi
 
@@ -249,15 +241,15 @@ fi
 
 cd "$WORKSPACE/python/cudf"
 gpuci_logger "Python py.test for cuDF"
-py.test -n 8 --cache-clear --basetemp="$WORKSPACE/cudf-cuda-tmp" --ignore="$WORKSPACE/python/cudf/cudf/benchmarks" --junitxml="$WORKSPACE/junit-cudf.xml" -v --cov-config=.coveragerc --cov=cudf --cov-report=xml:"$WORKSPACE/python/cudf/cudf-coverage.xml" --cov-report term --dist=loadscope
+py.test -n 8 --cache-clear --basetemp="$WORKSPACE/cudf-cuda-tmp" --ignore="$WORKSPACE/python/cudf/cudf/benchmarks" --junitxml="$WORKSPACE/junit-cudf.xml" -v --cov-config=.coveragerc --cov=cudf --cov-report=xml:"$WORKSPACE/python/cudf/cudf-coverage.xml" --cov-report term --dist=loadscope cudf
 
 cd "$WORKSPACE/python/dask_cudf"
 gpuci_logger "Python py.test for dask-cudf"
-py.test -n 8 --cache-clear --basetemp="$WORKSPACE/dask-cudf-cuda-tmp" --junitxml="$WORKSPACE/junit-dask-cudf.xml" -v --cov-config=.coveragerc --cov=dask_cudf --cov-report=xml:"$WORKSPACE/python/dask_cudf/dask-cudf-coverage.xml" --cov-report term
+py.test -n 8 --cache-clear --basetemp="$WORKSPACE/dask-cudf-cuda-tmp" --junitxml="$WORKSPACE/junit-dask-cudf.xml" -v --cov-config=.coveragerc --cov=dask_cudf --cov-report=xml:"$WORKSPACE/python/dask_cudf/dask-cudf-coverage.xml" --cov-report term dask_cudf
 
 cd "$WORKSPACE/python/custreamz"
 gpuci_logger "Python py.test for cuStreamz"
-py.test -n 8 --cache-clear --basetemp="$WORKSPACE/custreamz-cuda-tmp" --junitxml="$WORKSPACE/junit-custreamz.xml" -v --cov-config=.coveragerc --cov=custreamz --cov-report=xml:"$WORKSPACE/python/custreamz/custreamz-coverage.xml" --cov-report term
+py.test -n 8 --cache-clear --basetemp="$WORKSPACE/custreamz-cuda-tmp" --junitxml="$WORKSPACE/junit-custreamz.xml" -v --cov-config=.coveragerc --cov=custreamz --cov-report=xml:"$WORKSPACE/python/custreamz/custreamz-coverage.xml" --cov-report term custreamz
 
 gpuci_logger "Test notebooks"
 "$WORKSPACE/ci/gpu/test-notebooks.sh" 2>&1 | tee nbtest.log

conda/recipes/libcudf_kafka/meta.yaml

@@ -27,8 +27,6 @@ requirements:
   host:
     - libcudf {{version}}
     - librdkafka >=1.7.0,<1.8.0a0
-  run:
-    - {{ pin_compatible('librdkafka', max_pin='x.x') }} #TODO: librdkafka should be automatically included here by run_exports but is not
 
 test:
   commands:

cpp/CMakeLists.txt

@@ -498,6 +498,8 @@ set_target_properties(
              # set target compile options
              CXX_STANDARD 17
              CXX_STANDARD_REQUIRED ON
+             # For std:: support of __int128_t. Can be removed once using cuda::std
+             CXX_EXTENSIONS ON
              CUDA_STANDARD 17
              CUDA_STANDARD_REQUIRED ON
              POSITION_INDEPENDENT_CODE ON

cpp/benchmarks/CMakeLists.txt

@@ -59,6 +59,12 @@ function(ConfigureBench CMAKE_BENCH_NAME)
     ${CMAKE_BENCH_NAME}
     PROPERTIES RUNTIME_OUTPUT_DIRECTORY "$<BUILD_INTERFACE:${CUDF_BINARY_DIR}/benchmarks>"
                INSTALL_RPATH "\$ORIGIN/../../../lib"
+               CXX_STANDARD 17
+               CXX_STANDARD_REQUIRED ON
+               # For std:: support of __int128_t. Can be removed once using cuda::std
+               CXX_EXTENSIONS ON
+               CUDA_STANDARD 17
+               CUDA_STANDARD_REQUIRED ON
   )
   target_link_libraries(
     ${CMAKE_BENCH_NAME} PRIVATE cudf_benchmark_common cudf_datagen benchmark::benchmark_main
@@ -254,7 +260,7 @@ ConfigureBench(
   string/convert_durations.cpp
   string/convert_fixed_point.cpp
   string/convert_numerics.cpp
-  string/copy.cpp
+  string/copy.cu
   string/extract.cpp
   string/factory.cu
   string/filter.cpp

cpp/benchmarks/common/generate_input.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, NVIDIA CORPORATION.
+ * Copyright (c) 2020-2022, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -69,15 +69,14 @@ distribution_id default_distribution_id()
   return distribution_id::GEOMETRIC;
 }
 
-template <typename T,
-          std::enable_if_t<!std::is_unsigned<T>::value && cudf::is_numeric<T>()>* = nullptr>
+template <typename T, std::enable_if_t<!std::is_unsigned_v<T> && cudf::is_numeric<T>()>* = nullptr>
 distribution_id default_distribution_id()
 {
   return distribution_id::NORMAL;
 }
 
 template <typename T,
-          std::enable_if_t<!std::is_same_v<T, bool> && std::is_unsigned<T>::value &&
+          std::enable_if_t<!std::is_same_v<T, bool> && std::is_unsigned_v<T> &&
                            cudf::is_numeric<T>()>* = nullptr>
 distribution_id default_distribution_id()
 {
@@ -226,7 +225,7 @@ class data_profile {
  public:
   template <
     typename T,
-    typename std::enable_if_t<!std::is_same_v<T, bool> && std::is_integral<T>::value, T>* = nullptr>
+    typename std::enable_if_t<!std::is_same_v<T, bool> && std::is_integral_v<T>, T>* = nullptr>
   distribution_params<T> get_distribution_params() const
   {
     auto it = int_params.find(cudf::type_to_id<T>());
@@ -239,7 +238,7 @@ class data_profile {
     }
   }
 
-  template <typename T, typename std::enable_if_t<std::is_floating_point<T>::value, T>* = nullptr>
+  template <typename T, typename std::enable_if_t<std::is_floating_point_v<T>, T>* = nullptr>
   distribution_params<T> get_distribution_params() const
   {
     auto it = float_params.find(cudf::type_to_id<T>());
@@ -307,7 +306,7 @@ class data_profile {
   // discrete distributions (integers, strings, lists). Otherwise the call with have no effect.
   template <typename T,
             typename Type_enum,
-            typename std::enable_if_t<std::is_integral<T>::value, T>* = nullptr>
+            typename std::enable_if_t<std::is_integral_v<T>, T>* = nullptr>
   void set_distribution_params(Type_enum type_or_group,
                                distribution_id dist,
                                T lower_bound,
@@ -331,7 +330,7 @@ class data_profile {
   // have continuous distributions (floating point types). Otherwise the call with have no effect.
   template <typename T,
             typename Type_enum,
-            typename std::enable_if_t<std::is_floating_point<T>::value, T>* = nullptr>
+            typename std::enable_if_t<std::is_floating_point_v<T>, T>* = nullptr>
   void set_distribution_params(Type_enum type_or_group,
                                distribution_id dist,
                                T lower_bound,

cpp/benchmarks/common/random_distribution_factory.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, NVIDIA CORPORATION.
+ * Copyright (c) 2020-2022, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -24,7 +24,7 @@
 /**
  * @brief Generates a normal(binomial) distribution between zero and upper_bound.
  */
-template <typename T, typename std::enable_if_t<std::is_integral<T>::value, T>* = nullptr>
+template <typename T, typename std::enable_if_t<std::is_integral_v<T>, T>* = nullptr>
 auto make_normal_dist(T upper_bound)
 {
   using uT = typename std::make_unsigned<T>::type;
@@ -42,7 +42,7 @@ auto make_normal_dist(T upper_bound)
   return std::normal_distribution<T>(mean, stddev);
 }
 
-template <typename T, typename std::enable_if_t<std::is_integral<T>::value, T>* = nullptr>
+template <typename T, typename std::enable_if_t<std::is_integral_v<T>, T>* = nullptr>
 auto make_uniform_dist(T range_start, T range_end)
 {
   return std::uniform_int_distribution<T>(range_start, range_end);
@@ -62,7 +62,7 @@ double geometric_dist_p(T range_size)
   return p ? p : std::numeric_limits<double>::epsilon();
 }
 
-template <typename T, typename std::enable_if_t<std::is_integral<T>::value, T>* = nullptr>
+template <typename T, typename std::enable_if_t<std::is_integral_v<T>, T>* = nullptr>
 auto make_geometric_dist(T range_start, T range_end)
 {
   using uT = typename std::make_unsigned<T>::type;
@@ -82,7 +82,7 @@ auto make_geometric_dist(T range_start, T range_end)
 template <typename T>
 using distribution_fn = std::function<T(std::mt19937&)>;
 
-template <typename T, typename std::enable_if_t<std::is_integral<T>::value, T>* = nullptr>
+template <typename T, typename std::enable_if_t<std::is_integral_v<T>, T>* = nullptr>
 distribution_fn<T> make_distribution(distribution_id did, T lower_bound, T upper_bound)
 {
   switch (did) {

cpp/benchmarks/io/text/multibyte_split.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021, NVIDIA CORPORATION.
+ * Copyright (c) 2021-2022, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -135,12 +135,14 @@ static void BM_multibyte_split(benchmark::State& state)
     default: CUDF_FAIL();
   }
 
+  auto mem_stats_logger = cudf::memory_stats_logger();
   for (auto _ : state) {
     cuda_event_timer raii(state, true);
     auto output = cudf::io::text::multibyte_split(*source, delim);
   }
 
   state.SetBytesProcessed(state.iterations() * device_input.size());
+  state.counters["peak_memory_usage"] = mem_stats_logger.peak_memory_usage();
 }
 
 class MultibyteSplitBenchmark : public cudf::benchmark {