Merge branch 'fea-host_read_async' of https://github.com/vuule/cudf i…

…nto fea-host_read_async
rapidsai · Feb 20, 2025 · 2ebad98 · 2ebad98
2 parents aca9233 + 84ff62c
commit 2ebad98
Show file tree

Hide file tree

Showing 76 changed files with 1,693 additions and 2,477 deletions.
diff --git a/.github/workflows/spark-rapids-jni.yaml b/.github/workflows/spark-rapids-jni.yaml
@@ -7,7 +7,7 @@ jobs:
   spark-rapids-jni-build:
     runs-on: linux-amd64-cpu8
     container:
-      image: rapidsai/ci-spark-rapids-jni:rockylinux8-cuda12.2.0
+      image: rapidsai/ci-spark-rapids-jni:rockylinux8-cuda12.8.0
     steps:
       - uses: actions/checkout@v4
         with:

diff --git a/ci/test_cudf_polars_polars_tests.sh b/ci/test_cudf_polars_polars_tests.sh
@@ -27,8 +27,6 @@ git clone https://github.com/pola-rs/polars.git --branch "${TAG}" --depth 1
 # Install requirements for running polars tests
 rapids-logger "Install polars test requirements"
 rapids-pip-retry install -r polars/py-polars/requirements-dev.txt -r polars/py-polars/requirements-ci.txt
-# TODO: Workaround until https://github.com/pola-rs/polars/issues/21274 is fixed.
-rapids-pip-retry install connectorx==0.4.1
 
 # shellcheck disable=SC2317
 function set_exitcode()

diff --git a/conda/environments/all_cuda-118_arch-x86_64.yaml b/conda/environments/all_cuda-118_arch-x86_64.yaml
@@ -66,7 +66,7 @@ dependencies:
 - pandas
 - pandas>=2.0,<2.2.4dev0
 - pandoc
-- polars>=1.20,<1.22
+- polars>=1.20,<1.23
 - pre-commit
 - ptxcompiler
 - pyarrow>=14.0.0,<20.0.0a0

diff --git a/conda/environments/all_cuda-128_arch-x86_64.yaml b/conda/environments/all_cuda-128_arch-x86_64.yaml
@@ -64,7 +64,7 @@ dependencies:
 - pandas
 - pandas>=2.0,<2.2.4dev0
 - pandoc
-- polars>=1.20,<1.22
+- polars>=1.20,<1.23
 - pre-commit
 - pyarrow>=14.0.0,<20.0.0a0
 - pydata-sphinx-theme>=0.15.4

diff --git a/conda/recipes/cudf-polars/meta.yaml b/conda/recipes/cudf-polars/meta.yaml
@@ -43,7 +43,7 @@ requirements:
   run:
     - python
     - pylibcudf ={{ version }}
-    - polars >=1.20,<1.22
+    - polars >=1.20,<1.23
     - {{ pin_compatible('cuda-version', max_pin='x', min_pin='x') }}
 
 test:

diff --git a/cpp/include/cudf/column/column_device_view.cuh b/cpp/include/cudf/column/column_device_view.cuh
@@ -59,8 +59,8 @@ namespace CUDF_EXPORT cudf {
  *
  */
 struct nullate {
-  struct YES : cuda::std::bool_constant<true> {};
-  struct NO : cuda::std::bool_constant<false> {};
+  struct YES : cuda::std::true_type {};
+  struct NO : cuda::std::false_type {};
   /**
    * @brief `nullate::DYNAMIC` defers the determination of nullability to run time rather than
    * compile time. The calling code is responsible for specifying whether or not nulls are

diff --git a/cpp/include/cudf/datetime.hpp b/cpp/include/cudf/datetime.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2024, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2025, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -54,195 +54,6 @@ enum class datetime_component : uint8_t {
   NANOSECOND
 };
 
-/**
- * @brief  Extracts year from any datetime type and returns an int16_t
- * cudf::column.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t years
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_year(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts month from any datetime type and returns an int16_t
- * cudf::column.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t months
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_month(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts day from any datetime type and returns an int16_t
- * cudf::column.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t days
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_day(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts a weekday from any datetime type and returns an int16_t
- * cudf::column.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t days
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_weekday(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts hour from any datetime type and returns an int16_t
- * cudf::column.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t hours
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_hour(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts minute from any datetime type and returns an int16_t
- * cudf::column.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t minutes
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_minute(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts second from any datetime type and returns an int16_t
- * cudf::column.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t seconds
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_second(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts millisecond fraction from any datetime type and returns an int16_t
- * cudf::column.
- *
- * A millisecond fraction is only the 3 digits that make up the millisecond portion of a duration.
- * For example, the millisecond fraction of 1.234567890 seconds is 234.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t milliseconds
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_millisecond_fraction(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts microsecond fraction from any datetime type and returns an int16_t
- * cudf::column.
- *
- * A microsecond fraction is only the 3 digits that make up the microsecond portion of a duration.
- * For example, the microsecond fraction of 1.234567890 seconds is 567.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t microseconds
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_microsecond_fraction(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
-/**
- * @brief  Extracts nanosecond fraction from any datetime type and returns an int16_t
- * cudf::column.
- *
- * A nanosecond fraction is only the 3 digits that make up the nanosecond portion of a duration.
- * For example, the nanosecond fraction of 1.234567890 seconds is 890.
- *
- * @deprecated Deprecated in 24.12, to be removed in 25.02
- *
- * @param column cudf::column_view of the input datetime values
- * @param stream CUDA stream used for device memory operations and kernel launches
- * @param mr Device memory resource used to allocate device memory of the returned column
- *
- * @returns cudf::column of the extracted int16_t nanoseconds
- * @throw cudf::logic_error if input column datatype is not TIMESTAMP
- */
-[[deprecated]] std::unique_ptr<cudf::column> extract_nanosecond_fraction(
-  cudf::column_view const& column,
-  rmm::cuda_stream_view stream      = cudf::get_default_stream(),
-  rmm::device_async_resource_ref mr = cudf::get_current_device_resource_ref());
-
 /**
  * @brief Extracts the specified datetime component from any datetime type and
  * returns an int16_t cudf::column.

diff --git a/cpp/include/cudf/detail/datetime.hpp b/cpp/include/cudf/detail/datetime.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021-2024, NVIDIA CORPORATION.
+ * Copyright (c) 2021-2025, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -25,96 +25,6 @@
 namespace CUDF_EXPORT cudf {
 namespace datetime {
 namespace detail {
-/**
- * @copydoc cudf::extract_year(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_year(cudf::column_view const& column,
-                                           rmm::cuda_stream_view stream,
-                                           rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_month(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_month(cudf::column_view const& column,
-                                            rmm::cuda_stream_view stream,
-                                            rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_day(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_day(cudf::column_view const& column,
-                                          rmm::cuda_stream_view stream,
-                                          rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_weekday(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_weekday(cudf::column_view const& column,
-                                              rmm::cuda_stream_view stream,
-                                              rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_hour(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_hour(cudf::column_view const& column,
-                                           rmm::cuda_stream_view stream,
-                                           rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_minute(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_minute(cudf::column_view const& column,
-                                             rmm::cuda_stream_view stream,
-                                             rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_second(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_second(cudf::column_view const& column,
-                                             rmm::cuda_stream_view stream,
-                                             rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_millisecond_fraction(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_millisecond_fraction(cudf::column_view const& column,
-                                                           rmm::cuda_stream_view stream,
-                                                           rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_microsecond_fraction(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_microsecond_fraction(cudf::column_view const& column,
-                                                           rmm::cuda_stream_view stream,
-                                                           rmm::device_async_resource_ref mr);
-
-/**
- * @copydoc cudf::extract_nanosecond_fraction(cudf::column_view const&, rmm::cuda_stream_view,
- * rmm::device_async_resource_ref)
- *
- */
-std::unique_ptr<cudf::column> extract_nanosecond_fraction(cudf::column_view const& column,
-                                                          rmm::cuda_stream_view stream,
-                                                          rmm::device_async_resource_ref mr);
-
 /**
  * @copydoc cudf::extract_datetime_component(cudf::column_view const&, datetime_component,
  * rmm::cuda_stream_view, rmm::device_async_resource_ref)

diff --git a/cpp/include/cudf/detail/utilities/integer_utils.hpp b/cpp/include/cudf/detail/utilities/integer_utils.hpp
@@ -120,7 +120,7 @@ CUDF_HOST_DEVICE constexpr S div_rounding_up_unsafe(S const& dividend, T const&
 
 namespace detail {
 template <typename I>
-CUDF_HOST_DEVICE constexpr I div_rounding_up_safe(cuda::std::integral_constant<bool, false>,
+CUDF_HOST_DEVICE constexpr I div_rounding_up_safe(cuda::std::false_type,
                                                   I dividend,
                                                   I divisor) noexcept
 {
@@ -130,7 +130,7 @@ CUDF_HOST_DEVICE constexpr I div_rounding_up_safe(cuda::std::integral_constant<b
 }
 
 template <typename I>
-CUDF_HOST_DEVICE constexpr I div_rounding_up_safe(cuda::std::integral_constant<bool, true>,
+CUDF_HOST_DEVICE constexpr I div_rounding_up_safe(cuda::std::true_type,
                                                   I dividend,
                                                   I divisor) noexcept
 {
@@ -160,7 +160,7 @@ CUDF_HOST_DEVICE constexpr I div_rounding_up_safe(cuda::std::integral_constant<b
 template <typename I>
 CUDF_HOST_DEVICE constexpr I div_rounding_up_safe(I dividend, I divisor) noexcept
 {
-  using i_is_a_signed_type = cuda::std::integral_constant<bool, cuda::std::is_signed_v<I>>;
+  using i_is_a_signed_type = cuda::std::bool_constant<cuda::std::is_signed_v<I>>;
   return detail::div_rounding_up_safe(i_is_a_signed_type{}, dividend, divisor);
 }