cuda-cccl 0.3.0__cp312-cp312-manylinux_2_24_aarch64.whl → 0.3.1__cp312-cp312-manylinux_2_24_aarch64.whl

cuda/cccl/headers/include/cuda/std/__bit/countl.h

@@ -100,7 +100,14 @@ template <typename _Tp>
 template <typename _Tp>
 [[nodiscard]] _CCCL_HIDE_FROM_ABI _CCCL_DEVICE int __cccl_countl_zero_impl_device(_Tp __v) noexcept
 {
-  return (sizeof(_Tp) == sizeof(uint32_t)) ? ::__clz(static_cast<int>(__v)) : ::__clzll(static_cast<long long>(__v));
+  if constexpr (sizeof(_Tp) == sizeof(uint32_t))
+  {
+    return static_cast<int>(::__clz(static_cast<int>(__v)));
+  }
+  else
+  {
+    return static_cast<int>(::__clzll(static_cast<long long>(__v)));
+  }
 }
 #endif // _CCCL_CUDA_COMPILATION()
 

cuda/cccl/headers/include/cuda/std/__bit/countr.h

@@ -114,11 +114,11 @@ template <typename _Tp>
 {
   if constexpr (sizeof(_Tp) == sizeof(uint32_t))
   {
-    return ::__clz(static_cast<int>(::__brev(__v)));
+    return static_cast<int>(::__clz(static_cast<int>(::__brev(__v))));
   }
   else
   {
-    return ::__clzll(static_cast<long long>(::__brevll(__v)));
+    return static_cast<int>(::__clzll(static_cast<long long>(::__brevll(__v))));
   }
 }
 #endif // _CCCL_CUDA_COMPILATION()

cuda/cccl/headers/include/cuda/std/__bit/reference.h

@@ -275,10 +275,10 @@ _CCCL_API constexpr __bit_iterator<_Cp, false> __copy_aligned(
     // do first word
     if (__first.__ctz_ != 0)
     {
-      unsigned __clz       = __bits_per_word - __first.__ctz_;
-      difference_type __dn = ::cuda::std::min(static_cast<difference_type>(__clz), __n);
+      unsigned __clz_f     = __bits_per_word - __first.__ctz_;
+      difference_type __dn = ::cuda::std::min(static_cast<difference_type>(__clz_f), __n);
       __n -= __dn;
-      __storage_type __m = (~__storage_type(0) << __first.__ctz_) & (~__storage_type(0) >> (__clz - __dn));
+      __storage_type __m = (~__storage_type(0) << __first.__ctz_) & (~__storage_type(0) >> (__clz_f - __dn));
       __storage_type __b = *__first.__seg_ & __m;
       *__result.__seg_ &= ~__m;
       *__result.__seg_ |= __b;
@@ -420,8 +420,8 @@ _CCCL_API constexpr __bit_iterator<_Cp, false> __copy_backward_aligned(
     {
       difference_type __dn = ::cuda::std::min(static_cast<difference_type>(__last.__ctz_), __n);
       __n -= __dn;
-      unsigned __clz     = __bits_per_word - __last.__ctz_;
-      __storage_type __m = (~__storage_type(0) << (__last.__ctz_ - __dn)) & (~__storage_type(0) >> __clz);
+      unsigned __clz_f   = __bits_per_word - __last.__ctz_;
+      __storage_type __m = (~__storage_type(0) << (__last.__ctz_ - __dn)) & (~__storage_type(0) >> __clz_f);
       __storage_type __b = *__last.__seg_ & __m;
       *__result.__seg_ &= ~__m;
       *__result.__seg_ |= __b;
@@ -635,10 +635,10 @@ _CCCL_API inline __bit_iterator<_Cr, false> __swap_ranges_aligned(
     // do first word
     if (__first.__ctz_ != 0)
     {
-      unsigned __clz       = __bits_per_word - __first.__ctz_;
-      difference_type __dn = ::cuda::std::min(static_cast<difference_type>(__clz), __n);
+      unsigned __clz_f     = __bits_per_word - __first.__ctz_;
+      difference_type __dn = ::cuda::std::min(static_cast<difference_type>(__clz_f), __n);
       __n -= __dn;
-      __storage_type __m  = (~__storage_type(0) << __first.__ctz_) & (~__storage_type(0) >> (__clz - __dn));
+      __storage_type __m  = (~__storage_type(0) << __first.__ctz_) & (~__storage_type(0) >> (__clz_f - __dn));
       __storage_type __b1 = *__first.__seg_ & __m;
       *__first.__seg_ &= ~__m;
       __storage_type __b2 = *__result.__seg_ & __m;
@@ -988,10 +988,10 @@ _CCCL_API constexpr bool __equal_aligned(
     // do first word
     if (__first1.__ctz_ != 0)
     {
-      unsigned __clz       = __bits_per_word - __first1.__ctz_;
-      difference_type __dn = ::cuda::std::min(static_cast<difference_type>(__clz), __n);
+      unsigned __clz_f     = __bits_per_word - __first1.__ctz_;
+      difference_type __dn = ::cuda::std::min(static_cast<difference_type>(__clz_f), __n);
       __n -= __dn;
-      __storage_type __m = (~__storage_type(0) << __first1.__ctz_) & (~__storage_type(0) >> (__clz - __dn));
+      __storage_type __m = (~__storage_type(0) << __first1.__ctz_) & (~__storage_type(0) >> (__clz_f - __dn));
       if ((*__first2.__seg_ & __m) != (*__first1.__seg_ & __m))
       {
         return false;

cuda/cccl/headers/include/cuda/std/__chrono/duration.h

@@ -43,19 +43,19 @@ template <class _Rep, class _Period = ratio<1>>
 class _CCCL_TYPE_VISIBILITY_DEFAULT duration;
 
 template <class _Tp>
-inline const bool __is_duration_v = false;
+inline constexpr bool __is_duration_v = false;
 
 template <class _Rep, class _Period>
-inline const bool __is_duration_v<duration<_Rep, _Period>> = true;
+inline constexpr bool __is_duration_v<duration<_Rep, _Period>> = true;
 
 template <class _Rep, class _Period>
-inline const bool __is_duration_v<const duration<_Rep, _Period>> = true;
+inline constexpr bool __is_duration_v<const duration<_Rep, _Period>> = true;
 
 template <class _Rep, class _Period>
-inline const bool __is_duration_v<volatile duration<_Rep, _Period>> = true;
+inline constexpr bool __is_duration_v<volatile duration<_Rep, _Period>> = true;
 
 template <class _Rep, class _Period>
-inline const bool __is_duration_v<const volatile duration<_Rep, _Period>> = true;
+inline constexpr bool __is_duration_v<const volatile duration<_Rep, _Period>> = true;
 
 } // namespace chrono
 
@@ -190,29 +190,29 @@ class _CCCL_TYPE_VISIBILITY_DEFAULT duration
   struct __no_overflow
   {
   private:
-    static const intmax_t __gcd_n1_n2 = __static_gcd<_R1::num, _R2::num>::value;
-    static const intmax_t __gcd_d1_d2 = __static_gcd<_R1::den, _R2::den>::value;
-    static const intmax_t __n1        = _R1::num / __gcd_n1_n2;
-    static const intmax_t __d1        = _R1::den / __gcd_d1_d2;
-    static const intmax_t __n2        = _R2::num / __gcd_n1_n2;
-    static const intmax_t __d2        = _R2::den / __gcd_d1_d2;
-    static const intmax_t max         = -((intmax_t(1) << (sizeof(intmax_t) * CHAR_BIT - 1)) + 1);
+    static constexpr intmax_t __gcd_n1_n2 = __static_gcd<_R1::num, _R2::num>::value;
+    static constexpr intmax_t __gcd_d1_d2 = __static_gcd<_R1::den, _R2::den>::value;
+    static constexpr intmax_t __n1        = _R1::num / __gcd_n1_n2;
+    static constexpr intmax_t __d1        = _R1::den / __gcd_d1_d2;
+    static constexpr intmax_t __n2        = _R2::num / __gcd_n1_n2;
+    static constexpr intmax_t __d2        = _R2::den / __gcd_d1_d2;
+    static constexpr intmax_t max         = -((intmax_t(1) << (sizeof(intmax_t) * CHAR_BIT - 1)) + 1);
 
     template <intmax_t _Xp, intmax_t _Yp, bool __overflow>
     struct __mul // __overflow == false
     {
-      static const intmax_t value = _Xp * _Yp;
+      static constexpr intmax_t value = _Xp * _Yp;
     };
 
     template <intmax_t _Xp, intmax_t _Yp>
     struct __mul<_Xp, _Yp, true>
     {
-      static const intmax_t value = 1;
+      static constexpr intmax_t value = 1;
     };
 
   public:
-    static const bool value = (__n1 <= max / __d2) && (__n2 <= max / __d1);
-    using type              = ratio<__mul<__n1, __d2, !value>::value, __mul<__n2, __d1, !value>::value>;
+    static constexpr bool value = (__n1 <= max / __d2) && (__n2 <= max / __d1);
+    using type                  = ratio<__mul<__n1, __d2, !value>::value, __mul<__n2, __d1, !value>::value>;
   };
 
 public:

cuda/cccl/headers/include/cuda/std/__chrono/steady_clock.h

@@ -40,11 +40,11 @@ namespace chrono
 class _CCCL_TYPE_VISIBILITY_DEFAULT steady_clock
 {
 public:
-  using duration                        = nanoseconds;
-  using rep                             = duration::rep;
-  using period                          = duration::period;
-  using time_point                      = ::cuda::std::chrono::time_point<steady_clock, duration>;
-  static constexpr const bool is_steady = true;
+  using duration                  = nanoseconds;
+  using rep                       = duration::rep;
+  using period                    = duration::period;
+  using time_point                = ::cuda::std::chrono::time_point<steady_clock, duration>;
+  static constexpr bool is_steady = true;
 
   [[nodiscard]] _CCCL_API static time_point now() noexcept;
 };

cuda/cccl/headers/include/cuda/std/__chrono/system_clock.h

@@ -39,11 +39,11 @@ namespace chrono
 class _CCCL_TYPE_VISIBILITY_DEFAULT system_clock
 {
 public:
-  using duration                        = ::cuda::std::chrono::nanoseconds;
-  using rep                             = duration::rep;
-  using period                          = duration::period;
-  using time_point                      = ::cuda::std::chrono::time_point<system_clock>;
-  static constexpr const bool is_steady = false;
+  using duration                  = ::cuda::std::chrono::nanoseconds;
+  using rep                       = duration::rep;
+  using period                    = duration::period;
+  using time_point                = ::cuda::std::chrono::time_point<system_clock>;
+  static constexpr bool is_steady = false;
 
   [[nodiscard]] _CCCL_API inline static time_point now() noexcept
   {

cuda/cccl/headers/include/cuda/std/__floating_point/fp.h

@@ -11,7 +11,7 @@
 #ifndef _CUDA_STD___FLOATING_POINT_FP_H
 #define _CUDA_STD___FLOATING_POINT_FP_H
 
-#include <cuda/__cccl_config>
+#include <cuda/std/detail/__config>
 
 #if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
 #  pragma GCC system_header

cuda/cccl/headers/include/cuda/std/__tuple_dir/make_tuple_types.h

@@ -20,7 +20,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__fwd/complex.h>
 #include <cuda/std/__fwd/array.h>
+#include <cuda/std/__fwd/complex.h>
 #include <cuda/std/__fwd/tuple.h>
 #include <cuda/std/__tuple_dir/tuple_element.h>
 #include <cuda/std/__tuple_dir/tuple_indices.h>
@@ -61,7 +63,27 @@ struct __make_tuple_types_flat<array<_Vt, _Np>, __tuple_indices<_Idx...>>
   template <size_t>
   using __value_type = _Vt;
   template <class _Tp, class _ApplyFn = __apply_cvref_fn<_Tp>>
-  using __apply_quals = __tuple_types<__type_call<_ApplyFn, __value_type<_Idx>>...>;
+  using __apply_quals _CCCL_NODEBUG_ALIAS = __tuple_types<__type_call<_ApplyFn, __value_type<_Idx>>...>;
+};
+
+template <class _Vt, size_t... _Idx>
+struct __make_tuple_types_flat<complex<_Vt>, __tuple_indices<_Idx...>>
+{
+  static_assert(sizeof...(_Idx) == 2, "__make_tuple_types: complex has only 2 members");
+  template <size_t>
+  using __value_type = _Vt;
+  template <class _Tp, class _ApplyFn = __apply_cvref_fn<_Tp>>
+  using __apply_quals _CCCL_NODEBUG_ALIAS = __tuple_types<__type_call<_ApplyFn, __value_type<_Idx>>...>;
+};
+
+template <class _Vt, size_t... _Idx>
+struct __make_tuple_types_flat<::cuda::complex<_Vt>, __tuple_indices<_Idx...>>
+{
+  static_assert(sizeof...(_Idx) == 2, "__make_tuple_types: complex has only 2 members");
+  template <size_t>
+  using __value_type = _Vt;
+  template <class _Tp, class _ApplyFn = __apply_cvref_fn<_Tp>>
+  using __apply_quals _CCCL_NODEBUG_ALIAS = __tuple_types<__type_call<_ApplyFn, __value_type<_Idx>>...>;
 };
 
 template <class _Tp,

cuda/cccl/headers/include/cuda/std/__tuple_dir/tuple_like.h

@@ -20,6 +20,7 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__fwd/complex.h>
 #include <cuda/std/__concepts/concept_macros.h>
 #include <cuda/std/__fwd/array.h>
 #include <cuda/std/__fwd/complex.h>
@@ -58,6 +59,9 @@ inline constexpr bool __tuple_like_impl<array<_Tp, _Size>> = true;
 template <class _Tp>
 inline constexpr bool __tuple_like_impl<complex<_Tp>> = true;
 
+template <class _Tp>
+inline constexpr bool __tuple_like_impl<::cuda::complex<_Tp>> = true;
+
 template <class _Ip, class _Sp, ::cuda::std::ranges::subrange_kind _Kp>
 inline constexpr bool __tuple_like_impl<::cuda::std::ranges::subrange<_Ip, _Sp, _Kp>> = true;
 

cuda/cccl/headers/include/cuda/std/__tuple_dir/tuple_like_ext.h

@@ -20,6 +20,7 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__fwd/complex.h>
 #include <cuda/std/__fwd/array.h>
 #include <cuda/std/__fwd/complex.h>
 #include <cuda/std/__fwd/pair.h>
@@ -54,6 +55,9 @@ inline constexpr bool __tuple_like_ext<array<_Tp, _Size>> = true;
 template <class _Tp>
 inline constexpr bool __tuple_like_ext<complex<_Tp>> = true;
 
+template <class _Tp>
+inline constexpr bool __tuple_like_ext<::cuda::complex<_Tp>> = true;
+
 template <class... _Tp>
 inline constexpr bool __tuple_like_ext<__tuple_types<_Tp...>> = true;
 

cuda/cccl/headers/include/cuda/std/string_view

@@ -57,7 +57,7 @@
 #include <cuda/std/version>
 
 #if !_CCCL_COMPILER(NVRTC)
-#  include <iosfwd>
+#  include <string_view>
 #endif // !_CCCL_COMPILER(NVRTC)
 
 #include <cuda/std/__cccl/prologue.h>
@@ -727,14 +727,21 @@ _CCCL_HOST_DEVICE basic_string_view(_Range&&) -> basic_string_view<::cuda::std::
 
 // operator <<
 
-#if 0 // todo: we need to implement char_traits stream types & functions
+#if !_CCCL_COMPILER(NVRTC)
+template <class _CharT>
+_CCCL_HOST_API ::std::basic_ostream<_CharT>&
+operator<<(::std::basic_ostream<_CharT>& __os, basic_string_view<_CharT> __str)
+{
+  return __os << ::std::basic_string_view<_CharT>{__str.data(), __str.size()};
+}
+
 template <class _CharT, class _Traits>
-_CCCL_API inline ::std::basic_ostream<_CharT, _Traits>&
+_CCCL_HOST_API ::std::basic_ostream<_CharT, _Traits>&
 operator<<(::std::basic_ostream<_CharT, _Traits>& __os, basic_string_view<_CharT, _Traits> __str)
 {
-  return __os.write(__str.data(), static_cast<::std::streamsize>(__str.size()));
+  return __os << ::std::basic_string_view<_CharT, _Traits>{__str.data(), __str.size()};
 }
-#endif // 0
+#endif // !_CCCL_COMPILER(NVRTC)
 
 // literals
 

cuda/cccl/headers/include/cuda/std/version

@@ -141,7 +141,7 @@
 // #   define __cccl_lib_shared_mutex                       201505L
 // # define __cccl_lib_shared_ptr_arrays                    201611L
 // # define __cccl_lib_shared_ptr_weak_type                 201606L
-// # define __cccl_lib_string_view                          201606L
+#define __cccl_lib_string_view 201803L
 // # define __cccl_lib_to_chars                             201611L
 // #  define __cccl_lib_uncaught_exceptions           201411L
 // #  define __cccl_lib_unordered_map_try_emplace     201411L
@@ -171,7 +171,6 @@
 // # define __cccl_lib_constexpr_misc                       201811L
 // # define __cccl_lib_constexpr_numeric                    201911L
 // # define __cccl_lib_constexpr_string                     201907L
-// # define __cccl_lib_constexpr_string_view                201811L
 // # define __cccl_lib_constexpr_swap_algorithms            201806L
 // # define __cccl_lib_constexpr_tuple                      201811L
 // # define __cccl_lib_constexpr_utility                    201811L
@@ -204,8 +203,6 @@
 // # define __cccl_lib_source_location                      201907L
 // # define __cccl_lib_ssize                                201902L
 // # define __cccl_lib_starts_ends_with                     201711L
-// # undef  __cccl_lib_string_view
-// # define __cccl_lib_string_view                          201803L
 // # define __cccl_lib_syncbuf                              201803L
 // # define __cccl_lib_three_way_comparison                 201907L
 #  define __cccl_lib_unwrap_ref 201811L

cuda/cccl/headers/include/thrust/detail/integer_math.h

@@ -27,6 +27,8 @@
 #endif // no system header
 #include <thrust/detail/type_deduction.h>
 
+#include <cuda/std/__bit/countl.h>
+#include <cuda/std/__type_traits/make_unsigned.h>
 #include <cuda/std/limits>
 #include <cuda/std/type_traits>
 
@@ -36,25 +38,6 @@ THRUST_NAMESPACE_BEGIN
 namespace detail
 {
 
-template <typename Integer>
-_CCCL_HOST_DEVICE _CCCL_FORCEINLINE Integer clz(Integer x)
-{
-  Integer result;
-
-  NV_IF_TARGET(NV_IS_DEVICE,
-               (result = ::__clz(x);),
-               (int num_bits = 8 * sizeof(Integer); int num_bits_minus_one = num_bits - 1; result = num_bits;
-                for (int i = num_bits_minus_one; i >= 0; --i) {
-                  if ((Integer(1) << i) & x)
-                  {
-                    result = num_bits_minus_one - i;
-                    break;
-                  }
-                }));
-
-  return result;
-}
-
 template <typename Integer>
 _CCCL_HOST_DEVICE _CCCL_FORCEINLINE bool is_power_of_2(Integer x)
 {
@@ -85,7 +68,7 @@ _CCCL_HOST_DEVICE _CCCL_FORCEINLINE Integer log2(Integer x)
   Integer num_bits           = 8 * sizeof(Integer);
   Integer num_bits_minus_one = num_bits - 1;
 
-  return num_bits_minus_one - clz(x);
+  return num_bits_minus_one - ::cuda::std::countl_zero(::cuda::std::__to_unsigned_like(x));
 }
 
 template <typename Integer>

cuda/cccl/headers/include/thrust/iterator/iterator_traits.h

@@ -316,6 +316,17 @@ struct iterator_traversal<::cuda::zip_iterator<Iterators...>>
   using type = detail::minimum_type<iterator_traversal_t<Iterators>...>;
 };
 
+template <class Fn, class... Iterators>
+struct iterator_system<::cuda::zip_transform_iterator<Fn, Iterators...>>
+{
+  using type = detail::minimum_system_t<iterator_system_t<Iterators>...>;
+};
+template <class Fn, class... Iterators>
+struct iterator_traversal<::cuda::zip_transform_iterator<Fn, Iterators...>>
+{
+  using type = detail::minimum_type<iterator_traversal_t<Iterators>...>;
+};
+
 //! \} // end iterator_traits
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/system/cuda/detail/copy.h

@@ -48,6 +48,13 @@
 #include <thrust/system/cuda/detail/util.h>
 #include <thrust/type_traits/is_trivially_relocatable.h>
 
+#if _CCCL_HAS_CUDA_COMPILER()
+#  include <cub/device/dispatch/tuning/tuning_transform.cuh>
+#endif // _CCCL_HAS_CUDA_COMPILER()
+
+#include <cuda/__fwd/zip_iterator.h>
+#include <cuda/std/tuple>
+
 THRUST_NAMESPACE_BEGIN
 namespace cuda_cub
 {
@@ -61,6 +68,21 @@ template <class Derived, class InputIt, class OutputIt, class TransformOp>
 OutputIt _CCCL_API _CCCL_FORCEINLINE
 transform(execution_policy<Derived>& policy, InputIt first, InputIt last, OutputIt result, TransformOp transform_op);
 
+// Forward declare to work around a cyclic include, since "cuda/detail/transform.h" includes this header
+// We want this to unwrap zip_transform_iterator
+namespace __transform
+{
+_CCCL_EXEC_CHECK_DISABLE
+template <class Derived, class Offset, class... InputIts, class OutputIt, class TransformOp, class Predicate>
+OutputIt _CCCL_API _CCCL_FORCEINLINE cub_transform_many(
+  execution_policy<Derived>& policy,
+  ::cuda::std::tuple<InputIts...> firsts,
+  OutputIt result,
+  Offset num_items,
+  TransformOp transform_op,
+  Predicate pred);
+} // namespace __transform
+
 namespace __copy
 {
 template <class H, class D, class T, class Size>
@@ -190,6 +212,17 @@ device_to_device(execution_policy<Derived>& policy, InputIt first, InputIt last,
 
     return result + n;
   }
+  else if constexpr (::cuda::__is_zip_transform_iterator<InputIt>)
+  {
+    const auto n = ::cuda::std::distance(first, last);
+    return cuda_cub::__transform::cub_transform_many(
+      policy,
+      ::cuda::std::move(first).__base(),
+      result,
+      n,
+      ::cuda::std::move(first).__pred(),
+      cub::detail::transform::always_true_predicate{});
+  }
   else
   {
     return cuda_cub::transform(

cuda/cccl/parallel/experimental/__init__.py

@@ -1,77 +1,24 @@
-# Copyright (c) 2024, NVIDIA CORPORATION & AFFILIATES. ALL RIGHTS RESERVED.
+# Copyright (c) 2025, NVIDIA CORPORATION.
 #
-# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+# 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.
 
-from .algorithms import (
-    DoubleBuffer,
-    SortOrder,
-    binary_transform,
-    exclusive_scan,
-    histogram_even,
-    inclusive_scan,
-    make_binary_transform,
-    make_exclusive_scan,
-    make_histogram_even,
-    make_inclusive_scan,
-    make_merge_sort,
-    make_radix_sort,
-    make_reduce_into,
-    make_segmented_reduce,
-    make_three_way_partition,
-    make_unary_transform,
-    make_unique_by_key,
-    merge_sort,
-    radix_sort,
-    reduce_into,
-    segmented_reduce,
-    three_way_partition,
-    unary_transform,
-    unique_by_key,
-)
-from .iterators import (
-    CacheModifiedInputIterator,
-    ConstantIterator,
-    CountingIterator,
-    ReverseIterator,
-    TransformIterator,
-    TransformOutputIterator,
-    ZipIterator,
-)
-from .op import OpKind
-from .struct import gpu_struct
+# alias for backwards compatibility
 
-__all__ = [
-    "binary_transform",
-    "CacheModifiedInputIterator",
-    "ConstantIterator",
-    "CountingIterator",
-    "DoubleBuffer",
-    "exclusive_scan",
-    "gpu_struct",
-    "histogram_even",
-    "inclusive_scan",
-    "make_binary_transform",
-    "make_exclusive_scan",
-    "make_histogram_even",
-    "make_inclusive_scan",
-    "make_merge_sort",
-    "make_radix_sort",
-    "make_reduce_into",
-    "make_segmented_reduce",
-    "make_three_way_partition",
-    "make_unary_transform",
-    "make_unique_by_key",
-    "merge_sort",
-    "OpKind",
-    "radix_sort",
-    "reduce_into",
-    "ReverseIterator",
-    "segmented_reduce",
-    "SortOrder",
-    "TransformIterator",
-    "three_way_partition",
-    "TransformOutputIterator",
-    "unary_transform",
-    "unique_by_key",
-    "ZipIterator",
-]
+from warnings import warn
+
+from cuda.compute import *  # noqa: F403
+
+warn(
+    "The module cuda.cccl.parallel.experimental is deprecated. Use cuda.compute instead.",
+    FutureWarning,
+)

cuda/compute/__init__.py

@@ -0,0 +1,77 @@
+# Copyright (c) 2024, NVIDIA CORPORATION & AFFILIATES. ALL RIGHTS RESERVED.
+#
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+from .algorithms import (
+    DoubleBuffer,
+    SortOrder,
+    binary_transform,
+    exclusive_scan,
+    histogram_even,
+    inclusive_scan,
+    make_binary_transform,
+    make_exclusive_scan,
+    make_histogram_even,
+    make_inclusive_scan,
+    make_merge_sort,
+    make_radix_sort,
+    make_reduce_into,
+    make_segmented_reduce,
+    make_three_way_partition,
+    make_unary_transform,
+    make_unique_by_key,
+    merge_sort,
+    radix_sort,
+    reduce_into,
+    segmented_reduce,
+    three_way_partition,
+    unary_transform,
+    unique_by_key,
+)
+from .iterators import (
+    CacheModifiedInputIterator,
+    ConstantIterator,
+    CountingIterator,
+    ReverseIterator,
+    TransformIterator,
+    TransformOutputIterator,
+    ZipIterator,
+)
+from .op import OpKind
+from .struct import gpu_struct
+
+__all__ = [
+    "binary_transform",
+    "CacheModifiedInputIterator",
+    "ConstantIterator",
+    "CountingIterator",
+    "DoubleBuffer",
+    "exclusive_scan",
+    "gpu_struct",
+    "histogram_even",
+    "inclusive_scan",
+    "make_binary_transform",
+    "make_exclusive_scan",
+    "make_histogram_even",
+    "make_inclusive_scan",
+    "make_merge_sort",
+    "make_radix_sort",
+    "make_reduce_into",
+    "make_segmented_reduce",
+    "make_three_way_partition",
+    "make_unary_transform",
+    "make_unique_by_key",
+    "merge_sort",
+    "OpKind",
+    "radix_sort",
+    "reduce_into",
+    "ReverseIterator",
+    "segmented_reduce",
+    "SortOrder",
+    "TransformIterator",
+    "TransformOutputIterator",
+    "three_way_partition",
+    "unary_transform",
+    "unique_by_key",
+    "ZipIterator",
+]

cuda/{cccl/parallel/experimental → compute}/_bindings_impl.pyx

@@ -4,7 +4,7 @@
 
 # Python signatures are declared in the companion Python stub file _bindings.pyi
 # Make sure to update PYI with change to Python API to ensure that Python
-# static type checker tools like mypy green-lights cuda.cccl.parallel
+# static type checker tools like mypy green-lights cuda.compute
 
 from libc.string cimport memset, memcpy
 from libc.stdint cimport uint8_t, uint32_t, uint64_t, int64_t, uintptr_t

cuda/{cccl/parallel/experimental → compute}/algorithms/_histogram.py

@@ -148,7 +148,7 @@ def make_histogram_even(
     Example:
         Below, ``make_histogram_even`` is used to create a histogram object that can be reused.
 
-        .. literalinclude:: ../../python/cuda_cccl/tests/parallel/examples/histogram/histogram_object.py
+        .. literalinclude:: ../../python/cuda_cccl/tests/compute/examples/histogram/histogram_object.py
           :language: python
           :start-after: # example-begin
 
@@ -190,7 +190,7 @@ def histogram_even(
     Example:
         Below, ``histogram_even`` is used to compute a histogram with evenly-spaced bins.
 
-        .. literalinclude:: ../../python/cuda_cccl/tests/parallel/examples/histogram/histogram_even_basic.py
+        .. literalinclude:: ../../python/cuda_cccl/tests/compute/examples/histogram/histogram_even_basic.py
             :language: python
             :start-after: # example-begin
             :caption: Basic histogram example.