cuda-cccl 0.3.2__cp313-cp313-manylinux_2_24_aarch64.whl → 0.3.4__cp313-cp313-manylinux_2_24_aarch64.whl

cuda/cccl/headers/include/thrust/mr/pool.h

@@ -37,6 +37,8 @@
 #include <thrust/mr/memory_resource.h>
 #include <thrust/mr/pool_options.h>
 
+#include <cuda/__cmath/ilog.h>
+#include <cuda/__cmath/pow2.h>
 #include <cuda/std/__cccl/algorithm_wrapper.h>
 #include <cuda/std/cassert>
 #include <cuda/std/cstdint>
@@ -44,7 +46,6 @@
 THRUST_NAMESPACE_BEGIN
 namespace mr
 {
-
 /** \addtogroup memory_resources Memory Resources
  *  \ingroup memory_management
  *  \{
@@ -112,7 +113,7 @@ public:
   unsynchronized_pool_resource(Upstream* upstream, pool_options options = get_default_options())
       : m_upstream(upstream)
       , m_options(options)
-      , m_smallest_block_log2(detail::log2_ri(m_options.smallest_block_size))
+      , m_smallest_block_log2(::cuda::ceil_ilog2(m_options.smallest_block_size))
       , m_pools(upstream)
       , m_allocated()
       , m_oversized()
@@ -121,7 +122,7 @@ public:
     assert(m_options.validate());
 
     pool p = {block_descriptor_ptr(), 0};
-    m_pools.resize(detail::log2_ri(m_options.largest_block_size) - m_smallest_block_log2 + 1, p);
+    m_pools.resize(::cuda::ceil_ilog2(m_options.largest_block_size) - m_smallest_block_log2 + 1, p);
   }
 
   // TODO: C++11: use delegating constructors
@@ -133,7 +134,7 @@ public:
   unsynchronized_pool_resource(pool_options options = get_default_options())
       : m_upstream(get_global_resource<Upstream>())
       , m_options(options)
-      , m_smallest_block_log2(detail::log2_ri(m_options.smallest_block_size))
+      , m_smallest_block_log2(::cuda::ceil_ilog2(m_options.smallest_block_size))
       , m_pools(get_global_resource<Upstream>())
       , m_allocated()
       , m_oversized()
@@ -142,7 +143,7 @@ public:
     assert(m_options.validate());
 
     pool p = {block_descriptor_ptr(), 0};
-    m_pools.resize(detail::log2_ri(m_options.largest_block_size) - m_smallest_block_log2 + 1, p);
+    m_pools.resize(::cuda::ceil_ilog2(m_options.largest_block_size) - m_smallest_block_log2 + 1, p);
   }
 
   /*! Destructor. Releases all held memory to upstream.
@@ -260,7 +261,7 @@ public:
   do_allocate(std::size_t bytes, std::size_t alignment = THRUST_MR_DEFAULT_ALIGNMENT) override
   {
     bytes = (std::max) (bytes, m_options.smallest_block_size);
-    assert(detail::is_power_of_2(alignment));
+    assert(::cuda::is_power_of_two(alignment));
 
     // an oversized and/or overaligned allocation requested; needs to be allocated separately
     if (bytes > m_options.largest_block_size || alignment > m_options.alignment)
@@ -371,7 +372,7 @@ public:
 
     // the request is NOT for oversized and/or overaligned memory
     // allocate a block from an appropriate bucket
-    std::size_t bytes_log2 = thrust::detail::log2_ri(bytes);
+    std::size_t bytes_log2 = ::cuda::ceil_ilog2(bytes);
     std::size_t bucket_idx = bytes_log2 - m_smallest_block_log2;
     pool& bucket           = thrust::raw_reference_cast(m_pools[bucket_idx]);
 
@@ -439,7 +440,7 @@ public:
   virtual void do_deallocate(void_ptr p, std::size_t n, std::size_t alignment = THRUST_MR_DEFAULT_ALIGNMENT) override
   {
     n = (std::max) (n, m_options.smallest_block_size);
-    assert(detail::is_power_of_2(alignment));
+    assert(::cuda::is_power_of_two(alignment));
 
     // verify that the pointer is at least as aligned as claimed
     assert(reinterpret_cast<::cuda::std::intmax_t>(void_ptr_traits::get(p)) % alignment == 0);
@@ -504,7 +505,7 @@ public:
     }
 
     // push the block to the front of the appropriate bucket's free list
-    std::size_t n_log2     = thrust::detail::log2_ri(n);
+    std::size_t n_log2     = ::cuda::ceil_ilog2(n);
     std::size_t bucket_idx = n_log2 - m_smallest_block_log2;
     pool& bucket           = thrust::raw_reference_cast(m_pools[bucket_idx]);
 
@@ -521,6 +522,5 @@ public:
 
 /*! \} // memory_resources
  */
-
 } // namespace mr
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/mr/pool_options.h

@@ -32,14 +32,13 @@
 #endif // no system header
 
 #include <thrust/detail/config/memory_resource.h>
-#include <thrust/detail/integer_math.h>
 
+#include <cuda/__cmath/pow2.h>
 #include <cuda/std/cstddef>
 
 THRUST_NAMESPACE_BEGIN
 namespace mr
 {
-
 /*! \addtogroup memory_resources Memory Resources
  *  \ingroup memory_management
  *  \{
@@ -104,15 +103,15 @@ struct pool_options
    */
   bool validate() const
   {
-    if (!detail::is_power_of_2(smallest_block_size))
+    if (smallest_block_size != 0 && !::cuda::is_power_of_two(smallest_block_size))
     {
       return false;
     }
-    if (!detail::is_power_of_2(largest_block_size))
+    if (largest_block_size != 0 && !::cuda::is_power_of_two(largest_block_size))
     {
       return false;
     }
-    if (!detail::is_power_of_2(alignment))
+    if (alignment != 0 && !::cuda::is_power_of_two(alignment))
     {
       return false;
     }
@@ -169,6 +168,5 @@ struct pool_options
 
 /*! \} // memory_resources
  */
-
 } // namespace mr
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/mr/sync_pool.h

@@ -36,7 +36,6 @@
 THRUST_NAMESPACE_BEGIN
 namespace mr
 {
-
 /*! \addtogroup memory_resources Memory Resources
  *  \ingroup memory_management
  *  \{
@@ -109,6 +108,5 @@ private:
 
 /*! \} // memory_resources
  */
-
 } // namespace mr
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/mr/tls_pool.h

@@ -34,7 +34,6 @@
 THRUST_NAMESPACE_BEGIN
 namespace mr
 {
-
 /*! \addtogroup memory_resources Memory Resources
  *  \ingroup memory_management
  *  \{
@@ -59,6 +58,5 @@ _CCCL_HOST thrust::mr::unsynchronized_pool_resource<Upstream>& tls_pool(Upstream
 
 /*! \}
  */
-
 } // namespace mr
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/mr/validator.h

@@ -34,7 +34,6 @@
 THRUST_NAMESPACE_BEGIN
 namespace mr
 {
-
 template <typename MR>
 struct validator
 {
@@ -51,6 +50,5 @@ struct validator2
 template <typename T>
 struct validator2<T, T> : private validator<T>
 {};
-
 } // namespace mr
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/per_device_resource.h

@@ -27,8 +27,20 @@
 #endif // no system header
 #include <thrust/detail/execution_policy.h>
 #include <thrust/mr/allocator.h>
-#include <thrust/system/detail/adl/per_device_resource.h>
+
+// Include all active backend system implementations (generic, sequential, host and device)
 #include <thrust/system/detail/generic/per_device_resource.h>
+#include <thrust/system/detail/sequential/per_device_resource.h>
+#include __THRUST_HOST_SYSTEM_ALGORITH_DETAIL_HEADER_INCLUDE(per_device_resource.h)
+#include __THRUST_DEVICE_SYSTEM_ALGORITH_DETAIL_HEADER_INCLUDE(per_device_resource.h)
+
+// Some build systems need a hint to know which files we actually include
+#if 0
+#  include <thrust/system/cpp/detail/per_device_resource.h>
+#  include <thrust/system/cuda/detail/per_device_resource.h>
+#  include <thrust/system/omp/detail/per_device_resource.h>
+#  include <thrust/system/tbb/detail/per_device_resource.h>
+#endif
 
 THRUST_NAMESPACE_BEGIN
 

cuda/cccl/headers/include/thrust/random/detail/discard_block_engine.inl

@@ -32,7 +32,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 template <typename Engine, size_t p, size_t r>
 _CCCL_HOST_DEVICE discard_block_engine<Engine, p, r>::discard_block_engine()
     : m_e()
@@ -178,7 +177,6 @@ operator!=(const discard_block_engine<Engine, p, r>& lhs, const discard_block_en
 {
   return !(lhs == rhs);
 }
-
 } // namespace random
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/linear_congruential_engine.inl

@@ -34,7 +34,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 template <typename UIntType, UIntType a, UIntType c, UIntType m>
 _CCCL_HOST_DEVICE linear_congruential_engine<UIntType, a, c, m>::linear_congruential_engine(result_type s)
 {
@@ -149,7 +148,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, linear_congruential_engine<UIn
 {
   return detail::random_core_access::stream_in(is, e);
 }
-
 } // namespace random
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/linear_congruential_engine_discard.h

@@ -32,12 +32,8 @@
 
 THRUST_NAMESPACE_BEGIN
 
-namespace random
+namespace random::detail
 {
-
-namespace detail
-{
-
 template <typename UIntType, UIntType a, unsigned long long c, UIntType m>
 struct linear_congruential_engine_discard_implementation
 {
@@ -96,9 +92,6 @@ struct linear_congruential_engine_discard
     linear_congruential_engine_discard_implementation<result_type, a, c, m>::discard(lcg.m_x, z);
   }
 }; // end linear_congruential_engine_discard
-
-} // namespace detail
-
-} // namespace random
+} // namespace random::detail
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/linear_feedback_shift_engine.inl

@@ -32,7 +32,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 template <typename UIntType, size_t w, size_t k, size_t q, size_t s>
 _CCCL_HOST_DEVICE linear_feedback_shift_engine<UIntType, w, k, q, s>::linear_feedback_shift_engine(result_type value)
 {
@@ -145,7 +144,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, linear_feedback_shift_engine<U
 {
   return thrust::random::detail::random_core_access::stream_in(is, e);
 }
-
 } // namespace random
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/linear_feedback_shift_engine_wordmask.h

@@ -28,12 +28,8 @@
 
 THRUST_NAMESPACE_BEGIN
 
-namespace random
+namespace random::detail
 {
-
-namespace detail
-{
-
 template <typename T, int w, int i = w - 1>
 struct linear_feedback_shift_engine_wordmask
 {
@@ -45,9 +41,6 @@ struct linear_feedback_shift_engine_wordmask<T, w, 0>
 {
   static const T value = 0;
 }; // end linear_feedback_shift_engine_wordmask
-
-} // namespace detail
-
-} // namespace random
+} // namespace random::detail
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/mod.h

@@ -28,12 +28,8 @@
 
 THRUST_NAMESPACE_BEGIN
 
-namespace random
+namespace random::detail
 {
-
-namespace detail
-{
-
 template <typename T, T a, T c, T m, bool = (m == 0)>
 struct static_mod
 {
@@ -93,9 +89,6 @@ _CCCL_HOST_DEVICE T mod(T x)
   static_mod<T, a, c, m> f;
   return f(x);
 } // end static_mod
-
-} // namespace detail
-
-} // namespace random
+} // namespace random::detail
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/normal_distribution.inl

@@ -36,7 +36,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 template <typename RealType>
 _CCCL_HOST_DEVICE normal_distribution<RealType>::normal_distribution(RealType a, RealType b)
     : super_t()
@@ -181,7 +180,6 @@ std::basic_istream<CharT, Traits>& operator>>(std::basic_istream<CharT, Traits>&
 {
   return thrust::random::detail::random_core_access::stream_in(is, d);
 }
-
 } // namespace random
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/normal_distribution_base.h

@@ -39,11 +39,8 @@
 #include <cuda/std/limits>
 
 THRUST_NAMESPACE_BEGIN
-namespace random
+namespace random::detail
 {
-namespace detail
-{
-
 // this version samples the normal distribution directly
 // and uses the non-standard math function erfcinv
 template <typename RealType>
@@ -154,7 +151,5 @@ struct normal_distribution_base
   using type = normal_distribution_portable<RealType>;
 #endif
 };
-
-} // namespace detail
-} // namespace random
+} // namespace random::detail
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/random_core_access.h

@@ -28,12 +28,8 @@
 
 THRUST_NAMESPACE_BEGIN
 
-namespace random
+namespace random::detail
 {
-
-namespace detail
-{
-
 struct random_core_access
 {
   template <typename OStream, typename EngineOrDistribution>
@@ -55,9 +51,6 @@ struct random_core_access
   }
 
 }; // end random_core_access
-
-} // namespace detail
-
-} // namespace random
+} // namespace random::detail
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/subtract_with_carry_engine.inl

@@ -35,7 +35,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 template <typename UIntType, size_t w, size_t s, size_t r>
 _CCCL_HOST_DEVICE subtract_with_carry_engine<UIntType, w, s, r>::subtract_with_carry_engine(result_type value)
 {
@@ -195,7 +194,6 @@ _CCCL_HOST_DEVICE bool operator!=(const subtract_with_carry_engine<UIntType, w,
 {
   return !(lhs == rhs);
 }
-
 } // namespace random
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/uniform_int_distribution.inl

@@ -34,7 +34,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 template <typename IntType>
 _CCCL_HOST_DEVICE uniform_int_distribution<IntType>::uniform_int_distribution(IntType a, IntType b)
     : m_param(a, b)
@@ -192,7 +191,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, uniform_int_distribution<IntTy
 {
   return thrust::random::detail::random_core_access::stream_in(is, d);
 }
-
 } // namespace random
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/uniform_real_distribution.inl

@@ -34,7 +34,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 template <typename RealType>
 _CCCL_HOST_DEVICE uniform_real_distribution<RealType>::uniform_real_distribution(RealType a, RealType b)
     : m_param(a, b)
@@ -194,7 +193,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, uniform_real_distribution<Real
 {
   return thrust::random::detail::random_core_access::stream_in(is, d);
 }
-
 } // namespace random
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/detail/xor_combine_engine.inl

@@ -33,7 +33,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 template <typename Engine1, size_t s1, typename Engine2, size_t s2>
 _CCCL_HOST_DEVICE xor_combine_engine<Engine1, s1, Engine2, s2>::xor_combine_engine()
     : m_b1()
@@ -177,7 +176,6 @@ _CCCL_HOST_DEVICE bool operator!=(const xor_combine_engine<Engine1, s1, Engine2,
 {
   return !(lhs == rhs);
 }
-
 } // namespace random
 
 THRUST_NAMESPACE_END

cuda/cccl/headers/include/thrust/random/discard_block_engine.h

@@ -41,7 +41,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 /*! \addtogroup random_number_engine_adaptors Random Number Engine Adaptor Class Templates
  *  \ingroup random
  *  \{
@@ -229,7 +228,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, discard_block_engine<Engine, p
 
 /*! \} // end random_number_engine_adaptors
  */
-
 } // namespace random
 
 // import names into thrust::

cuda/cccl/headers/include/thrust/random/linear_congruential_engine.h

@@ -40,7 +40,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 /*! \addtogroup random_number_engine_templates Random Number Engine Class Templates
  *  \ingroup random
  *  \{
@@ -276,7 +275,6 @@ using minstd_rand = linear_congruential_engine<std::uint32_t, 48271, 0, 21474836
 
 /*! \} // predefined_random
  */
-
 } // namespace random
 
 // import names into thrust::

cuda/cccl/headers/include/thrust/random/linear_feedback_shift_engine.h

@@ -49,7 +49,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 /*! \addtogroup random_number_engine_templates
  *  \{
  */
@@ -206,7 +205,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, linear_feedback_shift_engine<U
 
 /*! \} // end random_number_engine_templates
  */
-
 } // namespace random
 
 // import names into thrust::

cuda/cccl/headers/include/thrust/random/normal_distribution.h

@@ -39,7 +39,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 /*! \addtogroup random_number_distributions
  *  \{
  */
@@ -247,7 +246,6 @@ std::basic_istream<CharT, Traits>& operator>>(std::basic_istream<CharT, Traits>&
 
 /*! \} // end random_number_distributions
  */
-
 } // namespace random
 
 using random::normal_distribution;

cuda/cccl/headers/include/thrust/random/subtract_with_carry_engine.h

@@ -42,7 +42,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 /*! \addtogroup random_number_engine_templates
  *  \{
  */
@@ -234,7 +233,6 @@ using ranlux48_base = subtract_with_carry_engine<std::uint64_t, 48, 5, 12>;
 
 /*! \} // end predefined_random
  */
-
 } // namespace random
 
 // import names into thrust::

cuda/cccl/headers/include/thrust/random/uniform_int_distribution.h

@@ -40,7 +40,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 /*! \addtogroup random_number_distributions Random Number Distributions Class Templates
  *  \ingroup random
  *  \{
@@ -251,7 +250,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, uniform_int_distribution<IntTy
 
 /*! \} // end random_number_distributions
  */
-
 } // namespace random
 
 using random::uniform_int_distribution;

cuda/cccl/headers/include/thrust/random/uniform_real_distribution.h

@@ -38,7 +38,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 /*! \addtogroup random_number_distributions
  *  \{
  */
@@ -248,7 +247,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, uniform_real_distribution<Real
 
 /*! \} // end random_number_distributions
  */
-
 } // namespace random
 
 using random::uniform_real_distribution;

cuda/cccl/headers/include/thrust/random/xor_combine_engine.h

@@ -45,7 +45,6 @@ THRUST_NAMESPACE_BEGIN
 
 namespace random
 {
-
 /*! \addtogroup random_number_engine_adaptors
  *  \{
  */
@@ -244,7 +243,6 @@ operator>>(std::basic_istream<CharT, Traits>& is, xor_combine_engine<Engine1_, s
 
 /*! \} // end random_number_engine_adaptors
  */
-
 } // namespace random
 
 // import names into thrust::

cuda/cccl/headers/include/thrust/random.h

@@ -58,7 +58,6 @@ THRUST_NAMESPACE_BEGIN
  */
 namespace random
 {
-
 /*! \addtogroup predefined_random Random Number Engines with Predefined Parameters
  *  \ingroup random
  *  \{
@@ -105,7 +104,6 @@ using default_random_engine = minstd_rand;
 
 /*! \} // end predefined_random
  */
-
 } // namespace random
 
 /*! \} // end random

cuda/cccl/headers/include/thrust/system/cpp/detail/execution_policy.h

@@ -19,16 +19,11 @@
 THRUST_NAMESPACE_BEGIN
 namespace system::cpp
 {
-//! \addtogroup execution_policies
-//! \{
-
-//! \p thrust::cpp::tag is a type representing Thrust's standard C++ backend system in C++'s type system.
-//! Iterators "tagged" with a type which is convertible to \p cpp::tag assert that they may be
-//! "dispatched" to algorithm implementations in the \p cpp system.
+namespace detail
+{
+// note: the tag and execution policy need to be defined in the same namespace as the algorithms for ADL to find them
 struct tag;
 
-//! \p thrust::cpp::execution_policy is the base class for all Thrust parallel execution
-//! policies which are derived from Thrust's standard C++ backend system.
 template <typename DerivedPolicy>
 struct execution_policy;
 
@@ -53,14 +48,24 @@ struct execution_policy : system::detail::sequential::execution_policy<Derived>
   }
 };
 
-namespace detail
-{
 struct par_t
     : execution_policy<par_t>
     , thrust::detail::allocator_aware_execution_policy<execution_policy>
 {};
 } // namespace detail
 
+//! \addtogroup execution_policies
+//! \{
+
+//! \p thrust::cpp::tag is a type representing Thrust's standard C++ backend system in C++'s type system.
+//! Iterators "tagged" with a type which is convertible to \p cpp::tag assert that they may be
+//! "dispatched" to algorithm implementations in the \p cpp system.
+using detail::tag;
+
+//! \p thrust::cpp::execution_policy is the base class for all Thrust parallel execution
+//! policies which are derived from Thrust's standard C++ backend system.
+using detail::execution_policy;
+
 //! \p thrust::system::cpp::par is the parallel execution policy associated with Thrust's standard C++ backend system.
 //!
 //! Instead of relying on implicit algorithm dispatch through iterator system tags, users may directly target Thrust's
@@ -96,7 +101,6 @@ struct par_t
 _CCCL_GLOBAL_CONSTANT detail::par_t par;
 
 //! \}
-
 } // namespace system::cpp
 
 // aliases: