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/cub/util_type.cuh

@@ -1,30 +1,6 @@
-/******************************************************************************
- * Copyright (c) 2011, Duane Merrill.  All rights reserved.
- * Copyright (c) 2011-2024, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2011, Duane Merrill. All rights reserved.
+// SPDX-FileCopyrightText: Copyright (c) 2011-2024, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * \file
@@ -158,9 +134,10 @@ struct Log2<N, 0, COUNT>
 
 /**
  * \brief Statically determine if N is a power-of-two
+ * deprecated [since 3.2]
  */
 template <int N>
-struct PowerOfTwo
+struct [[deprecated("Use cuda::is_power_of_two(N) instead")]] PowerOfTwo
 {
   enum
   {
@@ -208,7 +185,6 @@ struct NullType
 
 namespace detail
 {
-
 template <bool Value>
 inline constexpr auto bool_constant_v = ::cuda::std::bool_constant<Value>{};
 
@@ -217,7 +193,6 @@ using constant_t = ::cuda::std::integral_constant<decltype(Value), Value>;
 
 template <auto Value>
 inline constexpr auto constant_v = constant_t<Value>{};
-
 } // namespace detail
 
 /**
@@ -269,7 +244,6 @@ FutureValue(IterT) -> FutureValue<detail::it_value_t<IterT>, IterT>;
 
 namespace detail
 {
-
 /**
  * \brief Allows algorithms to instantiate a single kernel to support both immediate value and future value.
  */
@@ -315,7 +289,6 @@ private:
     T m_immediate_value;
   };
 };
-
 } // namespace detail
 
 /******************************************************************************

cuda/cccl/headers/include/cub/util_vsmem.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 2023-24, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2023-2024, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * \file
@@ -58,7 +34,6 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail
 {
-
 /**
  * @brief Helper struct to wrap all the information needed to implement virtual shared memory that's passed to a kernel.
  *
@@ -245,7 +220,6 @@ using vsmem_helper_fallback_policy_t =
 template <typename DefaultPolicyT, template <typename...> class AgentT, typename... AgentParamsT>
 using vsmem_helper_default_fallback_policy_t =
   vsmem_helper_fallback_policy_t<DefaultPolicyT, policy_wrapper_t<DefaultPolicyT, 64, 1>, AgentT, AgentParamsT...>;
-
 } // namespace detail
 
 #endif // _CCCL_DOXYGEN_INVOKED

cuda/cccl/headers/include/cub/version.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 2011-2022, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2011-2022, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /*! \file version.cuh
  *  \brief Compile-time macros encoding CUB release version

cuda/cccl/headers/include/cub/warp/specializations/warp_exchange_shfl.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 2011-2023, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2011-2023, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 #pragma once
 
@@ -40,17 +16,17 @@
 #include <cub/util_ptx.cuh>
 #include <cub/util_type.cuh>
 
+#include <cuda/__cmath/pow2.h>
 #include <cuda/__ptx/instructions/get_sreg.h>
 
 CUB_NAMESPACE_BEGIN
 
 namespace detail
 {
-
 template <typename InputT, int ITEMS_PER_THREAD, int LOGICAL_WARP_THREADS = warp_threads>
 class WarpExchangeShfl
 {
-  static_assert(PowerOfTwo<LOGICAL_WARP_THREADS>::VALUE, "LOGICAL_WARP_THREADS must be a power of two");
+  static_assert(::cuda::is_power_of_two(LOGICAL_WARP_THREADS), "LOGICAL_WARP_THREADS must be a power of two");
 
   static_assert(ITEMS_PER_THREAD == LOGICAL_WARP_THREADS,
                 "WARP_EXCHANGE_SHUFFLE currently only works when ITEMS_PER_THREAD == "
@@ -217,16 +193,16 @@ class WarpExchangeShfl
       CompileTimeArray<OutputT, next_idx, SIZE>::template Foreach<NUM_ENTRIES>(xor_bit_set, mask);
     }
 
-    // terminate recursion
-    _CCCL_DEVICE void TransposeImpl(unsigned int, unsigned int, constant_t<0>) {}
-
     template <int NUM_ENTRIES>
     _CCCL_DEVICE void TransposeImpl(const unsigned int lane_id, const unsigned int mask, constant_t<NUM_ENTRIES>)
     {
-      const bool xor_bit_set = lane_id & NUM_ENTRIES;
-      Foreach<NUM_ENTRIES>(xor_bit_set, mask);
+      if constexpr (NUM_ENTRIES != 0)
+      {
+        const bool xor_bit_set = lane_id & NUM_ENTRIES;
+        Foreach<NUM_ENTRIES>(xor_bit_set, mask);
 
-      TransposeImpl(lane_id, mask, constant_v<NUM_ENTRIES / 2>);
+        TransposeImpl(lane_id, mask, constant_v<NUM_ENTRIES / 2>);
+      }
     }
 
   public:
@@ -323,7 +299,6 @@ public:
                   "                 OffsetT (&ranks)[ITEMS_PER_THREAD])");
   }
 };
-
 } // namespace detail
 
 CUB_NAMESPACE_END

cuda/cccl/headers/include/cub/warp/specializations/warp_exchange_smem.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 2011-2023, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2011-2023, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file
@@ -46,17 +22,17 @@
 #include <cub/util_ptx.cuh>
 #include <cub/util_type.cuh>
 
+#include <cuda/__cmath/pow2.h>
 #include <cuda/__ptx/instructions/get_sreg.h>
 
 CUB_NAMESPACE_BEGIN
 
 namespace detail
 {
-
 template <typename InputT, int ITEMS_PER_THREAD, int LOGICAL_WARP_THREADS = warp_threads>
 class WarpExchangeSmem
 {
-  static_assert(PowerOfTwo<LOGICAL_WARP_THREADS>::VALUE, "LOGICAL_WARP_THREADS must be a power of two");
+  static_assert(::cuda::is_power_of_two(LOGICAL_WARP_THREADS), "LOGICAL_WARP_THREADS must be a power of two");
 
   static constexpr int ITEMS_PER_TILE = ITEMS_PER_THREAD * LOGICAL_WARP_THREADS + 1;
 
@@ -66,7 +42,7 @@ class WarpExchangeSmem
 
   // Insert padding if the number of items per thread is a power of two
   // and > 4 (otherwise we can typically use 128b loads)
-  static constexpr bool INSERT_PADDING = (ITEMS_PER_THREAD > 4) && (PowerOfTwo<ITEMS_PER_THREAD>::VALUE);
+  static constexpr bool INSERT_PADDING = (ITEMS_PER_THREAD > 4) && (::cuda::is_power_of_two(ITEMS_PER_THREAD));
 
   static constexpr int PADDING_ITEMS = INSERT_PADDING ? (ITEMS_PER_TILE >> LOG_SMEM_BANKS) : 0;
 
@@ -171,7 +147,6 @@ public:
     }
   }
 };
-
 } // namespace detail
 
 CUB_NAMESPACE_END

cuda/cccl/headers/include/cub/warp/specializations/warp_reduce_shfl.cuh

@@ -1,30 +1,6 @@
-/******************************************************************************
- * Copyright (c) 2011, Duane Merrill.  All rights reserved.
- * Copyright (c) 2011-2018, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2011, Duane Merrill. All rights reserved.
+// SPDX-FileCopyrightText: Copyright (c) 2011-2018, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file
@@ -48,6 +24,7 @@
 #include <cub/util_ptx.cuh>
 #include <cub/util_type.cuh>
 
+#include <cuda/__cmath/pow2.h>
 #include <cuda/__functional/maximum.h>
 #include <cuda/__functional/minimum.h>
 #include <cuda/__ptx/instructions/get_sreg.h>
@@ -66,7 +43,6 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail
 {
-
 template <class A = int, class = A>
 struct reduce_add_exists : ::cuda::std::false_type
 {};
@@ -104,7 +80,7 @@ struct reduce_max_exists<T, decltype(__reduce_max_sync(0xFFFFFFFF, T{}))> : ::cu
 template <typename T, int LOGICAL_WARP_THREADS>
 struct WarpReduceShfl
 {
-  static_assert(PowerOfTwo<LOGICAL_WARP_THREADS>::VALUE, "LOGICAL_WARP_THREADS must be a power of two");
+  static_assert(::cuda::is_power_of_two(LOGICAL_WARP_THREADS), "LOGICAL_WARP_THREADS must be a power of two");
 
   //---------------------------------------------------------------------
   // Constants and type definitions
@@ -452,12 +428,12 @@ struct WarpReduceShfl
    * @param[in] offset
    *   Up-offset to pull from
    */
-  template <typename _T, typename ReductionOp>
-  _CCCL_DEVICE _CCCL_FORCEINLINE _T ReduceStep(_T input, ReductionOp reduction_op, int last_lane, int offset)
+  template <typename _Tp, typename ReductionOp>
+  _CCCL_DEVICE _CCCL_FORCEINLINE _Tp ReduceStep(_Tp input, ReductionOp reduction_op, int last_lane, int offset)
   {
-    _T output = input;
+    _Tp output = input;
 
-    _T temp = ShuffleDown<LOGICAL_WARP_THREADS>(output, offset, last_lane, member_mask);
+    _Tp temp = ShuffleDown<LOGICAL_WARP_THREADS>(output, offset, last_lane, member_mask);
 
     // Perform reduction op if valid
     if (offset + lane_id <= last_lane)
@@ -486,9 +462,9 @@ struct WarpReduceShfl
    * @param[in] is_small_unsigned
    *   Marker type indicating whether T is a small unsigned integer
    */
-  template <typename _T, typename ReductionOp>
-  _CCCL_DEVICE _CCCL_FORCEINLINE _T ReduceStep(
-    _T input, ReductionOp reduction_op, int last_lane, int offset, ::cuda::std::true_type /*is_small_unsigned*/)
+  template <typename _Tp, typename ReductionOp>
+  _CCCL_DEVICE _CCCL_FORCEINLINE _Tp ReduceStep(
+    _Tp input, ReductionOp reduction_op, int last_lane, int offset, ::cuda::std::true_type /*is_small_unsigned*/)
   {
     return ReduceStep(input, reduction_op, last_lane, offset);
   }
@@ -512,9 +488,9 @@ struct WarpReduceShfl
    * @param[in] is_small_unsigned
    *   Marker type indicating whether T is a small unsigned integer
    */
-  template <typename _T, typename ReductionOp>
-  _CCCL_DEVICE _CCCL_FORCEINLINE _T ReduceStep(
-    _T input, ReductionOp reduction_op, int last_lane, int offset, ::cuda::std::false_type /*is_small_unsigned*/)
+  template <typename _Tp, typename ReductionOp>
+  _CCCL_DEVICE _CCCL_FORCEINLINE _Tp ReduceStep(
+    _Tp input, ReductionOp reduction_op, int last_lane, int offset, ::cuda::std::false_type /*is_small_unsigned*/)
   {
     return ReduceStep(input, reduction_op, last_lane, offset);
   }

cuda/cccl/headers/include/cub/warp/specializations/warp_reduce_smem.cuh

@@ -1,30 +1,6 @@
-/******************************************************************************
- * Copyright (c) 2011, Duane Merrill.  All rights reserved.
- * Copyright (c) 2011-2025, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2011, Duane Merrill. All rights reserved.
+// SPDX-FileCopyrightText: Copyright (c) 2011-2025, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file
@@ -48,6 +24,7 @@
 #include <cub/util_ptx.cuh>
 #include <cub/util_type.cuh>
 
+#include <cuda/__cmath/pow2.h>
 #include <cuda/__ptx/instructions/get_sreg.h>
 #include <cuda/std/__bit/countr.h>
 #include <cuda/std/__type_traits/integral_constant.h>
@@ -77,7 +54,7 @@ struct WarpReduceSmem
   static constexpr bool IS_ARCH_WARP = (LOGICAL_WARP_THREADS == warp_threads);
 
   /// Whether the logical warp size is a power-of-two
-  static constexpr bool IS_POW_OF_TWO = PowerOfTwo<LOGICAL_WARP_THREADS>::VALUE;
+  static constexpr bool IS_POW_OF_TWO = ::cuda::is_power_of_two(LOGICAL_WARP_THREADS);
 
   /// The number of warp reduction steps
   static constexpr int STEPS = Log2<LOGICAL_WARP_THREADS>::VALUE;
@@ -286,13 +263,6 @@ struct WarpReduceSmem
   _CCCL_DEVICE _CCCL_FORCEINLINE T
   SegmentedReduce(T input, FlagT flag, ReductionOp reduction_op, ::cuda::std::false_type /*has_ballot*/)
   {
-    enum
-    {
-      UNSET = 0x0, // Is initially unset
-      SET   = 0x1, // Is initially set
-      SEEN  = 0x2, // Has seen another head flag from a successor peer
-    };
-
     // Alias flags onto shared data storage
     SmemFlag* flag_storage = temp_storage.flags;
 

cuda/cccl/headers/include/cub/warp/specializations/warp_scan_shfl.cuh

@@ -1,30 +1,6 @@
-/******************************************************************************
- * Copyright (c) 2011, Duane Merrill.  All rights reserved.
- * Copyright (c) 2011-2018, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2011, Duane Merrill. All rights reserved.
+// SPDX-FileCopyrightText: Copyright (c) 2011-2018, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file
@@ -408,13 +384,13 @@ struct WarpScanShfl
    * @param[in] offset
    *   Up-offset to pull from
    */
-  template <typename _T, typename ScanOpT>
-  _CCCL_DEVICE _CCCL_FORCEINLINE _T InclusiveScanStep(_T input, ScanOpT scan_op, int first_lane, int offset)
+  template <typename _Tp, typename ScanOpT>
+  _CCCL_DEVICE _CCCL_FORCEINLINE _Tp InclusiveScanStep(_Tp input, ScanOpT scan_op, int first_lane, int offset)
   {
-    _T temp = ShuffleUp<LOGICAL_WARP_THREADS>(input, offset, first_lane, member_mask);
+    _Tp temp = ShuffleUp<LOGICAL_WARP_THREADS>(input, offset, first_lane, member_mask);
 
     // Perform scan op if from a valid peer
-    _T output = scan_op(temp, input);
+    _Tp output = scan_op(temp, input);
     if (static_cast<int>(lane_id) < first_lane + offset)
     {
       output = input;
@@ -441,19 +417,19 @@ struct WarpScanShfl
    * @param[in] offset
    *   Up-offset to pull from
    */
-  template <typename _T, typename ScanOpT>
-  _CCCL_DEVICE _CCCL_FORCEINLINE _T
-  InclusiveScanStepPartial(_T input, ScanOpT scan_op, int valid_items, int first_lane, int offset)
+  template <typename _Tp, typename ScanOpT>
+  _CCCL_DEVICE _CCCL_FORCEINLINE _Tp
+  InclusiveScanStepPartial(_Tp input, ScanOpT scan_op, int valid_items, int first_lane, int offset)
   {
     _CCCL_ASSERT((first_lane >= 0) && (first_lane <= static_cast<int>(lane_id)),
                  "first_lane must be in range [0, lane_id]");
     _CCCL_ASSERT((offset > 0) && (offset < LOGICAL_WARP_THREADS),
                  "offset must be in the range [1, LOGICAL_WARP_THREADS)");
     _CCCL_ASSERT(::cuda::std::has_single_bit(static_cast<unsigned>(offset)), "offset must be a power of two");
-    _T temp = ::cuda::device::warp_shuffle_up<LOGICAL_WARP_THREADS>(input, offset, member_mask);
+    _Tp temp = ::cuda::device::warp_shuffle_up<LOGICAL_WARP_THREADS>(input, offset, member_mask);
 
     // Perform scan op if from a valid peer
-    _T output = input;
+    _Tp output = input;
     if (static_cast<int>(lane_id) >= first_lane + offset && static_cast<int>(lane_id) < valid_items)
     {
       output = scan_op(temp, input);
@@ -479,9 +455,9 @@ struct WarpScanShfl
    * @param[in] is_small_unsigned
    *   Marker type indicating whether T is a small integer
    */
-  template <typename _T, typename ScanOpT>
-  _CCCL_DEVICE _CCCL_FORCEINLINE _T
-  InclusiveScanStep(_T input, ScanOpT scan_op, int first_lane, int offset, ::cuda::std::true_type /*is_small_unsigned*/)
+  template <typename _Tp, typename ScanOpT>
+  _CCCL_DEVICE _CCCL_FORCEINLINE _Tp InclusiveScanStep(
+    _Tp input, ScanOpT scan_op, int first_lane, int offset, ::cuda::std::true_type /*is_small_unsigned*/)
   {
     return InclusiveScanStep(input, scan_op, first_lane, offset);
   }
@@ -505,9 +481,9 @@ struct WarpScanShfl
    * @param[in] is_small_unsigned
    *   Marker type indicating whether T is a small integer
    */
-  template <typename _T, typename ScanOpT>
-  _CCCL_DEVICE _CCCL_FORCEINLINE _T InclusiveScanStep(
-    _T input, ScanOpT scan_op, int first_lane, int offset, ::cuda::std::false_type /*is_small_unsigned*/)
+  template <typename _Tp, typename ScanOpT>
+  _CCCL_DEVICE _CCCL_FORCEINLINE _Tp InclusiveScanStep(
+    _Tp input, ScanOpT scan_op, int first_lane, int offset, ::cuda::std::false_type /*is_small_unsigned*/)
   {
     return InclusiveScanStep(input, scan_op, first_lane, offset);
   }
@@ -550,8 +526,8 @@ struct WarpScanShfl
    * @param[in] scan_op
    *   Binary scan operator
    */
-  template <typename _T, typename ScanOpT>
-  _CCCL_DEVICE _CCCL_FORCEINLINE void InclusiveScan(_T input, _T& inclusive_output, ScanOpT scan_op)
+  template <typename _Tp, typename ScanOpT>
+  _CCCL_DEVICE _CCCL_FORCEINLINE void InclusiveScan(_Tp input, _Tp& inclusive_output, ScanOpT scan_op)
   {
     inclusive_output = input;
 
@@ -586,9 +562,9 @@ struct WarpScanShfl
    * @param[in] valid_items
    *   Number of valid items in warp
    */
-  template <typename _T, typename ScanOpT>
+  template <typename _Tp, typename ScanOpT>
   _CCCL_DEVICE _CCCL_FORCEINLINE void
-  InclusiveScanPartial(_T input, _T& inclusive_output, ScanOpT scan_op, int valid_items)
+  InclusiveScanPartial(_Tp input, _Tp& inclusive_output, ScanOpT scan_op, int valid_items)
   {
     if (static_cast<int>(lane_id) < valid_items)
     {

cuda/cccl/headers/include/cub/warp/specializations/warp_scan_smem.cuh

@@ -1,30 +1,6 @@
-/******************************************************************************
- * Copyright (c) 2011, Duane Merrill.  All rights reserved.
- * Copyright (c) 2011-2018, NVIDIA CORPORATION.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in the
- *       documentation and/or other materials provided with the distribution.
- *     * Neither the name of the NVIDIA CORPORATION nor the
- *       names of its contributors may be used to endorse or promote products
- *       derived from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
- * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
+// SPDX-FileCopyrightText: Copyright (c) 2011, Duane Merrill. All rights reserved.
+// SPDX-FileCopyrightText: Copyright (c) 2011-2018, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file