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/agent/agent_rle.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
@@ -68,61 +44,58 @@ CUB_NAMESPACE_BEGIN
 /**
  * Parameterizable tuning policy type for AgentRle
  *
- * @tparam _BLOCK_THREADS
+ * @tparam BlockThreads
  *   Threads per thread block
  *
- * @tparam _ITEMS_PER_THREAD
+ * @tparam ItemsPerThread
  *   Items per thread (per tile of input)
  *
- * @tparam _LOAD_ALGORITHM
+ * @tparam LoadAlgorithm
  *   The BlockLoad algorithm to use
  *
- * @tparam _LOAD_MODIFIER
+ * @tparam LoadModifier
  *   Cache load modifier for reading input elements
  *
- * @tparam _STORE_WARP_TIME_SLICING
+ * @tparam StoreWarpTimeSlicing
  *   Whether or not only one warp's worth of shared memory should be allocated and time-sliced among
  *   block-warps during any store-related data transpositions
  *   (versus each warp having its own storage)
  *
- * @tparam _SCAN_ALGORITHM
+ * @tparam ScanAlgorithm
  *   The BlockScan algorithm to use
  *
  * @tparam DelayConstructorT
  *   Implementation detail, do not specify directly, requirements on the
  *   content of this type are subject to breaking change.
  */
-template <int _BLOCK_THREADS,
-          int _ITEMS_PER_THREAD,
-          BlockLoadAlgorithm _LOAD_ALGORITHM,
-          CacheLoadModifier _LOAD_MODIFIER,
-          bool _STORE_WARP_TIME_SLICING,
-          BlockScanAlgorithm _SCAN_ALGORITHM,
+template <int BlockThreads,
+          int ItemsPerThread,
+          BlockLoadAlgorithm LoadAlgorithm,
+          CacheLoadModifier LoadModifier,
+          bool StoreWarpTimeSlicing,
+          BlockScanAlgorithm ScanAlgorithm,
           typename DelayConstructorT = detail::fixed_delay_constructor_t<350, 450>>
 struct AgentRlePolicy
 {
-  enum
-  {
-    /// Threads per thread block
-    BLOCK_THREADS = _BLOCK_THREADS,
+  /// Threads per thread block
+  static constexpr int BLOCK_THREADS = BlockThreads;
 
-    /// Items per thread (per tile of input)
-    ITEMS_PER_THREAD = _ITEMS_PER_THREAD,
+  /// Items per thread (per tile of input)
+  static constexpr int ITEMS_PER_THREAD = ItemsPerThread;
 
-    /// Whether or not only one warp's worth of shared memory should be allocated and time-sliced
-    /// among block-warps during any store-related data transpositions (versus each warp having its
-    /// own storage)
-    STORE_WARP_TIME_SLICING = _STORE_WARP_TIME_SLICING,
-  };
+  /// Whether or not only one warp's worth of shared memory should be allocated and time-sliced
+  /// among block-warps during any store-related data transpositions (versus each warp having its
+  /// own storage)
+  static constexpr bool STORE_WARP_TIME_SLICING = StoreWarpTimeSlicing;
 
   /// The BlockLoad algorithm to use
-  static constexpr BlockLoadAlgorithm LOAD_ALGORITHM = _LOAD_ALGORITHM;
+  static constexpr BlockLoadAlgorithm LOAD_ALGORITHM = LoadAlgorithm;
 
   /// Cache load modifier for reading input elements
-  static constexpr CacheLoadModifier LOAD_MODIFIER = _LOAD_MODIFIER;
+  static constexpr CacheLoadModifier LOAD_MODIFIER = LoadModifier;
 
   /// The BlockScan algorithm to use
-  static constexpr BlockScanAlgorithm SCAN_ALGORITHM = _SCAN_ALGORITHM;
+  static constexpr BlockScanAlgorithm SCAN_ALGORITHM = ScanAlgorithm;
 
   struct detail
   {
@@ -136,7 +109,6 @@ struct AgentRlePolicy
 
 namespace detail::rle
 {
-
 /**
  * @brief AgentRle implements a stateful abstraction of CUDA thread blocks for participating in device-wide
  * run-length-encode
@@ -1118,7 +1090,6 @@ struct AgentRle
     }
   }
 };
-
 } // namespace detail::rle
 
 CUB_NAMESPACE_END

cuda/cccl/headers/include/cub/agent/agent_scan.cuh

@@ -1,30 +1,6 @@
-/******************************************************************************
- * Copyright (c) 2011, Duane Merrill.  All rights reserved.
- * 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, Duane Merrill. All rights reserved.
+// SPDX-FileCopyrightText: Copyright (c) 2011-2022, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file cub::AgentScan implements a stateful abstraction of CUDA thread blocks
@@ -68,47 +44,46 @@ CUB_NAMESPACE_BEGIN
 /**
  * @brief Parameterizable tuning policy type for AgentScan
  *
- * @tparam NOMINAL_BLOCK_THREADS_4B
+ * @tparam NominalBlockThreads4B
  *   Threads per thread block
  *
- * @tparam NOMINAL_ITEMS_PER_THREAD_4B
+ * @tparam NominalItemsPerThread4B
  *   Items per thread (per tile of input)
  *
  * @tparam ComputeT
  *   Dominant compute type
  *
- * @tparam _LOAD_ALGORITHM
+ * @tparam LoadAlgorithm
  *   The BlockLoad algorithm to use
  *
- * @tparam _LOAD_MODIFIER
+ * @tparam LoadModifier
  *   Cache load modifier for reading input elements
  *
- * @tparam _STORE_ALGORITHM
+ * @tparam StoreAlgorithm
  *   The BlockStore algorithm to use
  *
- * @tparam _SCAN_ALGORITHM
+ * @tparam ScanAlgorithm
  *   The BlockScan algorithm to use
  *
  * @tparam DelayConstructorT
  *   Implementation detail, do not specify directly, requirements on the
  *   content of this type are subject to breaking change.
  */
-template <
-  int NOMINAL_BLOCK_THREADS_4B,
-  int NOMINAL_ITEMS_PER_THREAD_4B,
-  typename ComputeT,
-  BlockLoadAlgorithm _LOAD_ALGORITHM,
-  CacheLoadModifier _LOAD_MODIFIER,
-  BlockStoreAlgorithm _STORE_ALGORITHM,
-  BlockScanAlgorithm _SCAN_ALGORITHM,
-  typename ScalingType       = detail::MemBoundScaling<NOMINAL_BLOCK_THREADS_4B, NOMINAL_ITEMS_PER_THREAD_4B, ComputeT>,
-  typename DelayConstructorT = detail::default_delay_constructor_t<ComputeT>>
+template <int NominalBlockThreads4B,
+          int NominalItemsPerThread4B,
+          typename ComputeT,
+          BlockLoadAlgorithm LoadAlgorithm,
+          CacheLoadModifier LoadModifier,
+          BlockStoreAlgorithm StoreAlgorithm,
+          BlockScanAlgorithm ScanAlgorithm,
+          typename ScalingType = detail::MemBoundScaling<NominalBlockThreads4B, NominalItemsPerThread4B, ComputeT>,
+          typename DelayConstructorT = detail::default_delay_constructor_t<ComputeT>>
 struct AgentScanPolicy : ScalingType
 {
-  static constexpr BlockLoadAlgorithm LOAD_ALGORITHM   = _LOAD_ALGORITHM;
-  static constexpr CacheLoadModifier LOAD_MODIFIER     = _LOAD_MODIFIER;
-  static constexpr BlockStoreAlgorithm STORE_ALGORITHM = _STORE_ALGORITHM;
-  static constexpr BlockScanAlgorithm SCAN_ALGORITHM   = _SCAN_ALGORITHM;
+  static constexpr BlockLoadAlgorithm LOAD_ALGORITHM   = LoadAlgorithm;
+  static constexpr CacheLoadModifier LOAD_MODIFIER     = LoadModifier;
+  static constexpr BlockStoreAlgorithm STORE_ALGORITHM = StoreAlgorithm;
+  static constexpr BlockScanAlgorithm SCAN_ALGORITHM   = ScanAlgorithm;
 
   struct detail
   {
@@ -143,7 +118,6 @@ CUB_DETAIL_POLICY_WRAPPER_DEFINE(
 
 namespace detail::scan
 {
-
 /**
  * @brief AgentScan implements a stateful abstraction of CUDA thread blocks for
  *        participating in device-wide prefix scan.
@@ -583,7 +557,6 @@ struct AgentScan
     }
   }
 };
-
 } // namespace detail::scan
 
 CUB_NAMESPACE_END

cuda/cccl/headers/include/cub/agent/agent_scan_by_key.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 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) 2022, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file AgentScanByKey implements a stateful abstraction of CUDA thread blocks
@@ -69,22 +45,22 @@ CUB_NAMESPACE_BEGIN
  *   Implementation detail, do not specify directly, requirements on the
  *   content of this type are subject to breaking change.
  */
-template <int _BLOCK_THREADS,
-          int _ITEMS_PER_THREAD                = 1,
-          BlockLoadAlgorithm _LOAD_ALGORITHM   = BLOCK_LOAD_DIRECT,
-          CacheLoadModifier _LOAD_MODIFIER     = LOAD_DEFAULT,
-          BlockScanAlgorithm _SCAN_ALGORITHM   = BLOCK_SCAN_WARP_SCANS,
-          BlockStoreAlgorithm _STORE_ALGORITHM = BLOCK_STORE_DIRECT,
-          typename DelayConstructorT           = detail::fixed_delay_constructor_t<350, 450>>
+template <int BlockThreads,
+          int ItemsPerThread                 = 1,
+          BlockLoadAlgorithm LoadAlgorithm   = BLOCK_LOAD_DIRECT,
+          CacheLoadModifier LoadModifier     = LOAD_DEFAULT,
+          BlockScanAlgorithm ScanAlgorithm   = BLOCK_SCAN_WARP_SCANS,
+          BlockStoreAlgorithm StoreAlgorithm = BLOCK_STORE_DIRECT,
+          typename DelayConstructorT         = detail::fixed_delay_constructor_t<350, 450>>
 struct AgentScanByKeyPolicy
 {
-  static constexpr int BLOCK_THREADS    = _BLOCK_THREADS;
-  static constexpr int ITEMS_PER_THREAD = _ITEMS_PER_THREAD;
+  static constexpr int BLOCK_THREADS    = BlockThreads;
+  static constexpr int ITEMS_PER_THREAD = ItemsPerThread;
 
-  static constexpr BlockLoadAlgorithm LOAD_ALGORITHM   = _LOAD_ALGORITHM;
-  static constexpr CacheLoadModifier LOAD_MODIFIER     = _LOAD_MODIFIER;
-  static constexpr BlockScanAlgorithm SCAN_ALGORITHM   = _SCAN_ALGORITHM;
-  static constexpr BlockStoreAlgorithm STORE_ALGORITHM = _STORE_ALGORITHM;
+  static constexpr BlockLoadAlgorithm LOAD_ALGORITHM   = LoadAlgorithm;
+  static constexpr CacheLoadModifier LOAD_MODIFIER     = LoadModifier;
+  static constexpr BlockScanAlgorithm SCAN_ALGORITHM   = ScanAlgorithm;
+  static constexpr BlockStoreAlgorithm STORE_ALGORITHM = StoreAlgorithm;
 
   struct detail
   {
@@ -98,7 +74,6 @@ struct AgentScanByKeyPolicy
 
 namespace detail::scan_by_key
 {
-
 /**
  * @brief AgentScanByKey implements a stateful abstraction of CUDA thread
  *        blocks for participating in device-wide prefix scan by key.
@@ -468,7 +443,6 @@ struct AgentScanByKey
     }
   }
 };
-
 } // namespace detail::scan_by_key
 
 CUB_NAMESPACE_END

cuda/cccl/headers/include/cub/agent/agent_segmented_radix_sort.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 2011-2021, 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) 2011-2021, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 #pragma once
 
@@ -47,12 +23,11 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail::radix_sort
 {
-
 /**
  * This agent will be implementing the `DeviceSegmentedRadixSort` when the
  * https://github.com/NVIDIA/cub/issues/383 is addressed.
  *
- * @tparam IS_DESCENDING
+ * @tparam IsDescending
  *   Whether or not the sorted-order is high-to-low
  *
  * @tparam SegmentedPolicyT
@@ -67,7 +42,7 @@ namespace detail::radix_sort
  * @tparam OffsetT
  *   Signed integer type for global offsets
  */
-template <bool IS_DESCENDING,
+template <bool IsDescending,
           typename SegmentedPolicyT,
           typename KeyT,
           typename ValueT,
@@ -89,7 +64,7 @@ struct AgentSegmentedRadixSort
   // Huge segment handlers
   using BlockUpsweepT   = AgentRadixSortUpsweep<SegmentedPolicyT, KeyT, OffsetT, DecomposerT>;
   using DigitScanT      = BlockScan<OffsetT, BLOCK_THREADS>;
-  using BlockDownsweepT = AgentRadixSortDownsweep<SegmentedPolicyT, IS_DESCENDING, KeyT, ValueT, OffsetT, DecomposerT>;
+  using BlockDownsweepT = AgentRadixSortDownsweep<SegmentedPolicyT, IsDescending, KeyT, ValueT, OffsetT, DecomposerT>;
 
   /// Number of bin-starting offsets tracked per thread
   static constexpr int BINS_TRACKED_PER_THREAD = BlockDownsweepT::BINS_TRACKED_PER_THREAD;
@@ -150,7 +125,7 @@ struct AgentSegmentedRadixSort
     // LOWEST   -> -nan          = 11...11b -> TwiddleIn ->  0 = 00...00b
 
     bit_ordered_type default_key_bits =
-      IS_DESCENDING ? traits::min_raw_binary_key(decomposer) : traits::max_raw_binary_key(decomposer);
+      IsDescending ? traits::min_raw_binary_key(decomposer) : traits::max_raw_binary_key(decomposer);
     KeyT oob_default = reinterpret_cast<KeyT&>(default_key_bits);
 
     if (!KEYS_ONLY)
@@ -172,7 +147,7 @@ struct AgentSegmentedRadixSort
         thread_values,
         begin_bit,
         end_bit,
-        bool_constant_v<IS_DESCENDING>,
+        bool_constant_v<IsDescending>,
         bool_constant_v<KEYS_ONLY>,
         decomposer);
 
@@ -204,7 +179,7 @@ struct AgentSegmentedRadixSort
 
     __syncthreads();
 
-    if (IS_DESCENDING)
+    if (IsDescending)
     {
       // Reverse bin counts
       _CCCL_PRAGMA_UNROLL_FULL()
@@ -238,7 +213,7 @@ struct AgentSegmentedRadixSort
     OffsetT bin_offset[BINS_TRACKED_PER_THREAD];
     DigitScanT(temp_storage.unbound_sort.scan).ExclusiveSum(bin_count, bin_offset);
 
-    if (IS_DESCENDING)
+    if (IsDescending)
     {
       // Reverse bin offsets
       _CCCL_PRAGMA_UNROLL_FULL()
@@ -283,7 +258,6 @@ struct AgentSegmentedRadixSort
     downsweep.ProcessRegion(OffsetT{}, num_items);
   }
 };
-
 } // namespace detail::radix_sort
 
 CUB_NAMESPACE_END

cuda/cccl/headers/include/cub/agent/agent_select_if.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
@@ -71,50 +47,47 @@ CUB_NAMESPACE_BEGIN
 /**
  * Parameterizable tuning policy type for AgentSelectIf
  *
- * @tparam _BLOCK_THREADS
+ * @tparam BlockThreads
  *   Threads per thread block
  *
- * @tparam _ITEMS_PER_THREAD
+ * @tparam ItemsPerThread
  *   Items per thread (per tile of input)
  *
- * @tparam _LOAD_ALGORITHM
+ * @tparam LoadAlgorithm
  *   The BlockLoad algorithm to use
  *
- * @tparam _LOAD_MODIFIER
+ * @tparam LoadModifier
  *   Cache load modifier for reading input elements
  *
- * @tparam _SCAN_ALGORITHM
+ * @tparam ScanAlgorithm
  *   The BlockScan algorithm to use
  *
  * @tparam DelayConstructorT
  *   Implementation detail, do not specify directly, requirements on the
  *   content of this type are subject to breaking change.
  */
-template <int _BLOCK_THREADS,
-          int _ITEMS_PER_THREAD,
-          BlockLoadAlgorithm _LOAD_ALGORITHM,
-          CacheLoadModifier _LOAD_MODIFIER,
-          BlockScanAlgorithm _SCAN_ALGORITHM,
+template <int BlockThreads,
+          int ItemsPerThread,
+          BlockLoadAlgorithm LoadAlgorithm,
+          CacheLoadModifier LoadModifier,
+          BlockScanAlgorithm ScanAlgorithm,
           typename DelayConstructorT = detail::fixed_delay_constructor_t<350, 450>>
 struct AgentSelectIfPolicy
 {
-  enum
-  {
-    /// Threads per thread block
-    BLOCK_THREADS = _BLOCK_THREADS,
+  /// Threads per thread block
+  static constexpr int BLOCK_THREADS = BlockThreads;
 
-    /// Items per thread (per tile of input)
-    ITEMS_PER_THREAD = _ITEMS_PER_THREAD,
-  };
+  /// Items per thread (per tile of input)
+  static constexpr int ITEMS_PER_THREAD = ItemsPerThread;
 
   /// The BlockLoad algorithm to use
-  static constexpr BlockLoadAlgorithm LOAD_ALGORITHM = _LOAD_ALGORITHM;
+  static constexpr BlockLoadAlgorithm LOAD_ALGORITHM = LoadAlgorithm;
 
   /// Cache load modifier for reading input elements
-  static constexpr CacheLoadModifier LOAD_MODIFIER = _LOAD_MODIFIER;
+  static constexpr CacheLoadModifier LOAD_MODIFIER = LoadModifier;
 
   /// The BlockScan algorithm to use
-  static constexpr BlockScanAlgorithm SCAN_ALGORITHM = _SCAN_ALGORITHM;
+  static constexpr BlockScanAlgorithm SCAN_ALGORITHM = ScanAlgorithm;
 
   struct detail
   {
@@ -128,7 +101,6 @@ struct AgentSelectIfPolicy
 
 namespace detail::select
 {
-
 template <typename EqualityOpT>
 struct guarded_inequality_op
 {
@@ -1111,7 +1083,6 @@ struct AgentSelectIf
     }
   }
 };
-
 } // namespace detail::select
 
 CUB_NAMESPACE_END

cuda/cccl/headers/include/cub/agent/agent_sub_warp_merge_sort.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 2011-2021, 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) 2011-2021, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 #pragma once
 
@@ -48,23 +24,23 @@
 
 CUB_NAMESPACE_BEGIN
 
-template <int BLOCK_THREADS_ARG,
-          int WARP_THREADS_ARG,
-          int ITEMS_PER_THREAD_ARG,
-          cub::WarpLoadAlgorithm LOAD_ALGORITHM_ARG   = cub::WARP_LOAD_DIRECT,
-          cub::CacheLoadModifier LOAD_MODIFIER_ARG    = cub::LOAD_LDG,
-          cub::WarpStoreAlgorithm STORE_ALGORITHM_ARG = cub::WARP_STORE_DIRECT>
+template <int BlockThreadsArg,
+          int WarpThreadsArg,
+          int ItemsPerThreadArg,
+          cub::WarpLoadAlgorithm LoadAlgorithmArg   = cub::WARP_LOAD_DIRECT,
+          cub::CacheLoadModifier LoadModifierArg    = cub::LOAD_LDG,
+          cub::WarpStoreAlgorithm StoreAlgorithmArg = cub::WARP_STORE_DIRECT>
 struct AgentSubWarpMergeSortPolicy
 {
-  static constexpr int BLOCK_THREADS      = BLOCK_THREADS_ARG;
-  static constexpr int WARP_THREADS       = WARP_THREADS_ARG;
-  static constexpr int ITEMS_PER_THREAD   = ITEMS_PER_THREAD_ARG;
+  static constexpr int BLOCK_THREADS      = BlockThreadsArg;
+  static constexpr int WARP_THREADS       = WarpThreadsArg;
+  static constexpr int ITEMS_PER_THREAD   = ItemsPerThreadArg;
   static constexpr int ITEMS_PER_TILE     = WARP_THREADS * ITEMS_PER_THREAD;
   static constexpr int SEGMENTS_PER_BLOCK = BLOCK_THREADS / WARP_THREADS;
 
-  static constexpr cub::WarpLoadAlgorithm LOAD_ALGORITHM   = LOAD_ALGORITHM_ARG;
-  static constexpr cub::CacheLoadModifier LOAD_MODIFIER    = LOAD_MODIFIER_ARG;
-  static constexpr cub::WarpStoreAlgorithm STORE_ALGORITHM = STORE_ALGORITHM_ARG;
+  static constexpr cub::WarpLoadAlgorithm LOAD_ALGORITHM   = LoadAlgorithmArg;
+  static constexpr cub::CacheLoadModifier LOAD_MODIFIER    = LoadModifierArg;
+  static constexpr cub::WarpStoreAlgorithm STORE_ALGORITHM = StoreAlgorithmArg;
 };
 
 #if defined(CUB_DEFINE_RUNTIME_POLICIES) || defined(CUB_ENABLE_POLICY_PTX_JSON)
@@ -86,7 +62,6 @@ CUB_DETAIL_POLICY_WRAPPER_DEFINE(
 
 namespace detail::sub_warp_merge_sort
 {
-
 /**
  * @brief AgentSubWarpSort implements a sub-warp merge sort.
  *
@@ -340,7 +315,6 @@ private:
     }
   }
 };
-
 } // namespace detail::sub_warp_merge_sort
 
 CUB_NAMESPACE_END