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/device/dispatch/dispatch_histogram.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
@@ -47,7 +23,7 @@
 #endif // no system header
 
 #include <cub/agent/agent_histogram.cuh>
-#include <cub/device/dispatch/kernels/histogram.cuh>
+#include <cub/device/dispatch/kernels/kernel_histogram.cuh>
 #include <cub/device/dispatch/tuning/tuning_histogram.cuh>
 #include <cub/grid/grid_queue.cuh>
 #include <cub/thread/thread_search.cuh>
@@ -78,10 +54,15 @@ namespace detail::histogram
 template <int NUM_CHANNELS, int NUM_ACTIVE_CHANNELS, typename SampleIteratorT, typename CounterT, typename OffsetT>
 struct DeviceHistogramKernelSource
 {
-  CUB_DEFINE_KERNEL_GETTER(HistogramInitKernel, DeviceHistogramInitKernel<NUM_ACTIVE_CHANNELS, CounterT, OffsetT>);
-
   // We define this differently than the other kernel getters because there are
   // different dispatch paths which affect which policy is used and the decode operators.
+
+  template <typename PolicyT>
+  _CCCL_HIDE_FROM_ABI CUB_RUNTIME_FUNCTION static constexpr auto HistogramInitKernel()
+  {
+    return &DeviceHistogramInitKernel<PolicyT, NUM_ACTIVE_CHANNELS, CounterT, OffsetT>;
+  }
+
   template <typename PolicyT, int PRIVATIZED_SMEM_BINS, typename PrivatizedDecodeOpT, typename OutputDecodeOpT>
   _CCCL_HIDE_FROM_ABI CUB_RUNTIME_FUNCTION static constexpr auto HistogramSweepKernel()
   {
@@ -251,7 +232,7 @@ struct dispatch_histogram
 #endif // CUB_DEBUG_LOG
 
       // Invoke histogram_init_kernel
-      launcher_factory(histogram_init_grid_dims, histogram_init_block_threads, 0, stream)
+      launcher_factory(histogram_init_grid_dims, histogram_init_block_threads, 0, stream, true)
         .doit(histogram_init_kernel, num_output_bins_wrapper, d_output_histograms, tile_queue);
 
       // Return if empty problem
@@ -274,7 +255,7 @@ struct dispatch_histogram
 #endif // CUB_DEBUG_LOG
 
       // Invoke histogram_sweep_kernel
-      launcher_factory(sweep_grid_dims, block_threads, 0, stream)
+      launcher_factory(sweep_grid_dims, block_threads, 0, stream, true)
         .doit(histogram_sweep_kernel,
               d_samples,
               num_output_bins_wrapper,
@@ -311,13 +292,12 @@ struct dispatch_histogram
   CUB_RUNTIME_FUNCTION _CCCL_FORCEINLINE cudaError_t Invoke(ActivePolicyT active_policy = {})
   {
     return Invoke<ActivePolicyT>(
-      kernel_source.HistogramInitKernel(),
+      kernel_source.template HistogramInitKernel<MaxPolicyT>(),
       kernel_source
         .template HistogramSweepKernel<MaxPolicyT, PRIVATIZED_SMEM_BINS, PrivatizedDecodeOpT, OutputDecodeOpT>(),
       active_policy);
   }
 };
-
 } // namespace detail::histogram
 
 /******************************************************************************

cuda/cccl/headers/include/cub/device/dispatch/dispatch_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
 
@@ -38,16 +14,15 @@
 #endif // no system header
 
 #include <cub/agent/agent_merge_sort.cuh>
-#include <cub/device/dispatch/kernels/merge_sort.cuh>
+#include <cub/device/dispatch/kernels/kernel_merge_sort.cuh>
 #include <cub/device/dispatch/tuning/tuning_merge_sort.cuh>
 #include <cub/util_device.cuh>
 #include <cub/util_math.cuh>
 #include <cub/util_namespace.cuh>
 #include <cub/util_vsmem.cuh>
 
-#include <thrust/detail/integer_math.h>
-
 #include <cuda/__cmath/ceil_div.h>
+#include <cuda/__cmath/ilog.h>
 #include <cuda/std/__algorithm/max.h>
 #include <cuda/std/__type_traits/is_same.h>
 #include <cuda/std/cstdint>
@@ -282,7 +257,7 @@ struct DispatchMergeSort
         break;
       }
 
-      const int num_passes = static_cast<int>(THRUST_NS_QUALIFIER::detail::log2_ri(num_tiles));
+      const int num_passes = ::cuda::ceil_ilog2(num_tiles);
 
       /*
        * The algorithm consists of stages. At each stage, there are input and output arrays. There are two pairs of

cuda/cccl/headers/include/cub/device/dispatch/dispatch_radix_sort.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
@@ -44,7 +20,7 @@
 #  pragma system_header
 #endif // no system header
 
-#include <cub/device/dispatch/kernels/radix_sort.cuh>
+#include <cub/device/dispatch/kernels/kernel_radix_sort.cuh>
 #include <cub/device/dispatch/tuning/tuning_radix_sort.cuh>
 #include <cub/util_debug.cuh>
 #include <cub/util_device.cuh>
@@ -156,7 +132,6 @@ struct DeviceSegmentedRadixSortKernelSource
     return sizeof(ValueT);
   }
 };
-
 } // namespace detail::radix_sort
 
 /******************************************************************************

cuda/cccl/headers/include/cub/device/dispatch/dispatch_reduce.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 cub::DeviceReduce provides device-wide, parallel operations for
@@ -46,8 +22,8 @@
 
 #include <cub/detail/launcher/cuda_runtime.cuh>
 #include <cub/detail/type_traits.cuh> // for cub::detail::invoke_result_t
-#include <cub/device/dispatch/kernels/reduce.cuh>
-#include <cub/device/dispatch/kernels/segmented_reduce.cuh>
+#include <cub/device/dispatch/kernels/kernel_reduce.cuh>
+#include <cub/device/dispatch/kernels/kernel_segmented_reduce.cuh>
 #include <cub/device/dispatch/tuning/tuning_reduce.cuh>
 #include <cub/grid/grid_even_share.cuh>
 #include <cub/thread/thread_operators.cuh>
@@ -70,7 +46,6 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail::reduce
 {
-
 template <typename MaxPolicyT,
           typename InputIteratorT,
           typename OutputIteratorT,
@@ -609,7 +584,6 @@ using DispatchTransformReduce =
 
 namespace detail::reduce
 {
-
 template <typename MaxPolicyT,
           typename InputIteratorT,
           typename OutputIteratorT,
@@ -990,7 +964,6 @@ struct DispatchSegmentedReduce
 
 namespace detail::reduce
 {
-
 // @brief Functor to generate a key-value pair from an index and value
 template <typename Iterator, typename OutputValueT>
 struct generate_idx_value

cuda/cccl/headers/include/cub/device/dispatch/dispatch_reduce_by_key.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::DeviceReduceByKey provides device-wide, parallel operations for
@@ -61,7 +37,6 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail::reduce
 {
-
 template <typename PrecedingKeyItT, typename AccumT, typename GlobalOffsetT>
 struct streaming_context
 {
@@ -259,7 +234,6 @@ __launch_bounds__(int(ChainedPolicyT::ActivePolicy::ReduceByKeyPolicyT::BLOCK_TH
   // If applicable, hints to discard modified cache lines for vsmem
   vsmem_helper_t::discard_temp_storage(temp_storage);
 }
-
 } // namespace detail::reduce
 
 /******************************************************************************

cuda/cccl/headers/include/cub/device/dispatch/dispatch_reduce_deterministic.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 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) 2025, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file This file device-wide, parallel operations for
@@ -72,10 +48,8 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail
 {
-
 namespace rfa
 {
-
 template <typename Invocable, typename InputT>
 using transformed_input_t = ::cuda::std::decay_t<::cuda::std::invoke_result_t<Invocable, InputT>>;
 
@@ -112,7 +86,6 @@ struct deterministic_sum_t
     return lhs + rhs;
   }
 };
-
 } // namespace rfa
 
 /******************************************************************************

cuda/cccl/headers/include/cub/device/dispatch/dispatch_reduce_nondeterministic.cuh

@@ -20,7 +20,7 @@
 
 #include <cub/detail/launcher/cuda_runtime.cuh>
 #include <cub/detail/type_traits.cuh> // for cub::detail::invoke_result_t
-#include <cub/device/dispatch/kernels/reduce.cuh>
+#include <cub/device/dispatch/kernels/kernel_reduce.cuh>
 #include <cub/device/dispatch/tuning/tuning_reduce.cuh>
 #include <cub/grid/grid_even_share.cuh>
 #include <cub/thread/thread_operators.cuh>
@@ -39,7 +39,6 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail::reduce
 {
-
 template <typename MaxPolicyT,
           typename InputIteratorT,
           typename OutputIteratorT,

cuda/cccl/headers/include/cub/device/dispatch/dispatch_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
@@ -69,7 +45,6 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail::rle
 {
-
 template <typename PrecedingKeyItT, typename RunLengthT, typename GlobalOffsetT>
 struct streaming_context
 {
@@ -414,26 +389,6 @@ struct DeviceRleDispatch
           ? (num_items - current_partition_offset)
           : capped_num_items_per_invocation;
 
-      streaming_context_t streaming_context{};
-      if constexpr (use_streaming_invocation)
-      {
-        auto tmp_num_uniques = static_cast<global_offset_t*>(allocations[1]);
-        auto tmp_prefix      = static_cast<length_t*>(allocations[2]);
-
-        const bool is_first_partition = (partition_idx == 0);
-        const bool is_last_partition  = (partition_idx + 1 == num_partitions);
-        const int buffer_selector     = partition_idx % 2;
-
-        streaming_context = streaming_context_t{
-          is_first_partition,
-          is_last_partition,
-          current_partition_offset,
-          &tmp_prefix[buffer_selector],
-          &tmp_prefix[buffer_selector ^ 0x01],
-          &tmp_num_uniques[buffer_selector],
-          &tmp_num_uniques[buffer_selector ^ 0x01]};
-      }
-
       // Construct the tile status interface
       const auto num_current_tiles = static_cast<int>(::cuda::ceil_div(current_num_items, tile_size));
 
@@ -490,17 +445,50 @@ struct DeviceRleDispatch
 #endif // CUB_DEBUG_LOG
 
       // Invoke device_rle_sweep_kernel
-      THRUST_NS_QUALIFIER::cuda_cub::detail::triple_chevron(num_current_tiles, block_threads, 0, stream)
-        .doit(device_rle_sweep_kernel,
-              d_in + current_partition_offset,
-              d_offsets_out,
-              d_lengths_out,
-              d_num_runs_out,
-              tile_status,
-              equality_op,
-              static_cast<local_offset_t>(current_num_items),
-              num_current_tiles,
-              streaming_context);
+      if constexpr (use_streaming_invocation)
+      {
+        auto tmp_num_uniques = static_cast<global_offset_t*>(allocations[1]);
+        auto tmp_prefix      = static_cast<length_t*>(allocations[2]);
+
+        const bool is_first_partition = (partition_idx == 0);
+        const bool is_last_partition  = (partition_idx + 1 == num_partitions);
+        const int buffer_selector     = partition_idx % 2;
+
+        streaming_context_t streaming_context{
+          is_first_partition,
+          is_last_partition,
+          current_partition_offset,
+          &tmp_prefix[buffer_selector],
+          &tmp_prefix[buffer_selector ^ 0x01],
+          &tmp_num_uniques[buffer_selector],
+          &tmp_num_uniques[buffer_selector ^ 0x01]};
+
+        THRUST_NS_QUALIFIER::cuda_cub::detail::triple_chevron(num_current_tiles, block_threads, 0, stream)
+          .doit(device_rle_sweep_kernel,
+                d_in + current_partition_offset,
+                d_offsets_out,
+                d_lengths_out,
+                d_num_runs_out,
+                tile_status,
+                equality_op,
+                static_cast<local_offset_t>(current_num_items),
+                num_current_tiles,
+                streaming_context);
+      }
+      else
+      {
+        THRUST_NS_QUALIFIER::cuda_cub::detail::triple_chevron(num_current_tiles, block_threads, 0, stream)
+          .doit(device_rle_sweep_kernel,
+                d_in + current_partition_offset,
+                d_offsets_out,
+                d_lengths_out,
+                d_num_runs_out,
+                tile_status,
+                equality_op,
+                static_cast<local_offset_t>(current_num_items),
+                num_current_tiles,
+                NullType{});
+      }
 
       // Check for failure to launch
       error = CubDebug(cudaPeekAtLastError());

cuda/cccl/headers/include/cub/device/dispatch/dispatch_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::DeviceScan provides device-wide, parallel operations for
@@ -48,7 +24,7 @@
 
 #include <cub/agent/agent_scan.cuh>
 #include <cub/detail/launcher/cuda_runtime.cuh>
-#include <cub/device/dispatch/kernels/scan.cuh>
+#include <cub/device/dispatch/kernels/kernel_scan.cuh>
 #include <cub/device/dispatch/tuning/tuning_scan.cuh>
 #include <cub/thread/thread_operators.cuh>
 #include <cub/util_debug.cuh>
@@ -74,7 +50,6 @@ enum class ForceInclusive
 
 namespace detail::scan
 {
-
 template <typename MaxPolicyT,
           typename InputIteratorT,
           typename OutputIteratorT,
@@ -111,7 +86,6 @@ struct DeviceScanKernelSource
     return ScanTileStateT();
   }
 };
-
 } // namespace detail::scan
 
 /******************************************************************************
@@ -237,6 +211,15 @@ struct DispatchScan
    * @param[in] stream
    *   **[optional]** CUDA stream to launch kernels within.
    *   Default is stream<sub>0</sub>.
+   *
+   * @param[in] kernel_source
+   *   Object specifying implementation kernels
+   *
+   * @param[in] launcher_factory
+   *   Object to execute implementation kernels on the given stream
+   *
+   * @param[in] max_policy
+   *   Struct encoding chain of algorithm tuning policies
    */
   CUB_RUNTIME_FUNCTION _CCCL_FORCEINLINE DispatchScan(
     void* d_temp_storage,
@@ -446,6 +429,14 @@ struct DispatchScan
    *   **[optional]** CUDA stream to launch kernels within.
    *   Default is stream<sub>0</sub>.
    *
+   * @param[in] kernel_source
+   *   Object specifying implementation kernels
+   *
+   * @param[in] launcher_factory
+   *   Object to execute implementation kernels on the given stream
+   *
+   * @param[in] max_policy
+   *   Struct encoding chain of algorithm tuning policies
    */
   template <typename MaxPolicyT = typename PolicyHub::MaxPolicy>
   CUB_RUNTIME_FUNCTION _CCCL_FORCEINLINE static cudaError_t Dispatch(

cuda/cccl/headers/include/cub/device/dispatch/dispatch_scan_by_key.cuh

@@ -1,29 +1,5 @@
-/******************************************************************************
- * Copyright (c) 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) 2021, NVIDIA CORPORATION. All rights reserved.
+// SPDX-License-Identifier: BSD-3
 
 /**
  * @file DeviceScan provides device-wide, parallel operations for computing a
@@ -68,7 +44,6 @@ CUB_NAMESPACE_BEGIN
 
 namespace detail::scan_by_key
 {
-
 /**
  * @brief Scan by key kernel entry point (multi-block)
  *