Commit 198c0c31 authored by Tomáš Oberhuber's avatar Tomáš Oberhuber
Browse files

Added reference algorithms - Nvidia bitonic sort and thrust radix sort.

parent 9a2f8f66

src/Benchmarks/Sorting/Measurer.h

+2 −0
Original line number Diff line number Diff line
@@ -12,6 +12,8 @@ 
#include "ReferenceAlgorithms/MancaQuicksort.h"

#endif

#include "ReferenceAlgorithms/CedermanQuicksort.h"

#include "ReferenceAlgorithms/ThrustRadixsort.h"

#include "ReferenceAlgorithms/NvidiaBitonicSort.h"

#endif



#include "timer.h"

src/Benchmarks/Sorting/ReferenceAlgorithms/NvidiaBitonicSort.h

0 → 100644
+16 −0
Original line number Diff line number Diff line 
#include <6_Advanced/sortingNetworks/bitonicSort.cu>

#include <TNL/Containers/Array.h>





struct NvidiaBitonicSort

{

   static void sort( Containers::ArrayView< int, Devices::Cuda >& view )

   {

      Array<int, Devices::Cuda> arr;

      arr = view;

      bitonicSort((unsigned *)view.getData(), (unsigned *)arr.getData(),

                  (unsigned *)view.getData(), (unsigned *)arr.getData(),

                  1, arr.getSize(), 1);

      cudaDeviceSynchronize();

   }

};

src/Benchmarks/Sorting/ReferenceAlgorithms/ThrustRadixsort.h

0 → 100644
+13 −0
Original line number Diff line number Diff line 
#include <thrust/sort.h>

#include <thrust/execution_policy.h>

#include <TNL/Containers/Array.h>





struct ThrustRadixsort

{

   static void sort( Containers::ArrayView< int, Devices::Cuda >& view )

   {

      thrust::sort(thrust::device, view.getData(), view.getData() + view.getSize());

      cudaDeviceSynchronize();

   }

};

src/Benchmarks/Sorting/tnl-benchmark-sort.h

+35 −26
Original line number Diff line number Diff line
@@ -85,38 +85,47 @@ int main(int argc, char *argv[]) 
{

   if(argc == 1)

   {

        std::cout << "STL sort on CPU ... " << std::endl;

        start< STLSort >( cout, "\t" );

#ifdef HAVE_CUDA

      std::cout << "Quicksort on GPU ... " << std::endl;

      start< Quicksort >(cout, "\t");

      std::cout << "Bitonic sort on GPU ... " << std::endl;

      start< BitonicSort >( cout, "\t" );

#ifdef HAVE_CUDA

#ifdef HAVE_CUDA_SAMPLES

      std::cout << "Manca quicksort on GPU ... " << std::endl;

      start< MancaQuicksort >( cout, "\t" );

      std::cout << "Nvidia bitonic sort on GPU ... " << std::endl;

      start< NvidiaBitonicSort >( cout, "\t" );

#endif

      std::cout << "Cederman quicksort on GPU ... " << std::endl;

      start< CedermanQuicksort >( cout, "\t" );

      std::cout << "Thrust radixsort on GPU ... " << std::endl;

      start< ThrustRadixsort >( cout, "\t" );

#endif

      std::cout << "STL sort on CPU ... " << std::endl;

      start< STLSort >( cout, "\t" );

   }

   else

   {

      std::ofstream out(argv[1]);

        std::cout << "STL sort on CPU ... " << std::endl;

        start< STLSort >( out, "," );

#ifdef HAVE_CUDA

      std::cout << "Quicksort on GPU ... " << std::endl;

      start< Quicksort >(out, ",");

      std::cout << "Bitonic sort on GPU ... " << std::endl;

      start< BitonicSort >(out, ",");

#ifdef HAVE_CUDA



#ifdef HAVE_CUDA_SAMPLES

      std::cout << "Manca quicksort on GPU ... " << std::endl;

      start< MancaQuicksort >( out, "," );

      std::cout << "Nvidia bitonic sort on GPU ... " << std::endl;

      start< NvidiaBitonicSort >( out, "," );

#endif

      std::cout << "Cederman quicksort on GPU ... " << std::endl;

      start< CedermanQuicksort >( out, "," );

      std::cout << "Thrust radixsort on GPU ... " << std::endl;

      start< ThrustRadixsort >( out, "," );

#endif

      std::cout << "STL sort on CPU ... " << std::endl;

      start< STLSort >( out, "," );

    }

    return 0;

}