testBlockedUnifCuda.cpp 10.7 KB
Newer Older
1 2 3 4
// ==== CMAKE =====
// @FUSE_BLAS
// @FUSE_FFT
// @FUSE_STARPU
BRAMAS Berenger's avatar
BRAMAS Berenger committed
5
// @FUSE_CUDA
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
// ================
// Keep in private GIT


#include "../../Src/Utils/FGlobal.hpp"

#include "../../Src/GroupTree/Core/FGroupTree.hpp"

#include "../../Src/Components/FSimpleLeaf.hpp"
#include "../../Src/Containers/FVector.hpp"

#include "../../Src/Kernels/P2P/FP2PParticleContainer.hpp"

#include "Kernels/Interpolation/FInterpMatrixKernel.hpp"
#include "../../Src/Kernels/Uniform/FUnifKernel.hpp"

#include "../../Src/GroupTree/Uniform/FUnifCellPOD.hpp"
23
#include "../../Src/GroupTree/Uniform/FUnifCudaCellPOD.hpp"
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

#include "../../Src/Utils/FMath.hpp"
#include "../../Src/Utils/FMemUtils.hpp"
#include "../../Src/Utils/FParameters.hpp"

#include "../../Src/Files/FRandomLoader.hpp"
#include "../../Src/Files/FFmaGenericLoader.hpp"

#include "../../Src/GroupTree/Core/FGroupSeqAlgorithm.hpp"
#include "../../Src/GroupTree/Core/FGroupTaskAlgorithm.hpp"

#include "../../Src/GroupTree/Core/FGroupTaskStarpuAlgorithm.hpp"
#include "../../Src/GroupTree/StarPUUtils/FStarPUKernelCapacities.hpp"

#include "../../Src/GroupTree/Core/FP2PGroupParticleContainer.hpp"

#include "../../Src/GroupTree/Cuda/FCudaDeviceWrapper.hpp"
#include "../../Src/GroupTree/Cuda/FCudaEmptyCellSymb.hpp"
#include "../../Src/GroupTree/Cuda/FCudaGroupOfParticles.hpp"
#include "../../Src/GroupTree/Cuda/FCudaGroupOfCells.hpp"

45
#include "../../Src/GroupTree/Uniform/FUnifCudaSharedData.hpp"
46 47 48 49 50

#include "../../Src/Utils/FParameterNames.hpp"

#include <memory>

51
template <class FReal, int ORDER>
52
class FUnifCuda;
53

54
#define RANDOM_PARTICLES
55 56 57 58

int main(int argc, char* argv[]){
    const FParameterNames LocalOptionBlocSize { {"-bs"}, "The size of the block of the blocked tree"};
    const FParameterNames LocalOptionNoValidate { {"-no-validation"}, "To avoid comparing with direct computation"};
59
    FHelpDescribeAndExit(argc, argv, "Perform Lagrange Kernel based simulation with CUDA+StarPU.",
60 61 62 63 64 65 66 67 68 69 70
                         FParameterDefinitions::OctreeHeight,
#ifdef RANDOM_PARTICLES
                         FParameterDefinitions::NbParticles,
#else
                         FParameterDefinitions::InputFile,
#endif
                         FParameterDefinitions::NbThreads,
                         LocalOptionBlocSize, LocalOptionNoValidate);

    // Initialize the types
    typedef double FReal;
71
    static const int ORDER = 5;
72 73 74 75 76 77 78 79
    typedef FInterpMatrixKernelR<FReal> MatrixKernelClass;

    typedef FUnifCellPODCore         GroupCellSymbClass;
    typedef FUnifCellPODPole<FReal,ORDER>  GroupCellUpClass;
    typedef FUnifCellPODLocal<FReal,ORDER> GroupCellDownClass;
    typedef FUnifCellPOD<FReal,ORDER>      GroupCellClass;

    typedef FP2PGroupParticleContainer<FReal>          GroupContainerClass;
80
    typedef FGroupTree< FReal, GroupCellSymbClass, GroupCellUpClass, GroupCellDownClass, GroupContainerClass, 1, 4, FReal>  GroupOctreeClass;
81

BRAMAS Berenger's avatar
BRAMAS Berenger committed
82 83 84
    //typedef FStarPUCudaM2LCapacities<FUnifKernel<FReal,GroupCellClass,GroupContainerClass,MatrixKernelClass,ORDER>> GroupKernelClass;
    //typedef FStarPUCudaP2PCapacities<FUnifKernel<FReal,GroupCellClass,GroupContainerClass,MatrixKernelClass,ORDER>> GroupKernelClass;
    typedef FStarPUCudaP2PM2LCapacities<FUnifKernel<FReal,GroupCellClass,GroupContainerClass,MatrixKernelClass,ORDER>> GroupKernelClass;
85 86 87
    typedef FStarPUCpuWrapper<typename GroupOctreeClass::CellGroupClass, GroupCellClass, GroupKernelClass, typename GroupOctreeClass::ParticleGroupClass, GroupContainerClass> GroupCpuWrapper;

    typedef FStarPUCudaWrapper<GroupKernelClass,
88 89 90
            FBasicCellPOD, FCudaUnifCellPODPole<FReal,ORDER>,FCudaUnifCellPODLocal<FReal,ORDER>,
            FCudaGroupOfCells<FBasicCellPOD, FCudaUnifCellPODPole<FReal,ORDER>,FCudaUnifCellPODLocal<FReal,ORDER> >,
            FCudaGroupOfParticles<FReal, 1, 4, FReal>, FCudaGroupAttachedLeaf<FReal, 1, 4, FReal>, FUnifCuda<FReal,ORDER> > GroupCudaWrapper;
91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133

    typedef FGroupTaskStarPUAlgorithm<GroupOctreeClass, typename GroupOctreeClass::CellGroupClass, GroupKernelClass, typename GroupOctreeClass::ParticleGroupClass,
            GroupCpuWrapper, GroupCudaWrapper > GroupAlgorithm;

    // Get params
    const int NbLevels      = FParameters::getValue(argc,argv,FParameterDefinitions::OctreeHeight.options, 5);
    const int groupSize     = FParameters::getValue(argc,argv,LocalOptionBlocSize.options, 250);

    // Load the particles
#ifdef RANDOM_PARTICLES
    FRandomLoader<FReal> loader(FParameters::getValue(argc,argv,FParameterDefinitions::NbParticles.options, 2000), 1.0, FPoint<FReal>(0,0,0), 0);
#else
    const char* const filename = FParameters::getStr(argc,argv,FParameterDefinitions::InputFile.options, "../Data/test20k.fma");
    FFmaGenericLoader<FReal> loader(filename);
#endif
    FAssertLF(loader.isOpen());
    FTic timer;

    FP2PParticleContainer<FReal> allParticles;
    for(FSize idxPart = 0 ; idxPart < loader.getNumberOfParticles() ; ++idxPart){
        FPoint<FReal> particlePosition;
        FReal physicalValue;
#ifdef RANDOM_PARTICLES
        physicalValue = 0.10;
        loader.fillParticle(&particlePosition);
#else
        loader.fillParticle(&particlePosition, &physicalValue);
#endif
        allParticles.push(particlePosition, physicalValue);
    }
    std::cout << "Particles loaded in " << timer.tacAndElapsed() << "s\n";

    // Put the data into the tree
    timer.tic();
    GroupOctreeClass groupedTree(NbLevels, loader.getBoxWidth(), loader.getCenterOfBox(), groupSize, &allParticles);
    groupedTree.printInfoBlocks();
    std::cout << "Tree created in " << timer.tacAndElapsed() << "s\n";

    // Run the algorithm
    const MatrixKernelClass MatrixKernel;
    GroupKernelClass groupkernel(NbLevels, loader.getBoxWidth(), loader.getCenterOfBox(), &MatrixKernel);
    GroupAlgorithm groupalgo(&groupedTree,&groupkernel);

134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
    {
        typedef FUnifM2LHandler<FReal, ORDER,MatrixKernelClass::Type> M2LHandlerClass;
        const M2LHandlerClass M2LHandler(&MatrixKernel,
                                         NbLevels,
                                         loader.getBoxWidth(),
                                         1);
        // Copy precomputed matrix to cuda
        FUnifCudaSharedData<FReal, ORDER> hostData;
        hostData.BoxWidth = loader.getBoxWidth();

        std::cout << "Copy precomputed tab of size " << hostData.ninteractions << " " << hostData.opt_rc << std::endl;

        for(int idxInter = 0 ; idxInter < hostData.ninteractions ; ++idxInter){
            for(int idx = 0 ; idx < hostData.opt_rc ; ++idx){
                hostData.FC[hostData.opt_rc*idxInter + idx].complex[0] = M2LHandler.getFc(idxInter, idx).getReal();
                hostData.FC[hostData.opt_rc*idxInter + idx].complex[1] = M2LHandler.getFc(idxInter, idx).getImag();
            }
        }
        std::cout << "Copy to GPU" << std::endl;

        groupalgo.forEachCudaWorker([&](void* cudaKernel){
            FUnifCudaFillObject(cudaKernel,hostData);
        });

        std::cout << "Done" << std::endl;
    }

161 162 163 164 165
    timer.tic();
    groupalgo.execute();
    std::cout << "Kernel executed in in " << timer.tacAndElapsed() << "s\n";

    // Validate the result
BRAMAS Berenger's avatar
BRAMAS Berenger committed
166
    if(FParameters::existParameter(argc, argv, LocalOptionNoValidate.options) == false){
167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
        FSize offsetParticles = 0;
        FReal*const allPhysicalValues = allParticles.getPhysicalValues();
        FReal*const allPosX = const_cast<FReal*>( allParticles.getPositions()[0]);
        FReal*const allPosY = const_cast<FReal*>( allParticles.getPositions()[1]);
        FReal*const allPosZ = const_cast<FReal*>( allParticles.getPositions()[2]);

        groupedTree.forEachCellLeaf<FP2PGroupParticleContainer<FReal> >([&](GroupCellClass cellTarget, FP2PGroupParticleContainer<FReal> * leafTarget){
            const FReal*const physicalValues = leafTarget->getPhysicalValues();
            const FReal*const posX = leafTarget->getPositions()[0];
            const FReal*const posY = leafTarget->getPositions()[1];
            const FReal*const posZ = leafTarget->getPositions()[2];
            const FSize nbPartsInLeafTarget = leafTarget->getNbParticles();

            for(FSize idxPart = 0 ; idxPart < nbPartsInLeafTarget ; ++idxPart){
                allPhysicalValues[offsetParticles + idxPart] = physicalValues[idxPart];
                allPosX[offsetParticles + idxPart] = posX[idxPart];
                allPosY[offsetParticles + idxPart] = posY[idxPart];
                allPosZ[offsetParticles + idxPart] = posZ[idxPart];
            }

            offsetParticles += nbPartsInLeafTarget;
        });

        FAssertLF(offsetParticles == loader.getNumberOfParticles());

        FReal*const allDirectPotentials = allParticles.getPotentials();
        FReal*const allDirectforcesX = allParticles.getForcesX();
        FReal*const allDirectforcesY = allParticles.getForcesY();
        FReal*const allDirectforcesZ = allParticles.getForcesZ();

        for(int idxTgt = 0 ; idxTgt < offsetParticles ; ++idxTgt){
            for(int idxMutual = idxTgt + 1 ; idxMutual < offsetParticles ; ++idxMutual){
                FP2PR::MutualParticles(
                    allPosX[idxTgt],allPosY[idxTgt],allPosZ[idxTgt], allPhysicalValues[idxTgt],
                    &allDirectforcesX[idxTgt], &allDirectforcesY[idxTgt], &allDirectforcesZ[idxTgt], &allDirectPotentials[idxTgt],
                    allPosX[idxMutual],allPosY[idxMutual],allPosZ[idxMutual], allPhysicalValues[idxMutual],
                    &allDirectforcesX[idxMutual], &allDirectforcesY[idxMutual], &allDirectforcesZ[idxMutual], &allDirectPotentials[idxMutual]
                );
            }
        }

        FMath::FAccurater<FReal> potentialDiff;
        FMath::FAccurater<FReal> fx, fy, fz;
        offsetParticles = 0;
        groupedTree.forEachCellLeaf<FP2PGroupParticleContainer<FReal> >([&](GroupCellClass cellTarget, FP2PGroupParticleContainer<FReal> * leafTarget){
            const FReal*const potentials = leafTarget->getPotentials();
            const FReal*const forcesX = leafTarget->getForcesX();
            const FReal*const forcesY = leafTarget->getForcesY();
            const FReal*const forcesZ = leafTarget->getForcesZ();
            const FSize nbPartsInLeafTarget = leafTarget->getNbParticles();

            for(int idxTgt = 0 ; idxTgt < nbPartsInLeafTarget ; ++idxTgt){
                potentialDiff.add(allDirectPotentials[idxTgt + offsetParticles], potentials[idxTgt]);
                fx.add(allDirectforcesX[idxTgt + offsetParticles], forcesX[idxTgt]);
                fy.add(allDirectforcesY[idxTgt + offsetParticles], forcesY[idxTgt]);
                fz.add(allDirectforcesZ[idxTgt + offsetParticles], forcesZ[idxTgt]);
            }

            offsetParticles += nbPartsInLeafTarget;
        });

        std::cout << "Error : Potential " << potentialDiff << "\n";
        std::cout << "Error : fx " << fx << "\n";
        std::cout << "Error : fy " << fy << "\n";
        std::cout << "Error : fz " << fz << "\n";
    }

    return 0;
}