FFmmAlgorithmThreadProc.hpp 66.3 KB
Newer Older
1
// ===================================================================================
2
// Copyright ScalFmm 2011 INRIA, Olivier Coulaud, Berenger Bramas, Matthias Messner
3 4 5 6
// olivier.coulaud@inria.fr, berenger.bramas@inria.fr
// This software is a computer program whose purpose is to compute the FMM.
//
// This software is governed by the CeCILL-C and LGPL licenses and
7 8
// abiding by the rules of distribution of free software.
//
9 10 11 12
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
// GNU General Public and CeCILL-C Licenses for more details.
13
// "http://www.cecill.info".
14
// "http://www.gnu.org/licenses".
15
// ===================================================================================
16 17
#ifndef FFMMALGORITHMTHREADPROC_HPP
#define FFMMALGORITHMTHREADPROC_HPP
18

COULAUD Olivier's avatar
COULAUD Olivier committed
19
#include <omp.h>
20

COULAUD Olivier's avatar
COULAUD Olivier committed
21
//
BRAMAS Berenger's avatar
BRAMAS Berenger committed
22
#include "../Utils/FAssert.hpp"
BRAMAS Berenger's avatar
BRAMAS Berenger committed
23
#include "../Utils/FLog.hpp"
24 25
#include "../Utils/FTrace.hpp"
#include "../Utils/FTic.hpp"
26

27 28
#include "../Utils/FGlobal.hpp"

29
#include "../Containers/FBoolArray.hpp"
30
#include "../Containers/FOctree.hpp"
berenger-bramas's avatar
berenger-bramas committed
31
#include "../Containers/FLightOctree.hpp"
32

33 34
#include "../Containers/FBufferWriter.hpp"
#include "../Containers/FBufferReader.hpp"
35 36
#include "../Containers/FMpiBufferWriter.hpp"
#include "../Containers/FMpiBufferReader.hpp"
37

berenger-bramas's avatar
berenger-bramas committed
38
#include "../Utils/FMpi.hpp"
39
#include <sys/time.h>
40

41
#include "FCoreCommon.hpp"
42

43
/**
44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62
 * @author Berenger Bramas (berenger.bramas@inria.fr)
 * @class FFmmAlgorithmThreadProc
 * @brief
 * Please read the license
 *
 * This class is a threaded FMM algorithm with mpi.
 * It just iterates on a tree and call the kernels with good arguments.
 * It used the inspector-executor model :
 * iterates on the tree and builds an array to work in parallel on this array
 *
 * Of course this class does not deallocate pointer given in arguements.
 *
 * Threaded & based on the inspector-executor model
 * schedule(runtime) export OMP_NUM_THREADS=2
 * export OMPI_CXX=`which g++-4.4`
 * mpirun -np 2 valgrind --suppressions=/usr/share/openmpi/openmpi-valgrind.supp
 * --tool=memcheck --leak-check=yes --show-reachable=yes --num-callers=20 --track-fds=yes
 * ./Tests/testFmmAlgorithmProc ../Data/testLoaderSmall.fma.tmp
 */
63
template<class OctreeClass, class CellClass, class ContainerClass, class KernelClass, class LeafClass>
BRAMAS Berenger's avatar
BRAMAS Berenger committed
64
class FFmmAlgorithmThreadProc : public FAbstractAlgorithm {
65 66
    // Can be deleted
    // const static int MaxSizePerCell = CellClass::GetSize();
67

68 69
    OctreeClass* const tree;                 //< The octree to work on
    KernelClass** kernels;                   //< The kernels
70

71
    const FMpi::FComm& comm;                 //< MPI comm
72

73 74 75
    typename OctreeClass::Iterator*     iterArray;  //Will be used to store pointers to cells/leafs to work with
    typename OctreeClass::Iterator* iterArrayComm;  //Will be used to store pointers to cells/leafs to send/rcv
    int numberOfLeafs;                          //< To store the size at the previous level
76

77
    const int MaxThreads;               //< the max number of thread allowed by openmp
berenger-bramas's avatar
berenger-bramas committed
78

79 80
    const int nbProcess;                //< Number of process
    const int idProcess;                //< Id of current process
81

82
    const int OctreeHeight;            //<Height of the tree
berenger-bramas's avatar
berenger-bramas committed
83

84

85 86 87 88
    /** An interval is the morton index interval
     * that a proc use (it holds data in this interval)
     */
    struct Interval{
89 90
        MortonIndex leftIndex;
        MortonIndex rightIndex;
91 92 93 94 95
    };
    /** My interval */
    Interval*const intervals;
    /** All process intervals */
    Interval*const workingIntervalsPerLevel;
96

97 98
    /** Get an interval from proc id and level */
    Interval& getWorkingInterval( int level,  int proc){
99
        return workingIntervalsPerLevel[OctreeHeight * proc + level];
100
    }
101

102
    const Interval& getWorkingInterval( int level,  int proc) const {
103
        return workingIntervalsPerLevel[OctreeHeight * proc + level];
104 105 106 107
    }

    /** To know if a proc has work at a given level (if it hold cells and was responsible of them) */
    bool procHasWorkAtLevel(const int idxLevel , const int idxProc) const {
108
        return getWorkingInterval(idxLevel, idxProc).leftIndex <= getWorkingInterval(idxLevel, idxProc).rightIndex;
109 110 111 112
    }

    /** Return true if the idxProc left cell at idxLevel+1 has the same parent as us for our right cell */
    bool procCoversMyRightBorderCell(const int idxLevel , const int idxProc) const {
113
        return (getWorkingInterval((idxLevel+1) , idProcess).rightIndex>>3) == (getWorkingInterval((idxLevel+1) ,idxProc).leftIndex >>3);
114 115 116 117
    }

    /** Return true if the idxProc right cell at idxLevel+1 has the same parent as us for our left cell */
    bool procCoversMyLeftBorderCell(const int idxLevel , const int idxProc) const {
118
        return (getWorkingInterval((idxLevel+1) , idxProc).rightIndex >>3) == (getWorkingInterval((idxLevel+1) , idProcess).leftIndex>>3);
119 120
    }

121 122 123
public:
    /** Get current proc interval at level */
    Interval& getWorkingInterval( int level){
124
        return getWorkingInterval(level, idProcess);
125 126 127 128
    }

    /** Does the current proc has some work at this level */
    bool hasWorkAtLevel( int level){
129
        return idProcess == 0 || (getWorkingInterval(level, idProcess - 1).rightIndex) < (getWorkingInterval(level, idProcess).rightIndex);
130 131 132 133 134 135 136 137
    }

    /** The constructor need the octree and the kernels used for computation
     * @param inTree the octree to work on
     * @param inKernels the kernels to call
     * An assert is launched if one of the arguments is null
     */
    FFmmAlgorithmThreadProc(const FMpi::FComm& inComm, OctreeClass* const inTree, KernelClass* const inKernels)
138 139 140 141
        : tree(inTree) , kernels(nullptr), comm(inComm), iterArray(nullptr),iterArrayComm(nullptr),numberOfLeafs(0),
          MaxThreads(omp_get_max_threads()), nbProcess(inComm.processCount()), idProcess(inComm.processId()),
          OctreeHeight(tree->getHeight()),intervals(new Interval[inComm.processCount()]),
          workingIntervalsPerLevel(new Interval[inComm.processCount() * tree->getHeight()])
142
    {
143
        FAssertLF(tree, "tree cannot be null");
144 145


146 147 148 149
        this->kernels = new KernelClass*[MaxThreads];
        for(int idxThread = 0 ; idxThread < MaxThreads ; ++idxThread){
            this->kernels[idxThread] = new KernelClass(*inKernels);
        }
150

151 152
        FLOG(FLog::Controller << "FFmmAlgorithmThreadProc\n");
        FLOG(FLog::Controller << "Max threads = "  << MaxThreads << ", Procs = " << nbProcess << ", I am " << idProcess << ".\n");
153 154 155
    }
    /** Default destructor */
    virtual ~FFmmAlgorithmThreadProc(){
156 157 158 159
        for(int idxThread = 0 ; idxThread < MaxThreads ; ++idxThread){
            delete this->kernels[idxThread];
        }
        delete [] this->kernels;
160

161 162
        delete [] intervals;
        delete [] workingIntervalsPerLevel;
163 164 165 166 167 168 169
    }

    /**
     * To execute the fmm algorithm
     * Call this function to run the complete algorithm
     */
    void execute(const unsigned operationsToProceed = FFmmNearAndFarFields){
170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
        // Count leaf
        this->numberOfLeafs = 0;
        {
            Interval myFullInterval;
            {//Building the interval with the first and last leaves (and count the number of leaves)
                typename OctreeClass::Iterator octreeIterator(tree);
                octreeIterator.gotoBottomLeft();
                myFullInterval.leftIndex = octreeIterator.getCurrentGlobalIndex();
                do{
                    ++this->numberOfLeafs;
                } while(octreeIterator.moveRight());
                myFullInterval.rightIndex = octreeIterator.getCurrentGlobalIndex();
            }
            // Allocate a number to store the pointer of the cells at a level
            iterArray     = new typename OctreeClass::Iterator[numberOfLeafs];
            iterArrayComm = new typename OctreeClass::Iterator[numberOfLeafs];
            FAssertLF(iterArray,     "iterArray     bad alloc");
            FAssertLF(iterArrayComm, "iterArrayComm bad alloc");

            // We get the leftIndex/rightIndex indexes from each procs
            FMpi::MpiAssert( MPI_Allgather( &myFullInterval, sizeof(Interval), MPI_BYTE, intervals, sizeof(Interval), MPI_BYTE, comm.getComm()),  __LINE__ );

            // Build my intervals for all levels
            std::unique_ptr<Interval[]> myIntervals(new Interval[OctreeHeight]);
            // At leaf level we know it is the full interval
            myIntervals[OctreeHeight - 1] = myFullInterval;

            // We can estimate the interval for each level by using the parent/child relation
            for(int idxLevel = OctreeHeight - 2 ; idxLevel >= 0 ; --idxLevel){
                myIntervals[idxLevel].leftIndex = myIntervals[idxLevel+1].leftIndex >> 3;
                myIntervals[idxLevel].rightIndex = myIntervals[idxLevel+1].rightIndex >> 3;
            }

            // Process 0 uses the estimates as real intervals, but other processes
            // should remove cells that belong to others
            if(idProcess != 0){
                //We test for each level if process on left (idProcess-1) own cell I thought I owned
                typename OctreeClass::Iterator octreeIterator(tree);
                octreeIterator.gotoBottomLeft();
                octreeIterator.moveUp();

                // At h-1 the working limit is the parent of the right cell of the proc on the left
                MortonIndex workingLimitAtLevel = intervals[idProcess-1].rightIndex >> 3;

                // We check if we have no more work to do
                int nullIntervalFromLevel = 0;

                for(int idxLevel = OctreeHeight - 2 ; idxLevel >= 1 && nullIntervalFromLevel == 0 ; --idxLevel){
                    while(octreeIterator.getCurrentGlobalIndex() <= workingLimitAtLevel){
                        if( !octreeIterator.moveRight() ){
                            // We cannot move right we are not owner of any more cell
                            nullIntervalFromLevel = idxLevel;
                            break;
                        }
                    }
                    // If we are responsible for some cells at this level keep the first index
                    if(nullIntervalFromLevel == 0){
                        myIntervals[idxLevel].leftIndex = octreeIterator.getCurrentGlobalIndex();
                        octreeIterator.moveUp();
                        workingLimitAtLevel >>= 3;
                    }
                }
                // In case we are not responsible for any cells we put the leftIndex = rightIndex+1
                for(int idxLevel = nullIntervalFromLevel ; idxLevel >= 1 ; --idxLevel){
                    myIntervals[idxLevel].leftIndex = myIntervals[idxLevel].rightIndex + 1;
                }
            }

            // We get the leftIndex/rightIndex indexes from each procs
            FMpi::MpiAssert( MPI_Allgather( myIntervals.get(), int(sizeof(Interval)) * OctreeHeight, MPI_BYTE,
                                            workingIntervalsPerLevel, int(sizeof(Interval)) * OctreeHeight, MPI_BYTE, comm.getComm()),  __LINE__ );
        }

        // run;
        if(operationsToProceed & FFmmP2M) bottomPass();

        if(operationsToProceed & FFmmM2M) upwardPass();

        if(operationsToProceed & FFmmM2L) transferPass();

        if(operationsToProceed & FFmmL2L) downardPass();

        if((operationsToProceed & FFmmP2P) || (operationsToProceed & FFmmL2P)) directPass();


        // delete array
        delete []     iterArray;
        delete [] iterArrayComm;
        iterArray     = nullptr;
        iterArrayComm = nullptr;
260
    }
261 262 263

private:

264 265 266
    /////////////////////////////////////////////////////////////////////////////
    // P2M
    /////////////////////////////////////////////////////////////////////////////
267

268 269 270 271 272
    /**
     * P2M Bottom Pass
     * No communication are involved in the P2M.
     * It is similar to multi threaded version.
     */
273
    void bottomPass(){
274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299
        FLOG( FLog::Controller.write("\tStart Bottom Pass\n").write(FLog::Flush) );
        FLOG(FTic counterTime);
        FLOG(FTic computationCounter);
        typename OctreeClass::Iterator octreeIterator(tree);

        // Copy the ptr to leaves in array
        octreeIterator.gotoBottomLeft();
        int leafs = 0;
        do{
            iterArray[leafs++] = octreeIterator;
        } while(octreeIterator.moveRight());

        FLOG(computationCounter.tic());
#pragma omp parallel
        {
            // Each thread get its own kernel
            KernelClass * const myThreadkernels = kernels[omp_get_thread_num()];
            // Parallel iteration on the leaves
#pragma omp for nowait
            for(int idxLeafs = 0 ; idxLeafs < leafs ; ++idxLeafs){
                myThreadkernels->P2M( iterArray[idxLeafs].getCurrentCell() , iterArray[idxLeafs].getCurrentListSrc());
            }
        }
        FLOG(computationCounter.tac());
        FLOG( FLog::Controller << "\tFinished (@Bottom Pass (P2M) = "  << counterTime.tacAndElapsed() << " s)\n" );
        FLOG( FLog::Controller << "\t\t Computation : " << computationCounter.elapsed() << " s\n" );
300 301 302 303 304 305 306 307
    }

    /////////////////////////////////////////////////////////////////////////////
    // Upward
    /////////////////////////////////////////////////////////////////////////////

    /** M2M */
    void upwardPass(){
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496
        const int MaxSizePerCell = CellClass::GetSize();
        FLOG( FLog::Controller.write("\tStart Upward Pass\n").write(FLog::Flush); );
        FLOG(FTic counterTime);
        FLOG(FTic computationCounter);
        FLOG(FTic singleCounter);
        FLOG(FTic parallelCounter);

        // Start from leal level (height-1)
        typename OctreeClass::Iterator octreeIterator(tree);
        octreeIterator.gotoBottomLeft();
        octreeIterator.moveUp();
        typename OctreeClass::Iterator avoidGotoLeftIterator(octreeIterator);

        // The proc to send the shared cells to
        // Starting to the proc on the left this variable will go to 0
        int currentProcIdToSendTo = (idProcess - 1);

        // There are a maximum of 1 sends and 8-1 receptions
        MPI_Request requests[8];
        MPI_Status status[8];

        // Maximum data per message is:
        FMpiBufferWriter sendBuffer(comm.getComm(), 7*MaxSizePerCell + 1);
        const int recvBufferOffset = (7 * MaxSizePerCell + 1);
        FMpiBufferReader recvBuffer(comm.getComm(), 7*recvBufferOffset);
        CellClass recvBufferCells[8];

        // The first proc that send to me a cell
        // This variable will go to nbProcess
        int firstProcThatSend = idProcess + 1;
        FLOG(computationCounter.tic());

        // We work from height-1 to 1
        for(int idxLevel = OctreeHeight - 2 ; idxLevel > 1 ; --idxLevel ){
            // Does my cells are covered by my neighbors working interval and so I have no more work?
            const bool noMoreWorkForMe = (idProcess != 0 && !procHasWorkAtLevel(idxLevel+1, idProcess));
            if(noMoreWorkForMe){
                FAssertLF(procHasWorkAtLevel(idxLevel, idProcess) == false);
                break;
            }

            // Copy and count ALL the cells (even the ones outside the working interval)
            int totalNbCellsAtLevel = 0;
            do{
                iterArray[totalNbCellsAtLevel++] = octreeIterator;
            } while(octreeIterator.moveRight());
            avoidGotoLeftIterator.moveUp();
            octreeIterator = avoidGotoLeftIterator;

            int iterMpiRequests   = 0; // The iterator for send/recv requests

            int nbCellsToSkip     = 0; // The number of cells to send
            // Skip all the cells that are out of my working interval
            while(nbCellsToSkip < totalNbCellsAtLevel && iterArray[nbCellsToSkip].getCurrentGlobalIndex() < getWorkingInterval(idxLevel, idProcess).leftIndex){
                ++nbCellsToSkip;
            }

            // We need to know if we will recv something in order to know if threads skip the last cell
            int nbCellsForThreads = totalNbCellsAtLevel; // totalNbCellsAtLevel or totalNbCellsAtLevel-1
            bool hasToReceive = false;
            if(idProcess != nbProcess-1 && procHasWorkAtLevel(idxLevel , idProcess)){
                // Find the first proc that may send to me
                while(firstProcThatSend < nbProcess && !procHasWorkAtLevel(idxLevel+1, firstProcThatSend) ){
                    firstProcThatSend += 1;
                }
                // Do we have to receive?
                if(firstProcThatSend < nbProcess && procHasWorkAtLevel(idxLevel+1, firstProcThatSend) && procCoversMyRightBorderCell(idxLevel, firstProcThatSend) ){
                    hasToReceive = true;
                    // Threads do not compute the last cell, we will do it once data are received
                    nbCellsForThreads -= 1;
                }
            }

            FLOG(parallelCounter.tic());
#pragma omp parallel
            {
                const int threadNumber = omp_get_thread_num();
                KernelClass* myThreadkernels = (kernels[threadNumber]);
                //This single section post and receive the comms, and then do the M2M associated with it.
#pragma omp single nowait
                {
                    FLOG(singleCounter.tic());
                    // Master proc never send
                    if(idProcess != 0){
                        // Skip process that have no work at that level
                        while( currentProcIdToSendTo && !procHasWorkAtLevel(idxLevel, currentProcIdToSendTo)  ){
                            --currentProcIdToSendTo;
                        }
                        // Does the next proc that has work is sharing the parent of my left cell
                        if(procHasWorkAtLevel(idxLevel, currentProcIdToSendTo) && procCoversMyLeftBorderCell(idxLevel, currentProcIdToSendTo)){
                            FAssertLF(nbCellsToSkip != 0);

                            char packageFlags = 0;
                            sendBuffer.write(packageFlags);

                            // Only the cell the most on the right out of my working interval should be taken in
                            // consideration (at pos nbCellsToSkip-1) other (x < nbCellsToSkip-1) have already been sent
                            const CellClass* const* const child = iterArray[nbCellsToSkip-1].getCurrentChild();
                            for(int idxChild = 0 ; idxChild < 8 ; ++idxChild){
                                // Check if child exists and it was part of my working interval
                                if( child[idxChild] && getWorkingInterval((idxLevel+1), idProcess).leftIndex <= child[idxChild]->getMortonIndex() ){
                                    // Add the cell to the buffer
                                    child[idxChild]->serializeUp(sendBuffer);
                                    packageFlags = char(packageFlags | (0x1 << idxChild));
                                }
                            }
                            // Add the flag as first value
                            sendBuffer.writeAt(0,packageFlags);
                            // Post the message
                            MPI_Isend(sendBuffer.data(), sendBuffer.getSize(), MPI_PACKED, currentProcIdToSendTo,
                                      FMpi::TagFmmM2M + idxLevel, comm.getComm(), &requests[iterMpiRequests++]);
                        }
                    }

                    //Post receive, Datas needed in several parts of the section
                    int nbProcThatSendToMe = 0;

                    if(hasToReceive){
                        //Test : if the firstProcThatSend father minimal value in interval is lesser than mine
                        int idProcSource = firstProcThatSend;
                        // Find the last proc that should send to me
                        while( idProcSource < nbProcess
                               && ( !procHasWorkAtLevel(idxLevel+1, idProcSource) || procCoversMyRightBorderCell(idxLevel, idProcSource) )){
                            if(procHasWorkAtLevel(idxLevel+1, idProcSource) && procCoversMyRightBorderCell(idxLevel, idProcSource)){
                                MPI_Irecv(&recvBuffer.data()[nbProcThatSendToMe * recvBufferOffset], recvBufferOffset, MPI_PACKED,
                                        idProcSource, FMpi::TagFmmM2M + idxLevel, comm.getComm(), &requests[iterMpiRequests++]);
                                nbProcThatSendToMe += 1;
                                FAssertLF(nbProcThatSendToMe <= 7);
                            }
                            ++idProcSource;
                        }
                    }

                    //Wait For the comms, and do the work
                    // Are we sending or waiting anything?
                    if(iterMpiRequests){
                        FAssertLF(iterMpiRequests <= 8);
                        MPI_Waitall( iterMpiRequests, requests, status);
                    }
                    // We had received something so we need to proceed the last M2M
                    if( hasToReceive ){
                        FAssertLF(iterMpiRequests != 0);
                        CellClass* currentChild[8];
                        memcpy(currentChild, iterArray[totalNbCellsAtLevel - 1].getCurrentChild(), 8 * sizeof(CellClass*));

                        // Retreive data and merge my child and the child from others
                        for(int idxProc = 0 ; idxProc < nbProcThatSendToMe ; ++idxProc){
                            recvBuffer.seek(idxProc * recvBufferOffset);
                            int packageFlags = int(recvBuffer.getValue<char>());

                            int position = 0;
                            while( packageFlags && position < 8){
                                while(!(packageFlags & 0x1)){
                                    packageFlags >>= 1;
                                    ++position;
                                }
                                FAssertLF(!currentChild[position], "Already has a cell here");
                                recvBufferCells[position].deserializeUp(recvBuffer);
                                currentChild[position] = (CellClass*) &recvBufferCells[position];

                                packageFlags >>= 1;
                                ++position;
                            }
                        }
                        // Finally compute
                        (*kernels[threadNumber]).M2M( iterArray[totalNbCellsAtLevel - 1].getCurrentCell() , currentChild, idxLevel);
                        firstProcThatSend += nbProcThatSendToMe - 1;
                    }
                    // Reset buffer
                    sendBuffer.reset();
                    recvBuffer.seek(0);
                    FLOG(singleCounter.tac());
                }//End Of Single section

                // All threads proceed the M2M
#pragma omp for nowait
                for( int idxCell = nbCellsToSkip ; idxCell < nbCellsForThreads ; ++idxCell){
                    myThreadkernels->M2M( iterArray[idxCell].getCurrentCell() , iterArray[idxCell].getCurrentChild(), idxLevel);
                }
            }//End of parallel section
            FLOG(parallelCounter.tac());
        }

        FLOG(counterTime.tac());
        FLOG(computationCounter.tac());
        FLOG( FLog::Controller << "\tFinished (@Upward Pass (M2M) = "  << counterTime.elapsed() << " s)\n" );
        FLOG( FLog::Controller << "\t\t Computation : " << computationCounter.elapsed() << " s\n" );
        FLOG( FLog::Controller << "\t\t Wait : " << singleCounter.cumulated() << " s\n" );
        FLOG( FLog::Controller << "\t\t Wait : " << parallelCounter.cumulated() << " s\n" );
497
    }
498

499 500 501
    /////////////////////////////////////////////////////////////////////////////
    // Downard
    /////////////////////////////////////////////////////////////////////////////
502 503


504
    void transferPass(){
505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861
        const int MaxSizePerCell = CellClass::GetSize();
        FLOG( FLog::Controller.write("\tStart Downward Pass (M2L)\n").write(FLog::Flush); );
        FLOG(FTic counterTime);
        FLOG(FTic computationCounter);
        FLOG(FTic sendCounter);
        FLOG(FTic receiveCounter);
        FLOG(FTic prepareCounter);
        FLOG(FTic gatherCounter);

        //////////////////////////////////////////////////////////////////
        // First know what to send to who
        //////////////////////////////////////////////////////////////////

        // pointer to send
        FVector<typename OctreeClass::Iterator> toSend[nbProcess * OctreeHeight];
        // index
        int*const indexToSend = new int[nbProcess * OctreeHeight];
        memset(indexToSend, 0, sizeof(int) * nbProcess * OctreeHeight);
        // To know which one has need someone
        FBoolArray** const leafsNeedOther = new FBoolArray*[OctreeHeight];
        memset(leafsNeedOther, 0, sizeof(FBoolArray*) * OctreeHeight);

        // All process say to each others
        // what the will send to who
        int*const globalReceiveMap = new int[nbProcess * nbProcess * OctreeHeight];
        memset(globalReceiveMap, 0, sizeof(int) * nbProcess * nbProcess * OctreeHeight);

        FMpiBufferWriter**const sendBuffer = new FMpiBufferWriter*[nbProcess * OctreeHeight];
        memset(sendBuffer, 0, sizeof(FMpiBufferWriter*) * nbProcess * OctreeHeight);

        FMpiBufferReader**const recvBuffer = new FMpiBufferReader*[nbProcess * OctreeHeight];
        memset(recvBuffer, 0, sizeof(FMpiBufferReader*) * nbProcess * OctreeHeight);

#pragma omp parallel
        {
#pragma omp master
            {
                {
                    FLOG(prepareCounter.tic());

                    std::unique_ptr<typename OctreeClass::Iterator[]> iterArrayLocal(new typename OctreeClass::Iterator[numberOfLeafs]);

                    // To know if a leaf has been already sent to a proc
                    bool*const alreadySent = new bool[nbProcess];
                    memset(alreadySent, 0, sizeof(bool) * nbProcess);

                    typename OctreeClass::Iterator octreeIterator(tree);
                    octreeIterator.moveDown();
                    typename OctreeClass::Iterator avoidGotoLeftIterator(octreeIterator);
                    // for each levels
                    for(int idxLevel = 2 ; idxLevel < OctreeHeight ; ++idxLevel ){
                        if(!procHasWorkAtLevel(idxLevel, idProcess)){
                            avoidGotoLeftIterator.moveDown();
                            octreeIterator = avoidGotoLeftIterator;
                            continue;
                        }

                        int numberOfCells = 0;

                        while(octreeIterator.getCurrentGlobalIndex() <  getWorkingInterval(idxLevel , idProcess).leftIndex){
                            octreeIterator.moveRight();
                        }

                        // for each cells
                        do{
                            iterArrayLocal[numberOfCells] = octreeIterator;
                            ++numberOfCells;
                        } while(octreeIterator.moveRight());
                        avoidGotoLeftIterator.moveDown();
                        octreeIterator = avoidGotoLeftIterator;

                        leafsNeedOther[idxLevel] = new FBoolArray(numberOfCells);

                        // Which cell potentialy needs other data and in the same time
                        // are potentialy needed by other
                        MortonIndex neighborsIndexes[189];
                        for(int idxCell = 0 ; idxCell < numberOfCells ; ++idxCell){
                            // Find the M2L neigbors of a cell
                            const int counter = iterArrayLocal[idxCell].getCurrentGlobalCoordinate().getInteractionNeighbors(idxLevel, neighborsIndexes);

                            memset(alreadySent, false, sizeof(bool) * nbProcess);
                            bool needOther = false;
                            // Test each negibors to know which one do not belong to us
                            for(int idxNeigh = 0 ; idxNeigh < counter ; ++idxNeigh){
                                if(neighborsIndexes[idxNeigh] < getWorkingInterval(idxLevel , idProcess).leftIndex
                                        || (getWorkingInterval(idxLevel , idProcess).rightIndex) < neighborsIndexes[idxNeigh]){
                                    int procToReceive = idProcess;
                                    while( 0 != procToReceive && neighborsIndexes[idxNeigh] < getWorkingInterval(idxLevel , procToReceive).leftIndex ){
                                        --procToReceive;
                                    }
                                    while( procToReceive != nbProcess -1 && (getWorkingInterval(idxLevel , procToReceive).rightIndex) < neighborsIndexes[idxNeigh]){
                                        ++procToReceive;
                                    }
                                    // Maybe already sent to that proc?
                                    if( !alreadySent[procToReceive]
                                            && getWorkingInterval(idxLevel , procToReceive).leftIndex <= neighborsIndexes[idxNeigh]
                                            && neighborsIndexes[idxNeigh] <= getWorkingInterval(idxLevel , procToReceive).rightIndex){

                                        alreadySent[procToReceive] = true;

                                        needOther = true;

                                        toSend[idxLevel * nbProcess + procToReceive].push(iterArrayLocal[idxCell]);
                                        ++indexToSend[idxLevel * nbProcess + procToReceive];
                                    }
                                }
                            }
                            if(needOther){
                                leafsNeedOther[idxLevel]->set(idxCell,true);
                            }
                        }
                    }
                    FLOG(prepareCounter.tac());

                    delete[] alreadySent;
                }

                //////////////////////////////////////////////////////////////////
                // Gather this information
                //////////////////////////////////////////////////////////////////

                FLOG(gatherCounter.tic());
                FMpi::MpiAssert( MPI_Allgather( indexToSend, nbProcess * OctreeHeight, MPI_INT, globalReceiveMap, nbProcess * OctreeHeight, MPI_INT, comm.getComm()),  __LINE__ );
                FLOG(gatherCounter.tac());

                //////////////////////////////////////////////////////////////////
                // Send and receive for real
                //////////////////////////////////////////////////////////////////

                FLOG(sendCounter.tic());
                // Then they can send and receive (because they know what they will receive)
                // To send in asynchrone way
                MPI_Request*const requests = new MPI_Request[2 * nbProcess * OctreeHeight];
                MPI_Status*const status = new MPI_Status[2 * nbProcess * OctreeHeight];
                int iterRequest = 0;

                const int SizeOfCellToSend = sizeof(MortonIndex) + sizeof(int) + MaxSizePerCell;

                for(int idxLevel = 2 ; idxLevel < OctreeHeight ; ++idxLevel ){
                    for(int idxProc = 0 ; idxProc < nbProcess ; ++idxProc){
                        const int toSendAtProcAtLevel = indexToSend[idxLevel * nbProcess + idxProc];
                        if(toSendAtProcAtLevel != 0){
                            sendBuffer[idxLevel * nbProcess + idxProc] = new FMpiBufferWriter(comm.getComm(),toSendAtProcAtLevel * SizeOfCellToSend);

                            for(int idxLeaf = 0 ; idxLeaf < toSendAtProcAtLevel; ++idxLeaf){
                                const MortonIndex cellIndex = toSend[idxLevel * nbProcess + idxProc][idxLeaf].getCurrentGlobalIndex();
                                sendBuffer[idxLevel * nbProcess + idxProc]->write(cellIndex);
                                toSend[idxLevel * nbProcess + idxProc][idxLeaf].getCurrentCell()->serializeUp(*sendBuffer[idxLevel * nbProcess + idxProc]);
                            }

                            FMpi::MpiAssert( MPI_Isend( sendBuffer[idxLevel * nbProcess + idxProc]->data(),
                                             sendBuffer[idxLevel * nbProcess + idxProc]->getSize(),MPI_PACKED, idxProc,
                                    FMpi::TagLast + idxLevel, comm.getComm(), &requests[iterRequest++]) , __LINE__ );
                        }

                        const int toReceiveFromProcAtLevel = globalReceiveMap[(idxProc * nbProcess * OctreeHeight) + idxLevel * nbProcess + idProcess];
                        if(toReceiveFromProcAtLevel){
                            recvBuffer[idxLevel * nbProcess + idxProc] = new FMpiBufferReader(comm.getComm(),toReceiveFromProcAtLevel * SizeOfCellToSend);

                            FMpi::MpiAssert( MPI_Irecv(recvBuffer[idxLevel * nbProcess + idxProc]->data(),
                                             recvBuffer[idxLevel * nbProcess + idxProc]->getCapacity(), MPI_PACKED,idxProc,
                                    FMpi::TagLast + idxLevel, comm.getComm(), &requests[iterRequest++]) , __LINE__ );
                        }
                    }
                }

                //////////////////////////////////////////////////////////////////
                // Wait received data and compute
                //////////////////////////////////////////////////////////////////

                // Wait to receive every things (and send every things)
                MPI_Waitall(iterRequest, requests, status);

                delete[] requests;
                delete[] status;

                FLOG(sendCounter.tac());
            }

            //////////////////////////////////////////////////////////////////
            // Do M2L
            //////////////////////////////////////////////////////////////////

            KernelClass * const myThreadkernels = kernels[omp_get_thread_num()];
            const CellClass* neighbors[343];

#pragma omp single nowait
            {
                typename OctreeClass::Iterator octreeIterator(tree);
                octreeIterator.moveDown();
                typename OctreeClass::Iterator avoidGotoLeftIterator(octreeIterator);
                // Now we can compute all the data
                // for each levels
                for(int idxLevel = 2 ; idxLevel < OctreeHeight ; ++idxLevel ){
                    if(!procHasWorkAtLevel(idxLevel, idProcess)){
                        avoidGotoLeftIterator.moveDown();
                        octreeIterator = avoidGotoLeftIterator;
                        continue;
                    }

                    int numberOfCells = 0;
                    while(octreeIterator.getCurrentGlobalIndex() <  getWorkingInterval(idxLevel , idProcess).leftIndex){
                        octreeIterator.moveRight();
                    }
                    // for each cells
                    do{
                        iterArray[numberOfCells] = octreeIterator;
                        ++numberOfCells;
                    } while(octreeIterator.moveRight());
                    avoidGotoLeftIterator.moveDown();
                    octreeIterator = avoidGotoLeftIterator;

                    FLOG(computationCounter.tic());
                    {
                        const int chunckSize = FMath::Max(1, numberOfCells/(omp_get_num_threads()*omp_get_num_threads()));
                        for(int idxCell = 0 ; idxCell < numberOfCells ; idxCell += chunckSize){
#pragma omp task
                            {
                                const int nbCellToCompute = FMath::Min(chunckSize, numberOfCells-idxCell);
                                for(int idxCellToCompute = idxCell ; idxCellToCompute < idxCell+nbCellToCompute ; ++idxCellToCompute){
                                    const int counter = tree->getInteractionNeighbors(neighbors,  iterArray[idxCellToCompute].getCurrentGlobalCoordinate(), idxLevel);
                                    if(counter) myThreadkernels->M2L( iterArray[idxCellToCompute].getCurrentCell() , neighbors, counter, idxLevel);
                                }
                            }
                        }
                    }

#pragma omp taskwait

                    for(int idxThread = 0 ; idxThread < omp_get_num_threads() ; ++idxThread){
#pragma omp task
                        {
                            kernels[idxThread]->finishedLevelM2L(idxLevel);
                        }
                    }

                    FLOG(computationCounter.tac());
                }
            }
        }


        {
            FLOG(receiveCounter.tic());
            typename OctreeClass::Iterator octreeIterator(tree);
            octreeIterator.moveDown();
            typename OctreeClass::Iterator avoidGotoLeftIterator(octreeIterator);
            // compute the second time
            // for each levels
            for(int idxLevel = 2 ; idxLevel < OctreeHeight ; ++idxLevel ){
                if(!procHasWorkAtLevel(idxLevel, idProcess)){
                    avoidGotoLeftIterator.moveDown();
                    octreeIterator = avoidGotoLeftIterator;
                    continue;
                }

                // put the received data into a temporary tree
                FLightOctree<CellClass> tempTree;
                for(int idxProc = 0 ; idxProc < nbProcess ; ++idxProc){
                    const int toReceiveFromProcAtLevel = globalReceiveMap[(idxProc * nbProcess * OctreeHeight) + idxLevel * nbProcess + idProcess];

                    for(int idxCell = 0 ; idxCell < toReceiveFromProcAtLevel ; ++idxCell){
                        const MortonIndex cellIndex = recvBuffer[idxLevel * nbProcess + idxProc]->FMpiBufferReader::getValue<MortonIndex>();

                        CellClass* const newCell = new CellClass;
                        newCell->setMortonIndex(cellIndex);
                        newCell->deserializeUp(*recvBuffer[idxLevel * nbProcess + idxProc]);

                        tempTree.insertCell(cellIndex, idxLevel, newCell);
                    }
                }


                // take cells from our octree only if they are
                // linked to received data
                int numberOfCells = 0;
                int realCellId = 0;

                while(octreeIterator.getCurrentGlobalIndex() <  getWorkingInterval(idxLevel , idProcess).leftIndex){
                    octreeIterator.moveRight();
                }
                // for each cells
                do{
                    // copy cells that need data from others
                    if(leafsNeedOther[idxLevel]->get(realCellId++)){
                        iterArray[numberOfCells++] = octreeIterator;
                    }
                } while(octreeIterator.moveRight());
                avoidGotoLeftIterator.moveDown();
                octreeIterator = avoidGotoLeftIterator;

                delete leafsNeedOther[idxLevel];
                leafsNeedOther[idxLevel] = nullptr;

                // Compute this cells
                FLOG(computationCounter.tic());
#pragma omp parallel
                {
                    KernelClass * const myThreadkernels = kernels[omp_get_thread_num()];
                    MortonIndex neighborsIndex[189];
                    int neighborsPosition[189];
                    const CellClass* neighbors[343];

#pragma omp for schedule(static) nowait
                    for(int idxCell = 0 ; idxCell < numberOfCells ; ++idxCell){
                        // compute indexes
                        memset(neighbors, 0, 343 * sizeof(CellClass*));
                        const int counterNeighbors = iterArray[idxCell].getCurrentGlobalCoordinate().getInteractionNeighbors(idxLevel, neighborsIndex, neighborsPosition);

                        int counter = 0;
                        // does we receive this index from someone?
                        for(int idxNeig = 0 ;idxNeig < counterNeighbors ; ++idxNeig){
                            if(neighborsIndex[idxNeig] < (getWorkingInterval(idxLevel , idProcess).leftIndex)
                                    || (getWorkingInterval(idxLevel , idProcess).rightIndex) < neighborsIndex[idxNeig]){

                                CellClass*const otherCell = tempTree.getCell(neighborsIndex[idxNeig], idxLevel);

                                if(otherCell){
                                    //otherCell->setMortonIndex(neighborsIndex[idxNeig]);
                                    neighbors[ neighborsPosition[idxNeig] ] = otherCell;
                                    ++counter;
                                }
                            }
                        }
                        // need to compute
                        if(counter){
                            myThreadkernels->M2L( iterArray[idxCell].getCurrentCell() , neighbors, counter, idxLevel);
                        }
                    }

                    myThreadkernels->finishedLevelM2L(idxLevel);
                }
                FLOG(computationCounter.tac());
            }
            FLOG(receiveCounter.tac());
        }

        for(int idxComm = 0 ; idxComm < nbProcess * OctreeHeight; ++idxComm){
            delete sendBuffer[idxComm];
            delete recvBuffer[idxComm];
        }
        for(int idxComm = 0 ; idxComm < OctreeHeight; ++idxComm){
            delete leafsNeedOther[idxComm];
        }
        delete[] sendBuffer;
        delete[] recvBuffer;
        delete[] indexToSend;
        delete[] leafsNeedOther;
        delete[] globalReceiveMap;


        FLOG( FLog::Controller << "\tFinished (@Downward Pass (M2L) = "  << counterTime.tacAndElapsed() << " s)\n" );
        FLOG( FLog::Controller << "\t\t Computation : " << computationCounter.cumulated() << " s\n" );
        FLOG( FLog::Controller << "\t\t Send : " << sendCounter.cumulated() << " s\n" );
        FLOG( FLog::Controller << "\t\t Receive : " << receiveCounter.cumulated() << " s\n" );
        FLOG( FLog::Controller << "\t\t Gather : " << gatherCounter.cumulated() << " s\n" );
        FLOG( FLog::Controller << "\t\t Prepare : " << prepareCounter.cumulated() << " s\n" );
862

863
    }
864

865 866 867 868
    //////////////////////////////////////////////////////////////////
    // ---------------- L2L ---------------
    //////////////////////////////////////////////////////////////////

869
    void downardPass(){ // second L2L
870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010
        const int MaxSizePerCell = CellClass::GetSize();
        FLOG( FLog::Controller.write("\tStart Downward Pass (L2L)\n").write(FLog::Flush); );
        FLOG(FTic counterTime);
        FLOG(FTic computationCounter);
        FLOG(FTic prepareCounter);
        FLOG(FTic waitCounter);

        // Start from leal level - 1
        typename OctreeClass::Iterator octreeIterator(tree);
        octreeIterator.moveDown();
        typename OctreeClass::Iterator avoidGotoLeftIterator(octreeIterator);

        // Max 1 receive and 7 send (but 7 times the same data)
        MPI_Request*const requests = new MPI_Request[8];
        MPI_Status*const status = new MPI_Status[8];

        const int heightMinusOne = OctreeHeight - 1;

        FMpiBufferWriter sendBuffer(comm.getComm(),MaxSizePerCell);
        FMpiBufferReader recvBuffer(comm.getComm(),MaxSizePerCell);

        int righestProcToSendTo   = nbProcess - 1;

        // for each levels exepted leaf level
        for(int idxLevel = 2 ; idxLevel < heightMinusOne ; ++idxLevel ){
            // If nothing to do in the next level skip the current one
            if(idProcess != 0 && !procHasWorkAtLevel(idxLevel+1, idProcess) ){
                avoidGotoLeftIterator.moveDown();
                octreeIterator = avoidGotoLeftIterator;
                continue;
            }

            // Copy all the cells in an array even the one that are out of my working interval
            int totalNbCellsAtLevel = 0;
            do{
                iterArray[totalNbCellsAtLevel++] = octreeIterator;
            } while(octreeIterator.moveRight());
            avoidGotoLeftIterator.moveDown();
            octreeIterator = avoidGotoLeftIterator;

            // Count the number of cells that are out of my working interval
            int nbCellsToSkip = 0;
            while(nbCellsToSkip < totalNbCellsAtLevel && iterArray[nbCellsToSkip].getCurrentGlobalIndex() < getWorkingInterval(idxLevel , idProcess).leftIndex){
                nbCellsToSkip += 1;
            }

            // Check if someone will send a cell to me
            bool hasToReceive = false;
            int idxProcToReceive = idProcess - 1;
            if(idProcess != 0 && nbCellsToSkip){
                // Starting from my left neighbor stop at the first proc that has work to do (not null interval)
                while(idxProcToReceive && !procHasWorkAtLevel(idxLevel, idxProcToReceive) ){
                    idxProcToReceive -= 1;
                }
                // Check if we find such a proc and that it share a cell with us on the border
                if(procHasWorkAtLevel(idxLevel, idxProcToReceive) && procCoversMyLeftBorderCell(idxLevel, idxProcToReceive)){
                    hasToReceive = true;
                }
            }

#pragma omp parallel
            {
                int threadNumber = omp_get_thread_num();
                KernelClass* myThreadkernels = (kernels[threadNumber]);
#pragma omp single nowait
                {
                    FLOG(prepareCounter.tic());
                    int iterRequests = 0;
                    // Post the receive
                    if(hasToReceive){
                        FMpi::MpiAssert( MPI_Irecv( recvBuffer.data(), recvBuffer.getCapacity(), MPI_PACKED, idxProcToReceive,
                                                    FMpi::TagFmmL2L + idxLevel, comm.getComm(), &requests[iterRequests++]), __LINE__ );
                    }

                    // We have to be sure that we are not sending if we have no work in the current level
                    if(idProcess != nbProcess - 1 && idProcess < righestProcToSendTo && procHasWorkAtLevel(idxLevel, idProcess)){
                        int idxProcSend = idProcess + 1;
                        int nbMessageSent = 0;
                        // From the proc on the right to righestProcToSendTo, check if we have to send something
                        while(idxProcSend <= righestProcToSendTo && ( !procHasWorkAtLevel(idxLevel+1, idxProcSend) || procCoversMyRightBorderCell(idxLevel, idxProcSend)) ){
                            // We know that if the proc has work at the next level it share a cell with us due to the while condition
                            if(procHasWorkAtLevel(idxLevel+1, idxProcSend)){
                                FAssertLF(procCoversMyRightBorderCell(idxLevel, idxProcSend));
                                // If first message then serialize the cell to send
                                if( nbMessageSent == 0 ){
                                    // We send our last cell
                                    iterArray[totalNbCellsAtLevel - 1].getCurrentCell()->serializeDown(sendBuffer);
                                }
                                // Post the send message
                                FMpi::MpiAssert( MPI_Isend(sendBuffer.data(), sendBuffer.getSize(), MPI_PACKED, idxProcSend,
                                                           FMpi::TagFmmL2L + idxLevel, comm.getComm(), &requests[iterRequests++]), __LINE__);
                                // Inc and check the counter
                                nbMessageSent += 1;
                                FAssertLF(nbMessageSent <= 7);
                            }
                            idxProcSend += 1;
                        }
                        // Next time we will not need to go further than idxProcSend
                        righestProcToSendTo = idxProcSend;
                    }
                    // Finalize the communication
                    if(iterRequests){
                        FLOG(waitCounter.tic());
                        FAssertLF(iterRequests <= 8);
                        FMpi::MpiAssert(MPI_Waitall( iterRequests, requests, status), __LINE__);
                        FLOG(waitCounter.tac());
                    }
                    // If we receive something proceed the L2L
                    if(hasToReceive){
                        FAssertLF(iterRequests != 0);
                        // In this case we know that we have to perform the L2L with the last cell that are
                        // exclude from our working interval nbCellsToSkip-1
                        iterArray[nbCellsToSkip-1].getCurrentCell()->deserializeDown(recvBuffer);
                        kernels[threadNumber]->L2L( iterArray[nbCellsToSkip-1].getCurrentCell() , iterArray[nbCellsToSkip-1].getCurrentChild(), idxLevel);
                    }
                    FLOG(prepareCounter.tac());
                }

#pragma omp single nowait
                {
                    FLOG(computationCounter.tic());
                }
                // Threads are working on all the cell of our working interval at that level
#pragma omp for nowait
                for(int idxCell = nbCellsToSkip ; idxCell < totalNbCellsAtLevel ; ++idxCell){
                    myThreadkernels->L2L( iterArray[idxCell].getCurrentCell() , iterArray[idxCell].getCurrentChild(), idxLevel);
                }
            }
            FLOG(computationCounter.tac());

            sendBuffer.reset();
            recvBuffer.seek(0);
        }

        delete[] requests;
        delete[] status;

        FLOG( FLog::Controller << "\tFinished (@Downward Pass (L2L) = "  << counterTime.tacAndElapsed() << " s)\n" );
        FLOG( FLog::Controller << "\t\t Computation : " << computationCounter.cumulated() << " s\n" );
        FLOG( FLog::Controller << "\t\t Prepare : " << prepareCounter.cumulated() << " s\n" );
        FLOG( FLog::Controller << "\t\t Wait : " << waitCounter.cumulated() << " s\n" );
1011 1012 1013 1014 1015 1016 1017
    }


    /////////////////////////////////////////////////////////////////////////////
    // Direct
    /////////////////////////////////////////////////////////////////////////////
    struct LeafData{
1018 1019 1020 1021
        FTreeCoordinate coord;
        CellClass* cell;
        ContainerClass* targets;
        ContainerClass* sources;
1022 1023 1024 1025 1026
    };


    /** P2P */
    void directPass(){
1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
        FLOG( FLog::Controller.write("\tStart Direct Pass\n").write(FLog::Flush); );
        FLOG( FTic counterTime);
        FLOG( FTic prepareCounter);
        FLOG( FTic gatherCounter);
        FLOG( FTic waitCounter);
        FLOG(FTic computationCounter);

        ///////////////////////////////////////////////////
        // Prepare data to send receive
        ///////////////////////////////////////////////////
        FLOG(prepareCounter.tic());

        // To send in asynchrone way
        MPI_Request requests[2 * nbProcess];
        MPI_Status status[2 * nbProcess];
        int iterRequest = 0;
        int nbMessagesToRecv = 0;

        FMpiBufferWriter**const sendBuffer = new FMpiBufferWriter*[nbProcess];
        memset(sendBuffer, 0, sizeof(FMpiBufferWriter*) * nbProcess);

        FMpiBufferReader**const recvBuffer = new FMpiBufferReader*[nbProcess];
        memset(recvBuffer, 0, sizeof(FMpiBufferReader*) * nbProcess);

        /* This a nbProcess x nbProcess matrix of integer
         * let U and V be id of processes :
         * globalReceiveMap[U*nbProcess + V] == size of information needed by V and own by U
         */
        int*const globalReceiveMap = new int[nbProcess * nbProcess];
        memset(globalReceiveMap, 0, sizeof(int) * nbProcess * nbProcess);

        FBoolArray leafsNeedOther(this->numberOfLeafs);
        int countNeedOther = 0;

        // To store the result
        OctreeClass otherP2Ptree( tree->getHeight(), tree->getSubHeight(), tree->getBoxWidth(), tree->getBoxCenter() );

        // init
        const int LeafIndex = OctreeHeight - 1;
        const int SizeShape = 3*3*3;

        int shapeLeaf[SizeShape];
        memset(shapeLeaf,0,SizeShape*sizeof(int));

        LeafData* const leafsDataArray = new LeafData[this->numberOfLeafs];

        FVector<LeafData> leafsNeedOtherData(countNeedOther);

        FVector<typename OctreeClass::Iterator>*const toSend = new FVector<typename OctreeClass::Iterator>[nbProcess];
        int partsToSend[nbProcess];
        memset(partsToSend, 0, sizeof(int) * nbProcess);

        #pragma omp parallel
        {
            #pragma omp master // MUST WAIT to fill leafsNeedOther
            {
                // Copy leafs
                {
                    typename OctreeClass::Iterator octreeIterator(tree);
                    octreeIterator.gotoBottomLeft();
                    int idxLeaf = 0;
                    do{
                        this->iterArray[idxLeaf++] = octreeIterator;
                    } while(octreeIterator.moveRight());
                }

                int alreadySent[nbProcess];

                //Will store the indexes of the neighbors of current cell
                MortonIndex indexesNeighbors[26];

                for(int idxLeaf = 0 ; idxLeaf < this->numberOfLeafs ; ++idxLeaf){
                    memset(alreadySent, 0, sizeof(int) * nbProcess);
                    bool needOther = false;
                    //Get the neighbors of current cell in indexesNeighbors, and their number in neighCount
                    const int neighCount = (iterArray[idxLeaf].getCurrentGlobalCoordinate()).getNeighborsIndexes(OctreeHeight,indexesNeighbors);
                    //Loop over the neighbor leafs
                    for(int idxNeigh = 0 ; idxNeigh < neighCount ; ++idxNeigh){
                        //Test if leaf belongs to someone else (false if it's mine)
                        if(indexesNeighbors[idxNeigh] < (intervals[idProcess].leftIndex) || (intervals[idProcess].rightIndex) < indexesNeighbors[idxNeigh]){
                            needOther = true;

                            // find the proc that will need current leaf
                            int procToReceive = idProcess;
                            while( procToReceive != 0 && indexesNeighbors[idxNeigh] < intervals[procToReceive].leftIndex){
                                --procToReceive; //scroll process "before" current process
                            }

                            while( procToReceive != nbProcess - 1 && (intervals[procToReceive].rightIndex) < indexesNeighbors[idxNeigh]){
                                ++procToReceive;//scroll process "after" current process
                            }
                            //  Test : Not Already Send && be sure someone hold this interval
                            if( !alreadySent[procToReceive] && intervals[procToReceive].leftIndex <= indexesNeighbors[idxNeigh] && indexesNeighbors[idxNeigh] <= intervals[procToReceive].rightIndex){

                                alreadySent[procToReceive] = 1;
                                toSend[procToReceive].push( iterArray[idxLeaf] );
                                partsToSend[procToReceive] += iterArray[idxLeaf].getCurrentListSrc()->getSavedSize();
                                partsToSend[procToReceive] += int(sizeof(MortonIndex));
                            }
                        }
                    }

                    if(needOther){ //means that something need to be sent (or received)
                        leafsNeedOther.set(idxLeaf,true);
                        ++countNeedOther;
                    }
                }

                // No idea why it is mandatory there, could it be a few line before,
                for(int idxProc = 0 ; idxProc < nbProcess ; ++idxProc){
                    if(partsToSend[idxProc]){
                        partsToSend[idxProc] += int(sizeof(int));
                    }
                }
            }

            #pragma omp barrier

            #pragma omp master // nowait
            {
                //Share to all processus globalReceiveMap
                FLOG(gatherCounter.tic());
                FMpi::MpiAssert( MPI_Allgather( partsToSend, nbProcess, MPI_INT, globalReceiveMap, nbProcess, MPI_INT, comm.getComm()),  __LINE__ );
                FLOG(gatherCounter.tac());

                //Prepare receive
                for(int idxProc = 0 ; idxProc < nbProcess ; ++idxProc){
                    if(globalReceiveMap[idxProc * nbProcess + idProcess]){ //if idxProc has sth for me.
                        //allocate buffer of right size
                        recvBuffer[idxProc] = new FMpiBufferReader(comm.getComm(),globalReceiveMap[idxProc * nbProcess + idProcess]);
                        FMpi::MpiAssert( MPI_Irecv(recvBuffer[idxProc]->data(), recvBuffer[idxProc]->getCapacity(), MPI_PACKED,
                                                   idxProc, FMpi::TagFmmP2P, comm.getComm(), &requests[iterRequest++]) , __LINE__ );
                    }
                }

                nbMessagesToRecv = iterRequest;
                // Prepare send
                for(int idxProc = 0 ; idxProc < nbProcess ; ++idxProc){
                    if(toSend[idxProc].getSize() != 0){
                        sendBuffer[idxProc] = new FMpiBufferWriter(comm.getComm(),globalReceiveMap[idProcess*nbProcess+idxProc]);
                        // << is equivalent to write().
                        (*sendBuffer[idxProc]) << toSend[idxProc].getSize();
                        for(int idxLeaf = 0 ; idxLeaf < toSend[idxProc].getSize() ; ++idxLeaf){
                            (*sendBuffer[idxProc]) << toSend[idxProc][idxLeaf].getCurrentGlobalIndex();
                            toSend[idxProc][idxLeaf].getCurrentListSrc()->save(*sendBuffer[idxProc]);
                        }

                        FMpi::MpiAssert( MPI_Isend( sendBuffer[idxProc]->data(), sendBuffer[idxProc]->getSize() , MPI_PACKED ,
                                                    idxProc, FMpi::TagFmmP2P, comm.getComm(), &requests[iterRequest++]) , __LINE__ );

                    }
                }

                delete[] toSend;


                //////////////////////////////////////////////////////////
                // Waitsend receive
                //////////////////////////////////////////////////////////

                int complete = 0;
                int*const indexMessage = new int[nbProcess * 2];
                while( complete != iterRequest){
                    memset(indexMessage, 0, sizeof(int) * nbProcess * 2);
                    int countMessages = 0;
                    // Wait data
                    FLOG(waitCounter.tic());
                    MPI_Waitsome(iterRequest, requests, &countMessages, indexMessage, status);

                    FLOG(waitCounter.tac());
                    complete += countMessages;


                    for(int idxRcv = 0 ; idxRcv < countMessages ; ++idxRcv){
                        if( indexMessage[idxRcv] < nbMessagesToRecv ){
                            const int idxProc = status[idxRcv].MPI_SOURCE;
                            int nbLeaves;
                            (*recvBuffer[idxProc]) >> nbLeaves;
                            for(int idxLeaf = 0 ; idxLeaf < nbLeaves ; ++idxLeaf){
                                MortonIndex leafIndex;
                                (*recvBuffer[idxProc]) >> leafIndex;
                                otherP2Ptree.createLeaf(leafIndex)->getSrc()->restore((*recvBuffer[idxProc]));
                            }
                            delete recvBuffer[idxProc];
                            recvBuffer[idxProc] = nullptr;
                        }
                    }
                }
                delete[] indexMessage;
            }

            ///////////////////////////////////////////////////
            // Prepare data for thread P2P
            ///////////////////////////////////////////////////

            #pragma omp single // MUST WAIT!
            {
                typename OctreeClass::Iterator octreeIterator(tree);
                octreeIterator.gotoBottomLeft();

                // to store which shape for each leaf
                typename OctreeClass::Iterator* const myLeafs = new typename OctreeClass::Iterator[this->numberOfLeafs];
                int*const shapeType = new int[this->numberOfLeafs];

                for(int idxLeaf = 0 ; idxLeaf < this->numberOfLeafs ; ++idxLeaf){
                    myLeafs[idxLeaf] = octreeIterator;

                    const FTreeCoordinate& coord = octreeIterator.getCurrentCell()->getCoordinate();
                    const int shape = (coord.getX()%3)*9 + (coord.getY()%3)*3 + (coord.getZ()%3);
                    shapeType[idxLeaf] = shape;

                    ++shapeLeaf[shape];

                    octreeIterator.moveRight();
                }

                int startPosAtShape[SizeShape];
                startPosAtShape[0] = 0;
                for(int idxShape = 1 ; idxShape < SizeShape ; ++idxShape){
                    startPosAtShape[idxShape] = startPosAtShape[idxShape-1] + shapeLeaf[idxShape-1];
                }

                int idxInArray = 0;
                for(int idxLeaf = 0 ; idxLeaf < this->numberOfLeafs ; ++idxLeaf, ++idxInArray){
                    const int shapePosition = shapeType[idxInArray];

                    leafsDataArray[startPosAtShape[shapePosition]].coord = myLeafs[idxInArray].getCurrentGlobalCoordinate();
                    leafsDataArray[startPosAtShape[shapePosition]].cell = myLeafs[idxInArray].getCurrentCell();
                    leafsDataArray[startPosAtShape[shapePosition]].targets = myLeafs[idxInArray].getCurrentListTargets();
                    leafsDataArray[startPosAtShape[shapePosition]].sources = myLeafs[idxInArray].getCurrentListSrc();
                    if( leafsNeedOther.get(idxLeaf) ) leafsNeedOtherData.push(leafsDataArray[startPosAtShape[shapePosition]]);

                    ++startPosAtShape[shapePosition];
                }

                delete[] shapeType;
                delete[] myLeafs;

                FLOG(prepareCounter.tac());
            }


            //////////////////////////////////////////////////////////
            // Computation P2P that DO NOT need others data
            //////////////////////////////////////////////////////////

            {
                KernelClass* myThreadkernels = (kernels[omp_get_thread_num()]);

                #pragma omp single nowait
                {
                    FLOG(computationCounter.tic());
                    int previous = 0;

                    for(int idxShape = 0 ; idxShape < SizeShape ; ++idxShape){
                        const int endAtThisShape = shapeLeaf[idxShape] + previous;
                        const int chunckSize = FMath::Max(1, (endAtThisShape-previous)/(omp_get_num_threads()*omp_get_num_threads()));

                        for(int idxLeafs = previous ; idxLeafs < endAtThisShape ; idxLeafs += chunckSize){
                            const int nbLeavesInTask = FMath::Min(endAtThisShape-idxLeafs, chunckSize);
                            #pragma omp task
                            {
                                // There is a maximum of 26 neighbors
                                ContainerClass* neighbors[27];

                                for(int idxTaskLeaf = idxLeafs ; idxTaskLeaf < (idxLeafs + nbLeavesInTask) ; ++idxTaskLeaf){
                                    LeafData& currentIter = leafsDataArray[idxTaskLeaf];
                                    myThreadkernels->L2P(currentIter.cell, currentIter.targets);

                                    // need the current particles and neighbors particles
                                    const int counter = tree->getLeafsNeighbors(neighbors, currentIter.coord, LeafIndex);
                                    myThreadkernels->P2P( currentIter.coord,currentIter.targets,
                                                         currentIter.sources, neighbors, counter);
                                }
                            }
                        }
                        previous = endAtThisShape;

                        #pragma omp taskwait
                    }
                    FLOG(computationCounter.tac());
                }
            }

            // Wait the come to finish (and the previous computation also)
            #pragma omp barrier


            //////////////////////////////////////////////////////////
            // Computation P2P that need others data
            //////////////////////////////////////////////////////////
            #pragma omp master
            { FLOG( computation2Counter.tic() ); }

            {
                KernelClass& myThreadkernels = (*kernels[omp_get_thread_num()]);
                // There is a maximum of 26 neighbors
                ContainerClass* neighbors[27];
                MortonIndex indexesNeighbors[27];
                int indexArray[26];
                // Box limite
                const int nbLeafToProceed = leafsNeedOtherData.getSize();

                #pragma omp for schedule(static)
                for(int idxLeafs = 0 ; idxLeafs < nbLeafToProceed ; ++idxLeafs){
                    LeafData currentIter = leafsNeedOtherData[idxLeafs];

                    // need the current particles and neighbors particles
                    int counter = 0;
                    memset( neighbors, 0, sizeof(ContainerClass*) * 27);

                    // Take possible data
                    const int nbNeigh = currentIter.coord.getNeighborsIndexes(OctreeHeight, indexesNeighbors, indexArray);

                    for(int idxNeigh = 0 ; idxNeigh < nbNeigh ; ++idxNeigh){
                        if(indexesNeighbors[idxNeigh] < (intervals[idProcess].leftIndex) || (intervals[idProcess].rightIndex) < indexesNeighbors[idxNeigh]){
                            ContainerClass*const hypotheticNeighbor = otherP2Ptree.getLeafSrc(indexesNeighbors[idxNeigh]);
                            if(hypotheticNeighbor){
                                neighbors[ indexArray[idxNeigh] ] = hypotheticNeighbor;
                                ++counter;
                            }
                        }
                    }
                    myThreadkernels.P2PRemote( currentIter.cell->getCoordinate(), currentIter.targets,
                                               currentIter.sources, neighbors, counter);

                }

            }
        }

        for(int idxProc = 0 ; idxProc < nbProcess ; ++idxProc){
            delete sendBuffer[idxProc];
            delete recvBuffer[idxProc];
        }
        delete[] globalReceiveMap;
        delete[] leafsDataArray;

        FLOG(computation2Counter.tac());


        FLOG( FLog::Controller << "\tFinished (@Direct Pass (L2P + P2P) = "  << counterTime.tacAndElapsed() << " s)\n" );
        FLOG( FLog::Controller << "\t\t Computation L2P + P2P : " << computationCounter.elapsed() << " s\n" );
        FLOG( FLog::Controller << "\t\t Computation P2P 2 : " << computation2Counter.elapsed() << " s\n" );
        FLOG( FLog::Controller << "\t\t Prepare P2P : " << prepareCounter.elapsed() << " s\n" );
        FLOG( FLog::Controller << "\t\t Gather P2P : " << gatherCounter.elapsed() << " s\n" );
        FLOG( FLog::Controller << "\t\t Wait : " << waitCounter.elapsed() << " s\n" );
1374

1375
    }
1376 1377 1378 1379 1380 1381 1382 1383
};






#endif //FFMMALGORITHMTHREAD_HPP