/* -*- Mode: C; c-basic-offset:4 ; indent-tabs-mode:nil ; -*- */

/*

 *

 *  (C) 2003 by Argonne National Laboratory.

 *      See COPYRIGHT in top-level directory.

 */

#include "mpi.h"

#include <stdio.h>

#include <stdlib.h>

#include "mpitest.h"

/*

static char MTEST_Descrip[] = "Test MPI_Reduce with non-commutative user-define operations and arbitrary root";

*/

/*

 * This tests that the reduce operation respects the noncommutative flag.

 * and that can distinguish between P_{root} P_{root+1}

 * ... P_{root-1} and P_0 ... P_{size-1} .  The MPI standard clearly

 * specifies that the result is P_0 ... P_{size-1}, independent of the root

 * (see 4.9.4 in MPI-1)

 */

/* This implements a simple matrix-matrix multiply.  This is an associative

   but not commutative operation.  The matrix size is set in matSize;

   the number of matrices is the count argument. The matrix is stored

   in C order, so that

     c(i,j) is cin[j+i*matSize]

 */

#define MAXCOL 256

static int matSize = 0;         /* Must be < MAXCOL */

void uop(void *cinPtr, void *coutPtr, int *count, MPI_Datatype * dtype);

void uop(void *cinPtr, void *coutPtr, int *count, MPI_Datatype * dtype)

{

    const int *cin;

    int *cout;

    int i, j, k, nmat;

    int tempCol[MAXCOL];

    if (*count != 1)

        printf("Panic!\n");

    for (nmat = 0; nmat < *count; nmat++) {

        cin = (const int *) cinPtr;

        cout = (int *) coutPtr;

        for (j = 0; j < matSize; j++) {

            for (i = 0; i < matSize; i++) {

                tempCol[i] = 0;

                for (k = 0; k < matSize; k++) {

                    /* col[i] += cin(i,k) * cout(k,j) */

                    tempCol[i] += cin[k + i * matSize] * cout[j + k * matSize];

                }

            }

            for (i = 0; i < matSize; i++) {

                cout[j + i * matSize] = tempCol[i];

            }

        }

        cinPtr = (int *) cinPtr + matSize * matSize;

        coutPtr = (int *) coutPtr + matSize * matSize;

    }

}

/* Initialize the integer matrix as a permutation of rank with rank+1.

   If we call this matrix P_r, we know that product of P_0 P_1 ... P_{size-1}

   is the matrix with rows ordered as

   1,size,2,3,4,...,size-1

   (The matrix is basically a circular shift right,

   shifting right n-1 steps for an n x n dimensional matrix, with the last

   step swapping rows 1 and size)

*/

static void initMat(MPI_Comm comm, int mat[])

{

    int i, size, rank;

    MPI_Comm_rank(comm, &rank;;

    MPI_Comm_size(comm, &size);

    /* Remember the matrix size */

    matSize = size;

    for (i = 0; i < matSize * matSize; i++)

        mat[i] = 0;

    for (i = 0; i < matSize; i++) {

        if (i == rank)

            mat[((i + 1) % matSize) + i * matSize] = 1;

        else if (i == ((rank + 1) % matSize))

            mat[((i + matSize - 1) % matSize) + i * matSize] = 1;

        else

            mat[i + i * matSize] = 1;

    }

}

/* Compare a matrix with the identity matrix */

/*

static int isIdentity(MPI_Comm comm, int mat[])

{

    int i, j, size, rank, errs = 0;

    MPI_Comm_rank(comm, &rank;;

    MPI_Comm_size(comm, &size);

    for (i=0; i

	for (j=0; j

	    if (j == i) {

		if (mat[j+i*size] != 1) {

		    printf("mat(%d,%d) = %d, should = 1\n",

			    i, j, mat[j+i*size]);

		    errs++;

		}

	    }

	    else {

		if (mat[j+i*size] != 0) {

		    printf("mat(%d,%d) = %d, should = 0\n",

			    i, j, mat[j+i*size]);

		    errs++;

		}

	    }

	}

    }

    return errs;

}

*/

/* Compare a matrix with the identity matrix with rows permuted to as rows

   1,size,2,3,4,5,...,size-1 */

static int isPermutedIdentity(MPI_Comm comm, int mat[])

{

    int i, j, size, rank, errs = 0;

    MPI_Comm_rank(comm, &rank;;

    MPI_Comm_size(comm, &size);

    /* Check the first two last rows */

    i = 0;

    for (j = 0; j < size; j++) {

        if (j == 0) {

            if (mat[j] != 1) {

                printf("mat(%d,%d) = %d, should = 1\n", i, j, mat[j]);

                errs++;

            }

        } else {

            if (mat[j] != 0) {

                printf("mat(%d,%d) = %d, should = 0\n", i, j, mat[j]);

                errs++;

            }

        }

    }

    i = 1;

    for (j = 0; j < size; j++) {

        if (j == size - 1) {

            if (mat[j + i * size] != 1) {

                printf("mat(%d,%d) = %d, should = 1\n", i, j, mat[j + i * size]);

                errs++;

            }

        } else {

            if (mat[j + i * size] != 0) {

                printf("mat(%d,%d) = %d, should = 0\n", i, j, mat[j + i * size]);

                errs++;

            }

        }

    }

    /* The remaint rows are shifted down by one */

    for (i = 2; i < size; i++) {

        for (j = 0; j < size; j++) {

            if (j == i - 1) {

                if (mat[j + i * size] != 1) {

                    printf("mat(%d,%d) = %d, should = 1\n", i, j, mat[j + i * size]);

                    errs++;

                }

            } else {

                if (mat[j + i * size] != 0) {

                    printf("mat(%d,%d) = %d, should = 0\n", i, j, mat[j + i * size]);

                    errs++;

                }

            }

        }

    }

    return errs;

}

int main(int argc, char *argv[])

{

    int errs = 0;

    int rank, size, root;

    int minsize = 2, count;

    MPI_Comm comm;

    int *buf, *bufout;

    MPI_Op op;

    MPI_Datatype mattype;

    MTest_Init(&argc, &argv);

    MPI_Op_create(uop, 0, &op);

    while (MTestGetIntracommGeneral(&comm, minsize, 1)) {

        if (comm == MPI_COMM_NULL)

            continue;

        MPI_Comm_size(comm, &size);

        MPI_Comm_rank(comm, &rank;;

        if (size > MAXCOL) {

            /* Skip because there are too many processes */

            MTestFreeComm(&comm);

            continue;

        }

        /* Only one matrix for now */

        count = 1;

        /* A single matrix, the size of the communicator */

        MPI_Type_contiguous(size * size, MPI_INT, &mattype);

        MPI_Type_commit(&mattype);

        buf = (int *) malloc(count * size * size * sizeof(int));

        if (!buf)

            MPI_Abort(MPI_COMM_WORLD, 1);

        bufout = (int *) malloc(count * size * size * sizeof(int));

        if (!bufout)

            MPI_Abort(MPI_COMM_WORLD, 1);

        for (root = 0; root < size; root++) {

            initMat(comm, buf);

            MPI_Reduce(buf, bufout, count, mattype, op, root, comm);

            if (rank == root) {

                errs += isPermutedIdentity(comm, bufout);

            }

            /* Try the same test, but using MPI_IN_PLACE */

            initMat(comm, bufout);

            if (rank == root) {

                MPI_Reduce(MPI_IN_PLACE, bufout, count, mattype, op, root, comm);

            } else {

                MPI_Reduce(bufout, NULL, count, mattype, op, root, comm);

            }

            if (rank == root) {

                errs += isPermutedIdentity(comm, bufout);

            }

        }

        MPI_Type_free(&mattype);

        free(buf);

        free(bufout);

        MTestFreeComm(&comm);

    }

    MPI_Op_free(&op);

    MTest_Finalize(errs);

    return MTestReturnValue(errs);

}