test/datatype/to_self.c

/* -*- Mode: C; c-basic-offset:4 ; -*- */
/*
 * Copyright (c) 2004-2019 The University of Tennessee and The University
 *                         of Tennessee Research Foundation.  All rights
 *                         reserved.
 * $COPYRIGHT$
 *
 * Additional copyrights may follow
 *
 * $HEADER$
 */

#include "mpi.h"

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <math.h>

#if 0 && OPEN_MPI
extern void ompi_datatype_dump( MPI_Datatype ddt );
#define MPI_DDT_DUMP(ddt) ompi_datatype_dump( (ddt) )
#else
#define MPI_DDT_DUMP(ddt)
#endif  /* OPEN_MPI */

static MPI_Datatype
create_merged_contig_with_gaps(int count)  /* count of the basic datatype */
{
    int array_of_blocklengths[] = {1, 1, 1};
    MPI_Aint array_of_displacements[] = {0, 8, 16};
    MPI_Datatype array_of_types[] = {MPI_DOUBLE, MPI_LONG, MPI_CHAR};
    MPI_Datatype type;

    MPI_Type_create_struct(3, array_of_blocklengths,
                           array_of_displacements, array_of_types,
                           &type);
    if( 1 < count ) {
        MPI_Datatype temp = type;
        MPI_Type_contiguous(count, temp, &type);
    }
    MPI_Type_commit(&type);
    MPI_DDT_DUMP( type );
    return type;
}

/* Create a non-contiguous resized datatype */
struct structure {
    double not_transfered;
    double transfered_1;
    double transfered_2;
};

static MPI_Datatype
create_struct_constant_gap_resized_ddt( int number,  /* IGNORED: number of repetitions */
                                        int contig_size,  /* IGNORED: number of elements in a contiguous chunk */
                                        int gap_size )    /* IGNORED: number of elements in a gap */
{
    struct structure data[1];
    MPI_Datatype struct_type, temp_type;
    MPI_Datatype types[2] = {MPI_DOUBLE, MPI_DOUBLE};
    int blocklens[2] = {1, 1};
    MPI_Aint disps[3];

    MPI_Get_address(&data[0].transfered_1, &disps[0]);
    MPI_Get_address(&data[0].transfered_2, &disps[1]);
    MPI_Get_address(&data[0], &disps[2]);
    disps[1] -= disps[2]; /*  8 */
    disps[0] -= disps[2]; /* 16 */

    MPI_Type_create_struct(2, blocklens, disps, types, &temp_type);
    MPI_Type_create_resized(temp_type, 0, sizeof(data[0]), &struct_type);
    MPI_Type_commit(&struct_type);
    MPI_Type_free(&temp_type);
    MPI_DDT_DUMP( struct_type );

    return struct_type;
}

/* Create a datatype similar to the one use by HPL */
static MPI_Datatype
create_indexed_constant_gap_ddt( int number,  /* number of repetitions */
                                 int contig_size,  /* number of elements in a contiguous chunk */
                                 int gap_size )    /* number of elements in a gap */
{
    MPI_Datatype dt, *types;
    int i, *bLength;
    MPI_Aint* displ;

    types = (MPI_Datatype*)malloc( sizeof(MPI_Datatype) * number );
    bLength = (int*)malloc( sizeof(int) * number );
    displ = (MPI_Aint*)malloc( sizeof(MPI_Aint) * number );

    types[0] = MPI_DOUBLE;
    bLength[0] = contig_size;
    displ[0] = 0;
    for( i = 1; i < number; i++ ) {
        types[i] = MPI_DOUBLE;
        bLength[i] = contig_size;
        displ[i] = displ[i-1] + sizeof(double) * (contig_size + gap_size);
    }
    MPI_Type_create_struct( number, bLength, displ, types, &dt );
    MPI_DDT_DUMP( dt );
    free(types);
    free(bLength);
    free(displ);
    MPI_Type_commit( &dt );
    return dt;
}

static MPI_Datatype
create_optimized_indexed_constant_gap_ddt( int number,  /* number of repetitions */
                                           int contig_size,  /* number of elements in a contiguous chunk */
                                           int gap_size )    /* number of elements in a gap */
{
    MPI_Datatype dt;

    MPI_Type_vector( number, contig_size, (contig_size + gap_size), MPI_DOUBLE, &dt );
    MPI_Type_commit( &dt );
    MPI_DDT_DUMP( dt );
    return dt;
}

typedef struct {
   int i[2];
   float f;
} internal_struct;
typedef struct {
   int v1;
   int gap1;
   internal_struct is[3];
} ddt_gap;

static MPI_Datatype
create_indexed_gap_ddt( void )
{
    ddt_gap dt[2];
    MPI_Datatype dt1, dt2, dt3;
    int bLength[2] = { 2, 1 };
    MPI_Datatype types[2] = { MPI_INT, MPI_FLOAT };
    MPI_Aint displ[2];

    MPI_Get_address( &(dt[0].is[0].i[0]), &(displ[0]) );
    MPI_Get_address( &(dt[0].is[0].f), &(displ[1]) );
    displ[1] -= displ[0];
    displ[0] -= displ[0];
    MPI_Type_create_struct( 2, bLength, displ, types, &dt1 );
    /*MPI_DDT_DUMP( dt1 );*/
    MPI_Type_contiguous( 3, dt1, &dt2 );
    /*MPI_DDT_DUMP( dt2 );*/
    bLength[0] = 1;
    bLength[1] = 1;
    MPI_Get_address( &(dt[0].v1), &(displ[0]) );
    MPI_Get_address( &(dt[0].is[0]), &(displ[1]) );
    displ[1] -= displ[0];
    displ[0] -= displ[0];
    types[0] = MPI_INT;
    types[1] = dt2;
    MPI_Type_create_struct( 2, bLength, displ, types, &dt3 );
    /*MPI_DDT_DUMP( dt3 );*/
    MPI_Type_free( &dt1 );
    MPI_Type_free( &dt2 );
    MPI_Type_contiguous( 10, dt3, &dt1 );
    MPI_DDT_DUMP( dt1 );
    MPI_Type_free( &dt3 );
    MPI_Type_commit( &dt1 );
    return dt1;
}

static MPI_Datatype
create_indexed_gap_optimized_ddt( void )
{
    MPI_Datatype dt1, dt2, dt3;
    int bLength[3];
    MPI_Datatype types[3];
    MPI_Aint displ[3];

    MPI_Type_contiguous( 40, MPI_BYTE, &dt1 );
    MPI_Type_create_resized( dt1, 0, 44, &dt2 );

    bLength[0] = 4;
    bLength[1] = 9;
    bLength[2] = 36;

    types[0] = MPI_BYTE;
    types[1] = dt2;
    types[2] = MPI_BYTE;

    displ[0] = 0;
    displ[1] = 8;
    displ[2] = 44 * 9 + 8;

    MPI_Type_create_struct( 3, bLength, displ, types, &dt3 );

    MPI_Type_free( &dt1 );
    MPI_Type_free( &dt2 );
    MPI_DDT_DUMP( dt3 );
    MPI_Type_commit( &dt3 );
    return dt3;
}


/********************************************************************
 *******************************************************************/

#define DO_CONTIG                         0x00000001
#define DO_CONSTANT_GAP                   0x00000002
#define DO_INDEXED_GAP                    0x00000004
#define DO_OPTIMIZED_INDEXED_GAP          0x00000008
#define DO_STRUCT_CONSTANT_GAP_RESIZED    0x00000010
#define DO_STRUCT_MERGED_WITH_GAP_RESIZED 0x00000020

#define DO_PACK                         0x01000000
#define DO_UNPACK                       0x02000000
#define DO_ISEND_RECV                   0x04000000
#define DO_ISEND_IRECV                  0x08000000
#define DO_IRECV_SEND                   0x10000000
#define DO_IRECV_ISEND                  0x20000000

#define MIN_LENGTH   1024
#define MAX_LENGTH   (1024*1024)

static int cycles  = 100;
static int trials  = 20;
static int warmups = 2;

static void print_result( int length, int trials, double* timers )
{
    double bandwidth, clock_prec, temp;
    double min_time, max_time, average, std_dev = 0.0;
    double ordered[trials];
    int t, pos, quartile_start, quartile_end;

    for( t = 0; t < trials; ordered[t] = timers[t], t++ );
    for( t = 0; t < trials-1; t++ ) {
        temp = ordered[t];
        pos = t;
        for( int i = t+1; i < trials; i++ ) {
            if( temp > ordered[i] ) {
                temp = ordered[i];
                pos = i;
            }
        }
        if( pos != t ) {
            temp = ordered[t];
            ordered[t] = ordered[pos];
            ordered[pos] = temp;
        }
    }
    quartile_start = trials - (3 * trials) / 4;
    quartile_end   = trials - (1 * trials) / 4;
    clock_prec = MPI_Wtick();
    min_time = ordered[quartile_start];
    max_time = ordered[quartile_start];
    average = ordered[quartile_start];
    for( t = quartile_start + 1; t < quartile_end; t++ ) {
        if( min_time > ordered[t] ) min_time = ordered[t];
        if( max_time < ordered[t] ) max_time = ordered[t];
        average += ordered[t];
    }
    average /= (quartile_end - quartile_start);
    for( t = quartile_start; t < quartile_end; t++ ) {
        std_dev += (ordered[t] - average) * (ordered[t] - average);
    }
    std_dev = sqrt( std_dev/(quartile_end - quartile_start) );

    bandwidth = (length * clock_prec) / (1024.0 * 1024.0) / (average * clock_prec);
    printf( "%8d\t%15g\t%10.4f MB/s [min %10g max %10g std %2.2f%%]\n", length, average, bandwidth,
            min_time, max_time, (100.0 * std_dev) / average );
}

static int pack( int cycles,
                 MPI_Datatype sdt, int scount, void* sbuf,
                 void* packed_buf )
{
    int position, myself, c, t, outsize;
    double timers[trials];

    MPI_Type_size( sdt, &outsize );
    outsize *= scount;

    MPI_Comm_rank( MPI_COMM_WORLD, &myself );

    for( t = 0; t < warmups; t++ ) {
        for( c = 0; c < cycles; c++ ) {
            position = 0;
            MPI_Pack(sbuf, scount, sdt, packed_buf, outsize, &position, MPI_COMM_WORLD);
        }
    }

    for( t = 0; t < trials; t++ ) {
        timers[t] = MPI_Wtime();
        for( c = 0; c < cycles; c++ ) {
            position = 0;
            MPI_Pack(sbuf, scount, sdt, packed_buf, outsize, &position, MPI_COMM_WORLD);
        }
        timers[t] = (MPI_Wtime() - timers[t]) / cycles;
    }
    print_result( outsize, trials, timers );
    return 0;
}

static int unpack( int cycles,
                   void* packed_buf,
                   MPI_Datatype rdt, int rcount, void* rbuf )
{
    int position, myself, c, t, insize;
    double timers[trials];

    MPI_Type_size( rdt, &insize );
    insize *= rcount;

    MPI_Comm_rank( MPI_COMM_WORLD, &myself );

    for( t = 0; t < warmups; t++ ) {
        for( c = 0; c < cycles; c++ ) {
            position = 0;
            MPI_Unpack(packed_buf, insize, &position, rbuf, rcount, rdt, MPI_COMM_WORLD);
        }
    }

    for( t = 0; t < trials; t++ ) {
        timers[t] = MPI_Wtime();
        for( c = 0; c < cycles; c++ ) {
            position = 0;
            MPI_Unpack(packed_buf, insize, &position, rbuf, rcount, rdt, MPI_COMM_WORLD);
        }
        timers[t] = (MPI_Wtime() - timers[t]) / cycles;
    }
    print_result( insize, trials, timers );
    return 0;
}

static int isend_recv( int cycles,
                       MPI_Datatype sdt, int scount, void* sbuf,
                       MPI_Datatype rdt, int rcount, void* rbuf )
{
    int myself, tag = 0, c, t, slength, rlength;
    MPI_Status status;
    MPI_Request req;
    double timers[trials];

    MPI_Type_size( sdt, &slength );
    slength *= scount;
    MPI_Type_size( rdt, &rlength );
    rlength *= rcount;

    MPI_Comm_rank( MPI_COMM_WORLD, &myself );

    for( t = 0; t < trials; t++ ) {
        timers[t] = MPI_Wtime();
        for( c = 0; c < cycles; c++ ) {
            MPI_Isend( sbuf, scount, sdt, myself, tag, MPI_COMM_WORLD, &req );
            MPI_Recv( rbuf, rcount, rdt, myself, tag, MPI_COMM_WORLD, &status );
            MPI_Wait( &req, &status );
        }
        timers[t] = (MPI_Wtime() - timers[t]) / cycles;
    }
    print_result( rlength, trials, timers );
    return 0;
}

static int irecv_send( int cycles,
                       MPI_Datatype sdt, int scount, void* sbuf,
                       MPI_Datatype rdt, int rcount, void* rbuf )
{
    int myself, tag = 0, c, t, slength, rlength;
    MPI_Request req;
    MPI_Status status;
    double timers[trials];

    MPI_Type_size( sdt, &slength );
    slength *= scount;
    MPI_Type_size( rdt, &rlength );
    rlength *= rcount;

    MPI_Comm_rank( MPI_COMM_WORLD, &myself );

    for( t = 0; t < trials; t++ ) {
        timers[t] = MPI_Wtime();
        for( c = 0; c < cycles; c++ ) {
            MPI_Irecv( rbuf, rcount, rdt, myself, tag, MPI_COMM_WORLD, &req );
            MPI_Send( sbuf, scount, sdt, myself, tag, MPI_COMM_WORLD );
            MPI_Wait( &req, &status );
        }
        timers[t] = (MPI_Wtime() - timers[t]) / cycles;
    }
    print_result( rlength, trials, timers );
    return 0;
}

static int isend_irecv_wait( int cycles,
                             MPI_Datatype sdt, int scount, void* sbuf,
                             MPI_Datatype rdt, int rcount, void* rbuf )
{
    int myself, tag = 0, c, t, slength, rlength;
    MPI_Request requests[2];
    MPI_Status statuses[2];
    double timers[trials];

    MPI_Type_size( sdt, &slength );
    slength *= scount;
    MPI_Type_size( rdt, &rlength );
    rlength *= rcount;

    MPI_Comm_rank( MPI_COMM_WORLD, &myself );

    for( t = 0; t < trials; t++ ) {
        timers[t] = MPI_Wtime();
        for( c = 0; c < cycles; c++ ) {
            MPI_Isend( sbuf, scount, sdt, myself, tag, MPI_COMM_WORLD, &requests[0] );
            MPI_Irecv( rbuf, rcount, rdt, myself, tag, MPI_COMM_WORLD, &requests[1] );
            MPI_Waitall( 2, requests, statuses );
        }
        timers[t] = (MPI_Wtime() - timers[t]) / cycles;
    }
    print_result( rlength, trials, timers );
    return 0;
}

static int irecv_isend_wait( int cycles,
                             MPI_Datatype sdt, int scount, void* sbuf,
                             MPI_Datatype rdt, int rcount, void* rbuf )
{
    int myself, tag = 0, c, t, slength, rlength;
    MPI_Request requests[2];
    MPI_Status statuses[2];
    double timers[trials];

    MPI_Type_size( sdt, &slength );
    slength *= scount;
    MPI_Type_size( rdt, &rlength );
    rlength *= rcount;

    MPI_Comm_rank( MPI_COMM_WORLD, &myself );

    for( t = 0; t < trials; t++ ) {
        timers[t] = MPI_Wtime();
        for( c = 0; c < cycles; c++ ) {
            MPI_Irecv( rbuf, rcount, rdt, myself, tag, MPI_COMM_WORLD, &requests[0] );
            MPI_Isend( sbuf, scount, sdt, myself, tag, MPI_COMM_WORLD, &requests[1] );
            MPI_Waitall( 2, requests, statuses );
        }
        timers[t] = (MPI_Wtime() - timers[t]) / cycles;
    }
    print_result( rlength, trials, timers);
    return 0;
}

static int do_test_for_ddt( int doop, MPI_Datatype sddt, MPI_Datatype rddt, int length )
{
    MPI_Aint lb, extent;
    char *sbuf, *rbuf;
    int i;

    MPI_Type_get_extent( sddt, &lb, &extent );
    sbuf = (char*)malloc( length );
    rbuf = (char*)malloc( length );
    if( doop & DO_PACK ) {
        printf("# Pack (max length %d)\n", length);
        for( i = 1; i <= (length/extent); i *= 2 ) {
            pack( cycles, sddt, i, sbuf, rbuf );
        }
    }

    if( doop & DO_UNPACK ) {
        printf("# Unpack (length %d)\n", length);
        for( i = 1; i <= (length/extent); i *= 2 ) {
            unpack( cycles, sbuf, rddt, i, rbuf );
        }
    }

    if( doop & DO_ISEND_RECV ) {
        printf( "# Isend recv (length %d)\n", length );
        for( i = 1; i <= (length/extent); i *= 2 ) {
            isend_recv( cycles, sddt, i, sbuf, rddt, i, rbuf );
        }
    }

    if( doop & DO_ISEND_IRECV ) {
        printf( "# Isend Irecv Wait (length %d)\n", length );
        for( i = 1; i <= (length/extent); i *= 2 ) {
            isend_irecv_wait( cycles, sddt, i, sbuf, rddt, i, rbuf );
        }
    }

    if( doop & DO_IRECV_SEND ) {
        printf( "# Irecv send (length %d)\n", length );
        for( i = 1; i <= (length/extent); i *= 2 ) {
            irecv_send( cycles, sddt, i, sbuf, rddt, i, rbuf );
        }
    }

    if( doop & DO_IRECV_SEND ) {
        printf( "# Irecv Isend Wait (length %d)\n", length );
        for( i = 1; i <= (length/extent); i *= 2 ) {
            irecv_isend_wait( cycles, sddt, i, sbuf, rddt, i, rbuf );
        }
    }
    free( sbuf );
    free( rbuf );
    return 0;
}

int main( int argc, char* argv[] )
{
    int run_tests = DO_STRUCT_MERGED_WITH_GAP_RESIZED;  /* do all datatype tests by default */
    int rank, size;
    MPI_Datatype ddt;

    run_tests |= DO_PACK | DO_UNPACK;

    MPI_Init (&argc, &argv);

    MPI_Comm_rank (MPI_COMM_WORLD, &rank);
    MPI_Comm_size (MPI_COMM_WORLD, &size);

    if( rank != 0 ) {
        MPI_Finalize();
        exit(0);
    }

    if( run_tests & DO_CONTIG ) {
        printf( "\ncontiguous datatype\n\n" );
        do_test_for_ddt( run_tests, MPI_INT, MPI_INT, MAX_LENGTH );
    }

    if( run_tests & DO_INDEXED_GAP ) {
        printf( "\nindexed gap\n\n" );
        ddt = create_indexed_gap_ddt();
        MPI_DDT_DUMP( ddt );
        do_test_for_ddt( run_tests, ddt, ddt, MAX_LENGTH );
        MPI_Type_free( &ddt );
    }

    if( run_tests & DO_OPTIMIZED_INDEXED_GAP ) {
        printf( "\noptimized indexed gap\n\n" );
        ddt = create_indexed_gap_optimized_ddt();
        MPI_DDT_DUMP( ddt );
        do_test_for_ddt( run_tests, ddt, ddt, MAX_LENGTH );
        MPI_Type_free( &ddt );
    }

    if( run_tests & DO_CONSTANT_GAP ) {
        printf( "\nconstant indexed gap\n\n" );
        ddt = create_indexed_constant_gap_ddt( 80, 100, 1 );
        MPI_DDT_DUMP( ddt );
        do_test_for_ddt( run_tests, ddt, ddt, MAX_LENGTH );
        MPI_Type_free( &ddt );
    }

    if( run_tests & DO_CONSTANT_GAP ) {
        printf( "\noptimized constant indexed gap\n\n" );
        ddt = create_optimized_indexed_constant_gap_ddt( 80, 100, 1 );
        MPI_DDT_DUMP( ddt );
        do_test_for_ddt( run_tests, ddt, ddt, MAX_LENGTH );
        MPI_Type_free( &ddt );
    }

    if( run_tests & DO_STRUCT_CONSTANT_GAP_RESIZED ) {
        printf( "\nstruct constant gap resized\n\n" );
        ddt = create_struct_constant_gap_resized_ddt( 0 /* unused */, 0 /* unused */, 0 /* unused */ );
        MPI_DDT_DUMP( ddt );
        do_test_for_ddt( run_tests, ddt, ddt, MAX_LENGTH );
        MPI_Type_free( &ddt );
    }

    if( run_tests & DO_STRUCT_MERGED_WITH_GAP_RESIZED ) {
        printf( "\nstruct constant gap resized\n\n" );
        ddt = create_merged_contig_with_gaps( 1 );
        MPI_DDT_DUMP( ddt );
        do_test_for_ddt( run_tests, ddt, ddt, MAX_LENGTH );
        MPI_Type_free( &ddt );
    }

    MPI_Finalize ();
    exit(0);
}