void SendData(ArgStruct *p)
{
if(p->bufflen%8==0)
shmem_put64(p->s_ptr,p->s_ptr,p->bufflen/8,p->prot.nbor);
else
shmem_putmem(p->s_ptr,p->s_ptr,p->bufflen,p->prot.nbor);
}
int main(int argc, char * argv[])
{
/* Get PE information */
shmem_init();
me = _my_pe();
npes = _num_pes();
/* Initialize and send on PE 0 */
if (me == 0) {
for (i = 0; i < 8; i++)
src[i] = i + 1;
/* Put source date at PE 0 to dest at PE 1+ */
for (i = 1; i < npes; i++)
shmem_put64(dest, src, 8 * sizeof(int) / 8, i);
}
/* Make sure the transfer is complete */
shmem_barrier_all();
/* Print from PE 1+ */
if (me > 0) {
printf("PE %d: %d", me, dest[0]);
for (i = 1; i < 8; i++)
printf(",%d", dest[i]);
printf("\n");
}
shmem_finalize();
return 0;
}
int64 do_all2all (uint64 *dst, uint64 *src, int64 len, int64 nwrd)
{
static char cvs_info[] = "BMKGRP $Date: $ $Revision: $ $RCSfile: all2all.c,v $ $Name: $";
int64 i, j, pe;
len = len - (len % (nwrd * SIZE)); // force even multiple
for (i = 0; i < len; i+=SIZE*nwrd) {
shmem_barrier_all();
for (j = 0; j < SIZE; j++) {
pe = PERM(SELF,SIZE,j);
/* shmem_put (&dst[i + SELF*nwrd], &src[i + pe*nwrd], nwrd, pe);*/
shmem_put64 (&dst[i + SELF*nwrd], &src[i + pe*nwrd], nwrd, pe);
}
}
return len;
}
int
main (int argc, char **argv)
{
int i;
int nextpe;
int me, npes;
int success1, success2, success3, success4, success5, success6, success7,
success8;
short src1[N];
int src2[N];
long src3[N];
long double src4[N];
long long src5[N];
double src6[N];
float src7[N];
char *src8;
short src9;
int src10;
long src11;
double src12;
float src13;
short *dest1;
int *dest2;
long *dest3;
long double *dest4;
long long *dest5;
double *dest6;
float *dest7;
char *dest8;
short *dest9;
int *dest10;
long *dest11;
double *dest12;
float *dest13;
shmem_init ();
me = shmem_my_pe ();
npes = shmem_n_pes ();
if (npes > 1) {
success1 = 0;
success2 = 0;
success3 = 0;
success4 = 0;
success5 = 0;
success6 = 0;
success7 = 0;
success8 = 0;
src8 = (char *) malloc (N * sizeof (char));
for (i = 0; i < N; i += 1) {
src1[i] = (short) me;
src2[i] = me;
src3[i] = (long) me;
src4[i] = (long double) me;
src5[i] = (long long) me;
src6[i] = (double) me;
src7[i] = (float) me;
src8[i] = (char) me;
}
src9 = (short) me;
src10 = me;
src11 = (long) me;
src12 = (double) me;
src13 = (float) me;
dest1 = (short *) shmem_malloc (N * sizeof (*dest1));
dest2 = (int *) shmem_malloc (N * sizeof (*dest2));
dest3 = (long *) shmem_malloc (N * sizeof (*dest3));
dest4 = (long double *) shmem_malloc (N * sizeof (*dest4));
dest5 = (long long *) shmem_malloc (N * sizeof (*dest5));
dest6 = (double *) shmem_malloc (N * sizeof (*dest6));
dest7 = (float *) shmem_malloc (N * sizeof (*dest7));
dest8 = (char *) shmem_malloc (4 * sizeof (*dest8));
dest9 = (short *) shmem_malloc (sizeof (*dest9));
dest10 = (int *) shmem_malloc (sizeof (*dest10));
dest11 = (long *) shmem_malloc (sizeof (*dest11));
dest12 = (double *) shmem_malloc (sizeof (*dest12));
dest13 = (float *) shmem_malloc (sizeof (*dest13));
for (i = 0; i < N; i += 1) {
dest1[i] = -9;
dest2[i] = -9;
dest3[i] = -9;
dest4[i] = -9;
dest5[i] = -9;
dest6[i] = -9;
dest7[i] = -9.0;
dest8[i] = -9;
}
*dest9 = -9;
*dest10 = -9;
*dest11 = -9;
*dest12 = -9;
*dest13 = -9.0;
nextpe = (me + 1) % npes;
/* Testing shmem_short_put, shmem_int_put, shmem_long_put,
shmem_longdouble_put, shmem_longlong_put, shmem_double_put,
shmem_float_put, shmem_putmem */
shmem_barrier_all ();
shmem_short_put (dest1, src1, N, nextpe);
shmem_int_put (dest2, src2, N, nextpe);
shmem_long_put (dest3, src3, N, nextpe);
shmem_longdouble_put (dest4, src4, N, nextpe);
shmem_longlong_put (dest5, src5, N, nextpe);
shmem_double_put (dest6, src6, N, nextpe);
shmem_float_put (dest7, src7, N, nextpe);
shmem_putmem (dest8, src8, N * sizeof (char), nextpe);
shmem_barrier_all ();
if (me == 0) {
for (i = 0; i < N; i += 1) {
if (dest1[i] != (npes - 1)) {
success1 = 1;
}
if (dest2[i] != (npes - 1)) {
success2 = 1;
}
if (dest3[i] != (npes - 1)) {
success3 = 1;
}
if (dest4[i] != (npes - 1)) {
success4 = 1;
}
if (dest5[i] != (npes - 1)) {
success5 = 1;
}
if (dest6[i] != (npes - 1)) {
success6 = 1;
}
if (dest7[i] != (npes - 1)) {
success7 = 1;
}
if (dest8[i] != (npes - 1)) {
success8 = 1;
}
}
if (success1 == 0)
printf ("Test shmem_short_put: Passed\n");
else
printf ("Test shmem_short_put: Failed\n");
if (success2 == 0)
printf ("Test shmem_int_put: Passed\n");
else
printf ("Test shmem_int_put: Failed\n");
if (success3 == 0)
printf ("Test shmem_long_put: Passed\n");
else
printf ("Test shmem_long_put: Failed\n");
if (success4 == 0)
printf ("Test shmem_longdouble_put: Passed\n");
else
printf ("Test shmem_longdouble_put: Failed\n");
if (success5 == 0)
printf ("Test shmem_longlong_put: Passed\n");
else
printf ("Test shmem_longlong_put: Failed\n");
if (success6 == 0)
printf ("Test shmem_double_put: Passed\n");
else
printf ("Test shmem_double_put: Failed\n");
if (success7 == 0)
printf ("Test shmem_float_put: Passed\n");
else
printf ("Test shmem_float_put: Failed\n");
if (success8 == 0)
printf ("Test shmem_putmem: Passed\n");
else
printf ("Test shmem_putmem: Failed\n");
}
shmem_barrier_all ();
/* Testing shmem_put32, shmem_put64, shmem_put128 */
if (sizeof (int) == 4) {
for (i = 0; i < N; i += 1) {
dest2[i] = -9;
dest3[i] = -9;
dest4[i] = -9;
}
success2 = 0;
success3 = 0;
success4 = 0;
shmem_barrier_all ();
shmem_put32 (dest2, src2, N, nextpe);
shmem_put64 (dest3, src3, N, nextpe);
shmem_put128 (dest4, src4, N, nextpe);
shmem_barrier_all ();
if (me == 0) {
for (i = 0; i < N; i += 1) {
if (dest2[i] != (npes - 1)) {
success2 = 1;
}
if (dest3[i] != (npes - 1)) {
success3 = 1;
}
if (dest4[i] != (npes - 1)) {
success4 = 1;
}
}
if (success2 == 0)
printf ("Test shmem_put32: Passed\n");
else
printf ("Test shmem_put32: Failed\n");
if (success3 == 0)
printf ("Test shmem_put64: Passed\n");
else
printf ("Test shmem_put64: Failed\n");
if (success4 == 0)
printf ("Test shmem_put128: Passed\n");
else
printf ("Test shmem_put128: Failed\n");
}
}
else if (sizeof (int) == 8) {
for (i = 0; i < N; i += 1) {
dest1[i] = -9;
dest2[i] = -9;
dest3[i] = -9;
}
success1 = 0;
success2 = 0;
success3 = 0;
shmem_barrier_all ();
shmem_put32 (dest1, src1, N, nextpe);
shmem_put64 (dest2, src2, N, nextpe);
shmem_put128 (dest3, src3, N, nextpe);
shmem_barrier_all ();
if (me == 0) {
for (i = 0; i < N; i += 1) {
if (dest1[i] != (npes - 1)) {
success1 = 1;
}
if (dest2[i] != (npes - 1)) {
success2 = 1;
}
if (dest3[i] != (npes - 1)) {
success3 = 1;
}
}
if (success1 == 0)
printf ("Test shmem_put32: Passed\n");
else
printf ("Test shmem_put32: Failed\n");
if (success2 == 0)
printf ("Test shmem_put64: Passed\n");
else
printf ("Test shmem_put64: Failed\n");
if (success3 == 0)
printf ("Test shmem_put128: Passed\n");
else
printf ("Test shmem_put128: Failed\n");
}
}
/* Testing shmem_iput32, shmem_iput64, shmem_iput128 */
shmem_barrier_all ();
if (sizeof (int) == 4) {
for (i = 0; i < N; i += 1) {
dest2[i] = -9;
dest3[i] = -9;
dest4[i] = -9;
}
success2 = 0;
success3 = 0;
success4 = 0;
shmem_barrier_all ();
shmem_iput32 (dest2, src2, 1, 2, N, nextpe);
shmem_iput64 (dest3, src3, 1, 2, N, nextpe);
shmem_iput128 (dest4, src4, 1, 2, N, nextpe);
shmem_barrier_all ();
if (me == 0) {
for (i = 0; i < N / 2; i += 1) {
if (dest2[i] != (npes - 1)) {
success2 = 1;
}
if (dest3[i] != (npes - 1)) {
success3 = 1;
}
if (dest4[i] != (npes - 1)) {
success4 = 1;
}
}
if (success2 == 0)
printf ("Test shmem_iput32: Passed\n");
else
printf ("Test shmem_iput32: Failed\n");
if (success3 == 0)
printf ("Test shmem_iput64: Passed\n");
else
printf ("Test shmem_iput64: Failed\n");
if (success4 == 0)
printf ("Test shmem_iput128: Passed\n");
else
printf ("Test shmem_iput128: Failed\n");
}
}
else if (sizeof (int) == 8) {
for (i = 0; i < N; i += 1) {
dest1[i] = -9;
dest2[i] = -9;
dest3[i] = -9;
}
success1 = 0;
success2 = 0;
success3 = 0;
shmem_barrier_all ();
shmem_iput32 (dest1, src1, 1, 2, N, nextpe);
shmem_iput64 (dest2, src2, 1, 2, N, nextpe);
shmem_iput128 (dest3, src3, 1, 2, N, nextpe);
shmem_barrier_all ();
if (me == 0) {
for (i = 0; i < N / 2; i += 1) {
if (dest1[i] != (npes - 1)) {
success1 = 1;
}
if (dest2[i] != (npes - 1)) {
success2 = 1;
}
if (dest3[i] != (npes - 1)) {
success3 = 1;
}
}
if (success1 == 0)
printf ("Test shmem_iput32: Passed\n");
else
printf ("Test shmem_iput32: Failed\n");
if (success2 == 0)
printf ("Test shmem_iput64: Passed\n");
else
printf ("Test shmem_iput64: Failed\n");
if (success3 == 0)
printf ("Test shmem_iput128: Passed\n");
else
printf ("Test shmem_iput128: Failed\n");
}
}
/* Testing shmem_short_iput, shmem_int_iput, shmem_long_iput,
shmem_double_iput, shmem_float_iput */
for (i = 0; i < N; i += 1) {
dest1[i] = -9;
dest2[i] = -9;
dest3[i] = -9;
dest6[i] = -9;
dest7[i] = -9;
}
success1 = 0;
success2 = 0;
success3 = 0;
success6 = 0;
success7 = 0;
shmem_barrier_all ();
shmem_short_iput (dest1, src1, 1, 2, N, nextpe);
shmem_int_iput (dest2, src2, 1, 2, N, nextpe);
shmem_long_iput (dest3, src3, 1, 2, N, nextpe);
shmem_double_iput (dest6, src6, 1, 2, N, nextpe);
shmem_float_iput (dest7, src7, 1, 2, N, nextpe);
shmem_barrier_all ();
if (me == 0) {
for (i = 0; i < N / 2; i += 1) {
if (dest1[i] != (npes - 1)) {
success1 = 1;
}
if (dest2[i] != (npes - 1)) {
success2 = 1;
}
if (dest3[i] != (npes - 1)) {
success3 = 1;
}
if (dest6[i] != (npes - 1)) {
success6 = 1;
}
if (dest7[i] != (npes - 1)) {
success7 = 1;
}
}
if (success1 == 0)
printf ("Test shmem_short_iput: Passed\n");
else
printf ("Test shmem_short_iput: Failed\n");
if (success2 == 0)
printf ("Test shmem_int_iput: Passed\n");
else
printf ("Test shmem_int_iput: Failed\n");
if (success3 == 0)
printf ("Test shmem_long_iput: Passed\n");
else
printf ("Test shmem_long_iput: Failed\n");
if (success6 == 0)
printf ("Test shmem_double_iput: Passed\n");
else
printf ("Test shmem_double_iput: Failed\n");
if (success7 == 0)
printf ("Test shmem_float_iput: Passed\n");
else
printf ("Test shmem_float_iput: Failed\n");
}
/* Testing shmem_double_p, shmem_float_p, shmem_int_p, shmem_long_p,
shmem_short_p */
shmem_barrier_all ();
shmem_short_p (dest9, src9, nextpe);
shmem_int_p (dest10, src10, nextpe);
shmem_long_p (dest11, src11, nextpe);
shmem_double_p (dest12, src12, nextpe);
shmem_float_p (dest13, src13, nextpe);
shmem_barrier_all ();
if (me == 0) {
if (*dest9 == (npes - 1))
printf ("Test shmem_short_p: Passed\n");
else
printf ("Test shmem_short_p: Failed\n");
if (*dest10 == (npes - 1))
printf ("Test shmem_int_p: Passed\n");
else
printf ("Test shmem_int_p: Failed\n");
if (*dest11 == (npes - 1))
printf ("Test shmem_long_p: Passed\n");
else
printf ("Test shmem_long_p: Failed\n");
if (*dest12 == (npes - 1))
printf ("Test shmem_double_p: Passed\n");
else
printf ("Test shmem_double_p: Failed\n");
if (*dest13 == (npes - 1))
printf ("Test shmem_float_p: Passed\n");
else
printf ("Test shmem_float_p: Failed\n");
}
shmem_barrier_all ();
shmem_free (dest1);
shmem_free (dest2);
shmem_free (dest3);
shmem_free (dest4);
shmem_free (dest5);
shmem_free (dest6);
shmem_free (dest7);
shmem_free (dest8);
shmem_free (dest9);
shmem_free (dest10);
shmem_free (dest11);
shmem_free (dest12);
shmem_free (dest13);
}
else {
printf ("Number of PEs must be > 1 to test shmem put, test skipped\n");
}
shmem_finalize ();
return 0;
}
int main (int argc, char *argv[]) {
/**** Initialising ****/
const unsigned long long full_program_start = current_time_ns();
{
shmem_init ();
/* Variable Declarations */
int Numprocs,MyRank, Root = 0;
int i,j,k, NoofElements, NoofElements_Bloc,
NoElementsToSort;
int count, temp;
TYPE *Input, *InputData;
TYPE *Splitter, *AllSplitter;
TYPE *Buckets, *BucketBuffer, *LocalBucket;
TYPE *OutputBuffer, *Output;
MyRank = shmem_my_pe ();
Numprocs = shmem_n_pes ();
NoofElements = SIZE;
if(( NoofElements % Numprocs) != 0){
if(MyRank == Root)
printf("Number of Elements are not divisible by Numprocs \n");
shmem_finalize ();
exit(0);
}
/**** Reading Input ****/
Input = (TYPE *) shmem_malloc (NoofElements*sizeof(*Input));
if(Input == NULL) {
printf("Error : Can not allocate memory \n");
}
if (MyRank == Root){
/* Initialise random number generator */
printf ("Generating input Array for Sorting %d uint64_t numbers\n",SIZE);
srand48((TYPE)NoofElements);
for(i=0; i< NoofElements; i++) {
Input[i] = rand();
}
}
/**** Sending Data ****/
NoofElements_Bloc = NoofElements / Numprocs;
InputData = (TYPE *) shmem_malloc (NoofElements_Bloc * sizeof (*InputData));
if(InputData == NULL) {
printf("Error : Can not allocate memory \n");
}
//MPI_Scatter(Input, NoofElements_Bloc, TYPE_MPI, InputData,
// NoofElements_Bloc, TYPE_MPI, Root, MPI_COMM_WORLD);
shmem_barrier_all();
if(MyRank == Root) {
for(i=0; i<Numprocs; i++) {
TYPE* start = &Input[i * NoofElements_Bloc];
shmem_put64(InputData, start, NoofElements_Bloc, i);
}
}
shmem_barrier_all();
/**** Sorting Locally ****/
sorting(InputData, NoofElements_Bloc);
/**** Choosing Local Splitters ****/
Splitter = (TYPE *) shmem_malloc (sizeof (TYPE) * (Numprocs-1));
if(Splitter == NULL) {
printf("Error : Can not allocate memory \n");
}
for (i=0; i< (Numprocs-1); i++){
Splitter[i] = InputData[NoofElements/(Numprocs*Numprocs) * (i+1)];
}
/**** Gathering Local Splitters at Root ****/
AllSplitter = (TYPE *) shmem_malloc (sizeof (TYPE) * Numprocs * (Numprocs-1));
if(AllSplitter == NULL) {
printf("Error : Can not allocate memory \n");
}
//MPI_Gather (Splitter, Numprocs-1, TYPE_MPI, AllSplitter, Numprocs-1,
// TYPE_MPI, Root, MPI_COMM_WORLD);
shmem_barrier_all();
TYPE* target_index = &AllSplitter[MyRank * (Numprocs-1)];
shmem_put64(target_index, Splitter, Numprocs-1, Root);
shmem_barrier_all();
/**** Choosing Global Splitters ****/
if (MyRank == Root){
sorting (AllSplitter, Numprocs*(Numprocs-1));
for (i=0; i<Numprocs-1; i++)
Splitter[i] = AllSplitter[(Numprocs-1)*(i+1)];
}
/**** Broadcasting Global Splitters ****/
//MPI_Bcast (Splitter, Numprocs-1, TYPE_MPI, 0, MPI_COMM_WORLD);
{ int _i; for(_i=0; _i<_SHMEM_BCAST_SYNC_SIZE; _i++) { pSync[_i] = _SHMEM_SYNC_VALUE; } shmem_barrier_all(); }
shmem_broadcast64(Splitter, Splitter, Numprocs-1, 0, 0, 0, Numprocs, pSync);
shmem_barrier_all();
/**** Creating Numprocs Buckets locally ****/
Buckets = (TYPE *) shmem_malloc (sizeof (TYPE) * (NoofElements + Numprocs));
if(Buckets == NULL) {
printf("Error : Can not allocate memory \n");
}
j = 0;
k = 1;
for (i=0; i<NoofElements_Bloc; i++){
if(j < (Numprocs-1)){
if (InputData[i] < Splitter[j])
Buckets[((NoofElements_Bloc + 1) * j) + k++] = InputData[i];
else{
Buckets[(NoofElements_Bloc + 1) * j] = k-1;
k=1;
j++;
i--;
}
}
else
Buckets[((NoofElements_Bloc + 1) * j) + k++] = InputData[i];
}
Buckets[(NoofElements_Bloc + 1) * j] = k - 1;
shmem_free(Splitter);
shmem_free(AllSplitter);
/**** Sending buckets to respective processors ****/
BucketBuffer = (TYPE *) shmem_malloc (sizeof (TYPE) * (NoofElements + Numprocs));
if(BucketBuffer == NULL) {
printf("Error : Can not allocate memory \n");
}
//MPI_Alltoall (Buckets, NoofElements_Bloc + 1, TYPE_MPI, BucketBuffer,
// NoofElements_Bloc + 1, TYPE_MPI, MPI_COMM_WORLD);
shmem_barrier_all();
for(i=0; i<Numprocs; i++) {
shmem_put64(&BucketBuffer[MyRank*(NoofElements_Bloc + 1)], &Buckets[i*(NoofElements_Bloc + 1)], NoofElements_Bloc + 1, i);
}
shmem_barrier_all();
/**** Rearranging BucketBuffer ****/
LocalBucket = (TYPE *) shmem_malloc (sizeof (TYPE) * 2 * NoofElements / Numprocs);
if(LocalBucket == NULL) {
printf("Error : Can not allocate memory \n");
}
count = 1;
for (j=0; j<Numprocs; j++) {
k = 1;
for (i=0; i<BucketBuffer[(NoofElements/Numprocs + 1) * j]; i++)
LocalBucket[count++] = BucketBuffer[(NoofElements/Numprocs + 1) * j + k++];
}
LocalBucket[0] = count-1;
/**** Sorting Local Buckets using Bubble Sort ****/
/*sorting (InputData, NoofElements_Bloc, sizeof(int), intcompare); */
NoElementsToSort = LocalBucket[0];
sorting (&LocalBucket[1], NoElementsToSort);
/**** Gathering sorted sub blocks at root ****/
OutputBuffer = (TYPE *) shmem_malloc (sizeof(TYPE) * 2 * NoofElements);
if(OutputBuffer == NULL) {
printf("Error : Can not allocate memory \n");
}
//MPI_Gather (LocalBucket, 2*NoofElements_Bloc, TYPE_MPI, OutputBuffer,
// 2*NoofElements_Bloc, TYPE_MPI, Root, MPI_COMM_WORLD);
shmem_barrier_all();
target_index = &OutputBuffer[MyRank * (2*NoofElements_Bloc)];
shmem_put64(target_index, LocalBucket, 2*NoofElements_Bloc, Root);
shmem_barrier_all();
/**** Rearranging output buffer ****/
if (MyRank == Root){
Output = (TYPE *) malloc (sizeof (TYPE) * NoofElements);
count = 0;
for(j=0; j<Numprocs; j++){
k = 1;
for(i=0; i<OutputBuffer[(2 * NoofElements/Numprocs) * j]; i++)
Output[count++] = OutputBuffer[(2*NoofElements/Numprocs) * j + k++];
}
printf ( "Number of Elements to be sorted : %d \n", NoofElements);
TYPE prev = 0;
int fail = 0;
for (i=0; i<NoofElements; i++){
if(Output[i] < prev) { printf("Failed at index %d\n",i); fail = 1; }
prev = Output[i];
}
if(fail) printf("Sorting FAILED\n");
else printf("Sorting PASSED\n");
free(Output);
}/* MyRank==0*/
shmem_free(Input);
shmem_free(OutputBuffer);
shmem_free(InputData);
shmem_free(Buckets);
shmem_free(BucketBuffer);
shmem_free(LocalBucket);
/**** Finalize ****/
shmem_finalize();
} ;
const unsigned long long full_program_end = current_time_ns();
printf("full_program %llu ns\n", full_program_end - full_program_start);
}
int main(int argc, char **argv)
{
int j;
int my_pe,n_pes;
int *flag,*one;
size_t max_elements,max_elements_bytes;
char *srce_char,*targ_char;
short *srce_short,*targ_short;
int *srce_int,*targ_int;
long *srce_long,*targ_long;
start_pes(0);
my_pe = shmem_my_pe();
n_pes = shmem_n_pes();
flag = shmalloc((size_t) sizeof(int));
one = shmalloc((size_t) sizeof(int));
*one = 1;
/* fail if trying to use odd number of processors */
if ( (n_pes % 2) != 0 ){
fprintf(stderr, "FAIL - test requires even number of PEs\n");
exit(1);
}
if(my_pe == 0)
fprintf(stderr, "shmem_num_put(%s)\n", argv[0]);
/* shmem_putmem test */
*flag = 0;
max_elements = (size_t) (MAX_SIZE / sizeof(char));
max_elements_bytes = (size_t) (sizeof(char)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_putmem max_elements = %d\n",max_elements);
srce_char = shmalloc(max_elements_bytes);
targ_char = shmalloc(max_elements_bytes);
if((srce_char == NULL) || (targ_char == NULL))
shmalloc_error();
if ( (my_pe % 2) == 0 )
for(j = 0; j < max_elements; j++)
srce_char[j] = (char)(my_pe+j);
else
for(j = 0; j < max_elements; j++)
targ_char[j] = (char)(my_pe+j);
shmem_barrier_all();
if ( (my_pe % 2) == 0 ) {
shmem_putmem(targ_char,srce_char,max_elements,my_pe+1);
shmem_quiet();
shmem_int_put(flag,one,(size_t)1,my_pe+1);
} else {
shmem_int_wait(flag,0);
for(j = 0; j < max_elements; j++)
if ( targ_char[j] != (char)(my_pe+j-1) )
fprintf(stderr, "FAIL: PE [%d] targ_char[%d]=%d my_pe+j-1=%d\n",
my_pe,j,targ_char[j],my_pe+j-1);
}
shfree(srce_char); shfree(targ_char);
/* shmem_put16 test */
*flag = 0;
max_elements = (size_t) (MAX_SIZE / sizeof(short));
if(max_elements > 20000) max_elements=20000;
max_elements_bytes = (size_t) (sizeof(short)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_put16 max_elements = %d\n",max_elements);
srce_short = shmalloc(max_elements_bytes);
targ_short = shmalloc(max_elements_bytes);
if((srce_short == NULL) || (targ_short == NULL))
shmalloc_error();
if ( (my_pe % 2) == 0 )
for(j = 0; j < max_elements; j++)
srce_short[j] = (short)(my_pe+j);
else
for(j = 0; j < max_elements; j++)
targ_short[j] = (short)(my_pe+j);
shmem_barrier_all();
if ( (my_pe % 2) == 0 ) {
shmem_put16(targ_short,srce_short,max_elements,my_pe+1);
shmem_quiet();
shmem_int_put(flag,one,(size_t)1,my_pe+1);
} else {
shmem_int_wait(flag,0);
for(j = 0; j < max_elements; j++)
if ( targ_short[j] != (short)(my_pe+j-1) )
fprintf(stderr, "FAIL: PE [%d] targ_short[%d]=%d my_pe+j-1=%d\n",
my_pe,j,targ_short[j],my_pe+j-1);
}
shfree(srce_short); shfree(targ_short);
/* shmem_put32 test */
*flag = 0;
max_elements = (size_t) (MAX_SIZE / sizeof(int));
max_elements_bytes = (size_t) (sizeof(int)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_put32 max_elements = %d\n",max_elements);
srce_int = shmalloc(max_elements_bytes);
targ_int = shmalloc(max_elements_bytes);
if((srce_int == NULL) || (targ_int == NULL))
shmalloc_error();
if ( (my_pe % 2) == 0 )
for(j = 0; j < max_elements; j++)
srce_int[j] = (int)(my_pe+j);
else
for(j = 0; j < max_elements; j++)
targ_int[j] = (int)(my_pe+j);
shmem_barrier_all();
if ( (my_pe % 2) == 0 ) {
shmem_put32(targ_int,srce_int,max_elements,my_pe+1);
shmem_quiet();
shmem_int_put(flag,one,(size_t)1,my_pe+1);
} else {
shmem_int_wait(flag,0);
for(j = 0; j < max_elements; j++)
if ( targ_int[j] != (int)(my_pe+j-1) )
fprintf(stderr, "FAIL: PE [%d] targ_int[%d]=%d my_pe+j-1=%d\n",
my_pe,j,targ_int[j],my_pe+j-1);
}
shfree(srce_int); shfree(targ_int);
/* shmem_put64 test */
*flag = 0;
max_elements = (size_t) (MAX_SIZE / sizeof(long));
max_elements_bytes = (size_t) (sizeof(long)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_put64 max_elements = %d\n",max_elements);
srce_long = shmalloc(max_elements_bytes);
targ_long = shmalloc(max_elements_bytes);
if((srce_long == NULL) || (targ_long == NULL))
shmalloc_error();
if ( (my_pe % 2) == 0 )
for(j = 0; j < max_elements; j++)
srce_long[j] = (long)(my_pe+j);
else
for(j = 0; j < max_elements; j++)
targ_long[j] = (long)(my_pe+j);
shmem_barrier_all();
if ( (my_pe % 2) == 0 ) {
shmem_put64(targ_long,srce_long,max_elements,my_pe+1);
shmem_quiet();
shmem_int_put(flag,one,(size_t)1,my_pe+1);
} else {
shmem_int_wait(flag,0);
for(j = 0; j < max_elements; j++)
if ( targ_long[j] != (long)(my_pe+j-1) )
fprintf(stderr, "FAIL: PE [%d] targ_long[%d]=%d my_pe+j-1=%d\n",
my_pe,j,targ_long[j],my_pe+j-1);
}
shfree(srce_long); shfree(targ_long);
/* shmem_put128 test */
*flag = 0;
max_elements = (size_t) (MAX_SIZE / sizeof(long));
if ( (max_elements % 2) != 0)
max_elements = max_elements-1;
max_elements_bytes = (size_t) (sizeof(long)*max_elements);
max_elements = max_elements/2;
if(my_pe == 0)
fprintf(stderr,"shmem_put128 max_elements = %d\n",max_elements);
srce_long = shmalloc(max_elements_bytes);
targ_long = shmalloc(max_elements_bytes);
if((srce_long == NULL) || (targ_long == NULL))
shmalloc_error();
if ( (my_pe % 2) == 0 )
for(j = 0; j < 2*max_elements; j++)
srce_long[j] = (long)(my_pe+j);
else
for(j = 0; j < 2*max_elements; j++)
targ_long[j] = (long)(my_pe+j);
shmem_barrier_all();
if ( (my_pe % 2) == 0 ) {
shmem_put128(targ_long,srce_long,max_elements,my_pe+1);
shmem_quiet();
shmem_int_put(flag,one,(size_t)1,my_pe+1);
} else {
shmem_int_wait(flag,0);
for(j = 0; j < 2*max_elements; j++)
if ( targ_long[j] != (long)(my_pe+j-1) )
fprintf(stderr, "FAIL: PE [%d] targ_long[%d]=%d my_pe+j-1=%d\n",
my_pe,j,targ_long[j],my_pe+j-1);
}
shfree(srce_long); shfree(targ_long);
#ifdef SHMEM_C_GENERIC_32
/* shmem_put (GENERIC 32) test */
*flag = 0;
max_elements = (size_t) (MAX_SIZE / sizeof(int));
max_elements_bytes = (size_t) (sizeof(int)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_put (GENERIC 32) max_elements = %d\n",max_elements);
srce_int = shmalloc(max_elements_bytes);
targ_int = shmalloc(max_elements_bytes);
if((srce_int == NULL) || (targ_int == NULL))
shmalloc_error();
if ( (my_pe % 2) == 0 )
for(j = 0; j < max_elements; j++)
srce_int[j] = (int)(my_pe+j);
else
for(j = 0; j < max_elements; j++)
targ_int[j] = (int)(my_pe+j);
shmem_barrier_all();
if ( (my_pe % 2) == 0 ) {
shmem_put(targ_int,srce_int,max_elements,my_pe+1);
shmem_quiet();
shmem_int_put(flag,one,(size_t)1,my_pe+1);
} else {
shmem_int_wait(flag,0);
for(j = 0; j < max_elements; j++)
if ( targ_int[j] != (int)(my_pe+j-1) )
fprintf(stderr, "FAIL: PE [%d] targ_int[%d]=%d my_pe+j-1=%d\n",
my_pe,j,targ_int[j],my_pe+j-1);
}
shfree(srce_int); shfree(targ_int);
#else
/* shmem_put (GENERIC 64) test */
*flag = 0;
max_elements = (size_t) (MAX_SIZE / sizeof(long));
max_elements_bytes = (size_t) (sizeof(long)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_put (GENERIC 64) max_elements = %d\n",max_elements);
srce_long = shmalloc(max_elements_bytes);
targ_long = shmalloc(max_elements_bytes);
if((srce_long == NULL) || (targ_long == NULL))
shmalloc_error();
if ( (my_pe % 2) == 0 )
for(j = 0; j < max_elements; j++)
srce_long[j] = (long)(my_pe+j);
else
for(j = 0; j < max_elements; j++)
targ_long[j] = (long)(my_pe+j);
shmem_barrier_all();
if ( (my_pe % 2) == 0 ) {
shmem_put(targ_long,srce_long,max_elements,my_pe+1);
shmem_quiet();
shmem_int_put(flag,one,(size_t)1,my_pe+1);
} else {
shmem_int_wait(flag,0);
for(j = 0; j < max_elements; j++)
if ( targ_long[j] != (long)(my_pe+j-1) )
fprintf(stderr, "FAIL: PE [%d] targ_long[%d]=%d my_pe+j-1=%d\n",
my_pe,j,targ_long[j],my_pe+j-1);
}
shfree(srce_long); shfree(targ_long);
#endif
#ifdef NEEDS_FINALIZE
shmem_finalize();
#endif
return 0;
}