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;
shmem_init();
my_pe = shmem_my_pe();
n_pes = shmem_n_pes();
flag = shmem_malloc((size_t) sizeof(int));
one = shmem_malloc((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_nb(%s)\n", argv[0]);
/* shmem_putmem_nb 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_nb max_elements = %d\n",max_elements);
srce_char = shmem_malloc(max_elements_bytes);
targ_char = shmem_malloc(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_nb(targ_char,srce_char,max_elements,my_pe+1,NULL);
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);
}
shmem_free(srce_char); shmem_free(targ_char);
/* shmem_put16_nb 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_nb max_elements = %d\n",max_elements);
srce_short = shmem_malloc(max_elements_bytes);
targ_short = shmem_malloc(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_nb(targ_short,srce_short,max_elements,my_pe+1,NULL);
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);
}
shmem_free(srce_short); shmem_free(targ_short);
/* shmem_put32_nb 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_nb max_elements = %d\n",max_elements);
srce_int = shmem_malloc(max_elements_bytes);
targ_int = shmem_malloc(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_nb(targ_int,srce_int,max_elements,my_pe+1,NULL);
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);
}
shmem_free(srce_int); shmem_free(targ_int);
/* shmem_put64_nb 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_nb max_elements = %d\n",max_elements);
srce_long = shmem_malloc(max_elements_bytes);
targ_long = shmem_malloc(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_nb(targ_long,srce_long,max_elements,my_pe+1,NULL);
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);
}
shmem_free(srce_long); shmem_free(targ_long);
/* shmem_put128_nb 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_nb max_elements = %d\n",max_elements);
srce_long = shmem_malloc(max_elements_bytes);
targ_long = shmem_malloc(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_nb(targ_long,srce_long,max_elements,my_pe+1,NULL);
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);
}
shmem_free(srce_long); shmem_free(targ_long);
#ifdef SHMEM_C_GENERIC_32
/* shmem_put_nb (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_nb (GENERIC 32) max_elements = %d\n",max_elements);
srce_int = shmem_malloc(max_elements_bytes);
targ_int = shmem_malloc(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_nb(targ_int,srce_int,max_elements,my_pe+1,NULL);
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);
}
shmem_free(srce_int); shmem_free(targ_int);
#else
/* shmem_put_nb (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_nb (GENERIC 64) max_elements = %d\n",max_elements);
srce_long = shmem_malloc(max_elements_bytes);
targ_long = shmem_malloc(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_nb(targ_long,srce_long,max_elements,my_pe+1,NULL);
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);
}
shmem_free(srce_long); shmem_free(targ_long);
#endif
#ifdef NEEDS_FINALIZE
shmem_finalize();
#endif
return 0;
}
int main(int argc, char **argv)
{
int i,j;
int my_pe,n_pes,PE_root;
size_t max_elements,max_elements_bytes;
int *srce_int,*targ_int,ans_int;
long *srce_long,*targ_long,ans_long;
float *srce_float,*targ_float,ans_float;
double *srce_double,*targ_double,ans_double;
shmem_init();
my_pe = shmem_my_pe();
n_pes = shmem_n_pes();
/* fail if trying to use only one processor */
if ( n_pes <= 1 ){
fprintf(stderr, "FAIL - test requires at least two PEs\n");
exit(1);
}
if(my_pe == 0)
fprintf(stderr, "shmem_broadcast(%s) n_pes=%d\n", argv[0],n_pes);
/* initialize the pSync arrays */
for (i=0; i < _SHMEM_BCAST_SYNC_SIZE; i++) {
pSync1[i] = _SHMEM_SYNC_VALUE;
pSync2[i] = _SHMEM_SYNC_VALUE;
}
shmem_barrier_all(); /* Wait for all PEs to initialize pSync1 & pSync2 */
PE_root=1; /* we'll broadcast from this PE */
/* shmem_broadcast32 test */
max_elements = (size_t) (MAX_SIZE / sizeof(int));
max_elements_bytes = (size_t) (sizeof(int)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_broadcast32 max_elements = %d\n",
max_elements);
srce_int = shmem_malloc(max_elements_bytes);
targ_int = shmem_malloc(max_elements_bytes);
srce_float = shmem_malloc(max_elements_bytes);
targ_float = shmem_malloc(max_elements_bytes);
if((srce_int == NULL) || (targ_int == NULL) ||
(srce_float == NULL) || (targ_float == NULL))
shmalloc_error();
for(j = 0; j < max_elements; j++) {
srce_int[j] = (int)(my_pe+j);
srce_float[j] = (float)(my_pe+j);
targ_int[j] = (int)(100*my_pe+j);
targ_float[j] = (float)(100*my_pe+j);
}
shmem_barrier_all();
for(i = 0; i < IMAX; i+=2) {
/* i is even -- using int */
if (my_pe == PE_root)
for(j = 0; j < max_elements; j++) {
srce_int[j] = (int)(my_pe+i+j);
}
/* broadcast from PE_root to all PEs using pSync1 */
shmem_broadcast32(targ_int,srce_int,max_elements,PE_root,0,0,n_pes,pSync1);
for(j = 0; j < max_elements; j++) {
if (my_pe == PE_root) {
ans_int= (int)(100*my_pe+j);
} else {
ans_int= (int)(PE_root+i+j);
}
if ( targ_int[j] != ans_int )
fprintf(stderr, "FAIL: PE [%d] targ_int[%d]=%d ans_int=%d\n",
my_pe,j,targ_int[j],ans_int);
}
/* i+1 is odd -- using float */
if (my_pe == PE_root)
for(j = 0; j < max_elements; j++) {
srce_float[j] = (float)(PE_root+i+1+j);
}
/* broadcast from PE_root to all PEs using pSync2 */
shmem_broadcast32(targ_float,srce_float,max_elements,PE_root,0,0,n_pes,pSync2);
for(j = 0; j < max_elements; j++) {
if (my_pe == PE_root) {
ans_float= (float)(100*my_pe+j);
} else {
ans_float= (float)(PE_root+i+1+j);
}
if ( targ_float[j] != ans_float )
fprintf(stderr, "FAIL: PE [%d] targ_float[%d]=%10.0f ans_float=%10.0f\n",
my_pe,j,targ_float[j],ans_float);
}
}
shmem_free(srce_int); shmem_free(targ_int);
shmem_free(srce_float); shmem_free(targ_float);
/* shmem_broadcast64 test */
max_elements = (size_t) (MAX_SIZE / sizeof(long));
max_elements_bytes = (size_t) (sizeof(long)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_broadcast64 max_elements = %d\n",
max_elements);
srce_long = shmem_malloc(max_elements_bytes);
targ_long = shmem_malloc(max_elements_bytes);
srce_double = shmem_malloc(max_elements_bytes);
targ_double = shmem_malloc(max_elements_bytes);
if((srce_long == NULL) || (targ_long == NULL) ||
(srce_double == NULL) || (targ_double == NULL))
shmalloc_error();
for(j = 0; j < max_elements; j++) {
srce_long[j] = (long)(my_pe+j);
srce_double[j] = (double)(my_pe+j);
targ_long[j] = (long)(100*my_pe+j);
targ_double[j] = (double)(100*my_pe+j);
}
shmem_barrier_all();
for(i = 0; i < IMAX; i+=2) {
/* i is even -- using long */
if (my_pe == PE_root)
for(j = 0; j < max_elements; j++) {
srce_long[j] = (long)(my_pe+i+j);
}
/* broadcast from PE_root to all PEs using pSync1 */
shmem_broadcast64(targ_long,srce_long,max_elements,PE_root,0,0,n_pes,pSync1);
for(j = 0; j < max_elements; j++) {
if (my_pe == PE_root) {
ans_long= (long)(100*my_pe+j);
} else {
ans_long= (long)(PE_root+i+j);
}
if ( targ_long[j] != ans_long )
fprintf(stderr, "FAIL: PE [%d] targ_long[%d]=%d ans_long=%d\n",
my_pe,j,targ_long[j],ans_long);
}
/* i+1 is odd -- using double */
if (my_pe == PE_root)
for(j = 0; j < max_elements; j++) {
srce_double[j] = (double)(PE_root+i+1+j);
}
/* broadcast from PE_root to all PEs using pSync2 */
shmem_broadcast64(targ_double,srce_double,max_elements,PE_root,0,0,n_pes,pSync2);
for(j = 0; j < max_elements; j++) {
if (my_pe == PE_root) {
ans_double= (double)(100*my_pe+j);
} else {
ans_double= (double)(PE_root+i+1+j);
}
if ( targ_double[j] != ans_double )
fprintf(stderr, "FAIL: PE [%d] targ_double[%d]=%10.0f ans_double=%10.0f\n",
my_pe,j,targ_double[j],ans_double);
}
}
shmem_free(srce_long); shmem_free(targ_long);
shmem_free(srce_double); shmem_free(targ_double);
#ifndef OPENSHMEM
#ifdef SHMEM_C_GENERIC_32
/* shmemx_broadcast (GENERIC 32) test */
max_elements = (size_t) (MAX_SIZE / sizeof(int));
max_elements_bytes = (size_t) (sizeof(int)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmemx_broadcast (GENERIC 32) max_elements = %d\n",
max_elements);
srce_int = shmem_malloc(max_elements_bytes);
targ_int = shmem_malloc(max_elements_bytes);
if((srce_int == NULL) || (targ_int == NULL))
shmalloc_error();
for(j = 0; j < max_elements; j++) {
srce_int[j] = (int)(my_pe+j);
targ_int[j] = (int)(2*my_pe+j);
}
shmem_barrier_all();
/* broadcast from PE 1 to all PEs */
shmemx_broadcast(targ_int,srce_int,max_elements,1,0,0,n_pes,pSync1);
for(j = 0; j < max_elements; j++) {
if (my_pe == 1) {
ans_int= (int)(j+2);
} else {
ans_int= (int)(j+1);
}
if ( targ_int[j] != ans_int )
fprintf(stderr, "FAIL: PE [%d] targ_int[%d]=%d ans_int=%d\n",
my_pe,j,targ_int[j],ans_int);
}
shmem_free(srce_int); shmem_free(targ_int);
#else
/* shmemx_broadcast (GENERIC 64) test */
max_elements = (size_t) (MAX_SIZE / sizeof(long));
max_elements_bytes = (size_t) (sizeof(long)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmemx_broadcast (GENERIC 64) max_elements = %d\n",
max_elements);
srce_long = shmem_malloc(max_elements_bytes);
targ_long = shmem_malloc(max_elements_bytes);
if((srce_long == NULL) || (targ_long == NULL))
shmalloc_error();
for(j = 0; j < max_elements; j++) {
srce_long[j] = (long)(my_pe+j);
targ_long[j] = (long)(2*my_pe+j);
}
shmem_barrier_all();
/* broadcast from PE 1 to all PEs */
shmemx_broadcast(targ_long,srce_long,max_elements,1,0,0,n_pes,pSync1);
for(j = 0; j < max_elements; j++) {
if (my_pe == 1) {
ans_long = (long)(j+2);
} else {
ans_long = (long)(j+1);
}
if ( targ_long[j] != ans_long )
fprintf(stderr, "FAIL: PE [%d] targ_long[%d]=%d ans_long=%d\n",
my_pe,j,targ_long[j],ans_long);
}
shmem_free(srce_long); shmem_free(targ_long);
#endif
#endif
#ifdef NEEDS_FINALIZE
shmem_finalize();
#endif
return 0;
}
int main(int argc, char **argv)
{
int i,j,iter;
int my_pe,n_pes;
int *flag,*one;
size_t max_elements,max_elements_bytes;
size_t elements[16] = {1,2,4,8,12,16,24,32,64,128,256,512,1024,2048,4096,8192};
int num_elements = 16;
short *srce_short,*targ_short;
int *srce_int,*targ_int;
long *srce_long,*targ_long;
float *srce_float,*targ_float;
double *srce_double,*targ_double;
shmem_init();
my_pe = shmem_my_pe();
n_pes = shmem_n_pes();
flag = shmem_malloc((size_t) sizeof(int));
one = shmem_malloc((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_both_put_nb_size(%s)\n", argv[0]);
/* alloc arrays */
max_elements = (size_t) (MAX_SIZE / sizeof(int));
max_elements_bytes = (size_t) (sizeof(int)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_int_put_nb max_elements = %d\n",max_elements);
srce_int = shmem_malloc(max_elements_bytes);
targ_int = shmem_malloc(max_elements_bytes);
if((srce_int == NULL) || (targ_int == NULL))
shmalloc_error();
max_elements = (size_t) (MAX_SIZE / sizeof(short));
max_elements_bytes = (size_t) (sizeof(short)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_short_put max_elements = %d\n",max_elements);
srce_short = shmem_malloc(max_elements_bytes);
targ_short = shmem_malloc(max_elements_bytes);
if((srce_short == NULL) || (targ_short == NULL))
shmalloc_error();
max_elements = (size_t) (MAX_SIZE / sizeof(long));
max_elements_bytes = (size_t) (sizeof(long)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_long_put_nb max_elements = %d\n",max_elements);
srce_long = shmem_malloc(max_elements_bytes);
targ_long = shmem_malloc(max_elements_bytes);
if((srce_long == NULL) || (targ_long == NULL))
shmalloc_error();
max_elements = (size_t) (MAX_SIZE / sizeof(float));
max_elements_bytes = (size_t) (sizeof(float)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_float_put_nb max_elements = %d\n",max_elements);
srce_float = shmem_malloc(max_elements_bytes);
targ_float = shmem_malloc(max_elements_bytes);
if((srce_float == NULL) || (targ_float == NULL))
shmalloc_error();
max_elements = (size_t) (MAX_SIZE / sizeof(double));
max_elements_bytes = (size_t) (sizeof(double)*max_elements);
if(my_pe == 0)
fprintf(stderr,"shmem_double_put_nb max_elements = %d\n",max_elements);
srce_double = shmem_malloc(max_elements_bytes);
targ_double = shmem_malloc(max_elements_bytes);
if((srce_double == NULL) || (targ_double == NULL))
shmalloc_error();
if(my_pe == 0)
fprintf(stderr,"Actual value used for max_elements = %d\n",max_elements);
/* try the different sizes MAX_ITER times */
for (iter = 0; iter < MAX_ITER; iter++) {
for (i = 0; i < num_elements; i++) {
*flag = 0;
if (elements[i] <= max_elements) {
if ( (my_pe % 2) == 0 )
for(j = 0; j < elements[i]; j++) {
srce_short[j] = (short)(my_pe+j);
srce_int[j] = (int)(iter*10000+elements[i]*100+my_pe+j);
srce_long[j] = (long)(iter*10000+elements[i]*100+my_pe+j);
srce_float[j] = (float)(iter*10000+elements[i]*100+my_pe+j);
srce_double[j] = (double)(iter*10000+elements[i]*100+my_pe+j);
}
else
for(j = 0; j < elements[i]; j++) {
targ_short[j] = (short)(my_pe+j);
targ_int[j] = (int)(iter*10000+elements[i]*100+my_pe+j);
targ_long[j] = (long)(iter*10000+elements[i]*100+my_pe+j);
targ_float[j] = (float)(iter*10000+elements[i]*100+my_pe+j);
targ_double[j] = (double)(iter*10000+elements[i]*100+my_pe+j);
}
shmem_barrier_all();
if ( (my_pe % 2) == 0 ) {
#ifndef OPENSHMEM
shmemx_int_put_nb(targ_int,srce_int,elements[i],my_pe+1,NULL);
shmemx_long_put_nb(targ_long,srce_long,elements[i],my_pe+1,NULL);
shmemx_float_put_nb(targ_float,srce_float,elements[i],my_pe+1,NULL);
shmemx_double_put_nb(targ_double,srce_double,elements[i],my_pe+1,NULL);
#else
shmem_int_put_nbi(targ_int,srce_int,elements[i],my_pe+1);
shmem_long_put_nbi(targ_long,srce_long,elements[i],my_pe+1);
shmem_float_put_nbi(targ_float,srce_float,elements[i],my_pe+1);
shmem_double_put_nbi(targ_double,srce_double,elements[i],my_pe+1);
#endif
/* this one is blocking */
shmem_short_put(targ_short,srce_short,elements[i],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 < elements[i]; j++) {
if ( targ_short[j] != (short)(my_pe+j-1) )
fprintf(stderr,
"FAIL: PE [%d] iter=%d i=%d targ_short[%d]=%d not equal %d\n",
my_pe,iter,i,j,targ_short[j],my_pe+j-1);
if ( targ_int[j] != (int)(iter*10000+elements[i]*100+my_pe+j-1) )
fprintf(stderr,
"FAIL: PE [%d] iter=%d i=%d targ_int[%d]=%d not equal %d\n",
my_pe,iter,i,j,targ_int[j],iter*10000+elements[i]*100+my_pe+j-1);
if ( targ_long[j] != (long)(iter*10000+elements[i]*100+my_pe+j-1) )
fprintf(stderr,
"FAIL: PE [%d] iter=%d i=%d targ_long[%d]=%d not equal %d\n",
my_pe,iter,i,j,targ_long[j],iter*10000+elements[i]*100+my_pe+j-1);
if ( targ_float[j] != (float)(iter*10000+elements[i]*100+my_pe+j-1) )
fprintf(stderr,
"FAIL: PE [%d] iter=%d i=%d targ_long[%d]=%f not equal %d\n",
my_pe,iter,i,j,targ_float[j],iter*10000+elements[i]*100+my_pe+j-1);
if ( targ_double[j] != (double)(iter*10000+elements[i]*100+my_pe+j-1) )
fprintf(stderr,
"FAIL: PE [%d] iter=%d i=%d targ_double[%d]=%f not equal %d\n",
my_pe,iter,i,j,targ_double[j],iter*10000+elements[i]*100+my_pe+j-1);
}
}
}
}
}
shmem_free(srce_short); shmem_free(targ_short);
shmem_free(srce_int); shmem_free(targ_int);
shmem_free(srce_long); shmem_free(targ_long);
shmem_free(srce_float); shmem_free(targ_float);
shmem_free(srce_double); shmem_free(targ_double);
#ifdef NEEDS_FINALIZE
shmem_finalize();
#endif
return 0;
}
