Aquila-Consortium/libsharp2
0
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

introduce theta chunking for MPI jobs

Browse source
This commit is contained in:
Martin Reinecke 2013-01-16 12:58:29 +01:00
parent 842b790b1f
commit 2353da0042
1 changed files with 69 additions and 45 deletions
Download patch file Download diff file Expand all files Collapse all files

114
libsharp/sharp_mpi.c
View file

@ -216,72 +216,96 @@ static void sharp_execute_job_mpi (sharp_job *job, MPI_Comm comm)
MPI_Barrier(comm); MPI_Barrier(comm);
double timer=wallTime(); double timer=wallTime();
int lmax = job->ainfo->lmax; job->opcnt=0;
job->norm_l = sharp_Ylmgen_get_norm (lmax, job->spin);
sharp_mpi_info minfo; sharp_mpi_info minfo;
sharp_make_mpi_info(comm, job, &minfo); sharp_make_mpi_info(comm, job, &minfo);
/* clear output arrays if requested */ if (minfo.npairtotal>minfo.ntasks*300)
init_output (job);
alloc_phase_mpi (job,job->ainfo->nm,job->ginfo->npairs,minfo.mmax+1,
minfo.npairtotal);
double *cth = RALLOC(double,minfo.npairtotal),
*sth = RALLOC(double,minfo.npairtotal);
int *mlim = RALLOC(int,minfo.npairtotal);
for (int i=0; i<minfo.npairtotal; ++i)
{ {
cth[i] = cos(minfo.theta[i]); int nsub=(minfo.npairtotal+minfo.ntasks*200-1)/(minfo.ntasks*200);
sth[i] = sin(minfo.theta[i]); for (int isub=0; isub<nsub; ++isub)
mlim[i] = sharp_get_mlim(lmax, job->spin, sth[i], cth[i], 100.); {
sharp_job ljob=*job;
if ((isub>0)&&(job->type==SHARP_MAP2ALM)) ljob.flags|=SHARP_ADD;
sharp_geom_info lginfo;
lginfo.pair=RALLOC(sharp_ringpair,(job->ginfo->npairs/nsub)+1);
lginfo.npairs=0;
while (lginfo.npairs*nsub+isub<job->ginfo->npairs)
{
lginfo.pair[lginfo.npairs]=job->ginfo->pair[lginfo.npairs*nsub+isub];
++lginfo.npairs;
}
ljob.ginfo=&lginfo;
sharp_execute_job_mpi (&ljob,comm);
job->opcnt+=ljob.opcnt;
DEALLOC(lginfo.pair);
}
} }
else
{
int lmax = job->ainfo->lmax;
job->norm_l = sharp_Ylmgen_get_norm (lmax, job->spin);
/* map->phase where necessary */ /* clear output arrays if requested */
map2phase (job, minfo.mmax, 0, job->ginfo->npairs); init_output (job);
map2alm_comm (job, &minfo); alloc_phase_mpi (job,job->ainfo->nm,job->ginfo->npairs,minfo.mmax+1,
minfo.npairtotal);
double *cth = RALLOC(double,minfo.npairtotal),
*sth = RALLOC(double,minfo.npairtotal);
int *mlim = RALLOC(int,minfo.npairtotal);
for (int i=0; i<minfo.npairtotal; ++i)
{
cth[i] = cos(minfo.theta[i]);
sth[i] = sin(minfo.theta[i]);
mlim[i] = sharp_get_mlim(lmax, job->spin, sth[i], cth[i], 100.);
}
/* map->phase where necessary */
map2phase (job, minfo.mmax, 0, job->ginfo->npairs);
map2alm_comm (job, &minfo);
#pragma omp parallel if ((job->flags&SHARP_NO_OPENMP)==0) #pragma omp parallel if ((job->flags&SHARP_NO_OPENMP)==0)
{ {
sharp_job ljob = *job; sharp_job ljob = *job;
sharp_Ylmgen_C generator; sharp_Ylmgen_C generator;
sharp_Ylmgen_init (&generator,lmax,minfo.mmax,ljob.spin); sharp_Ylmgen_init (&generator,lmax,minfo.mmax,ljob.spin);
alloc_almtmp(&ljob,lmax); alloc_almtmp(&ljob,lmax);
#pragma omp for schedule(dynamic,1) #pragma omp for schedule(dynamic,1)
for (int mi=0; mi<job->ainfo->nm; ++mi) for (int mi=0; mi<job->ainfo->nm; ++mi)
{ {
/* alm->alm_tmp where necessary */ /* alm->alm_tmp where necessary */
alm2almtmp (&ljob, lmax, mi); alm2almtmp (&ljob, lmax, mi);
/* inner conversion loop */ /* inner conversion loop */
inner_loop (&ljob, minfo.ispair, cth, sth, 0, minfo.npairtotal, inner_loop (&ljob, minfo.ispair, cth, sth, 0, minfo.npairtotal,
&generator, mi, mlim); &generator, mi, mlim);
/* alm_tmp->alm where necessary */ /* alm_tmp->alm where necessary */
almtmp2alm (&ljob, lmax, mi); almtmp2alm (&ljob, lmax, mi);
} }
sharp_Ylmgen_destroy(&generator); sharp_Ylmgen_destroy(&generator);
dealloc_almtmp(&ljob); dealloc_almtmp(&ljob);
#pragma omp critical #pragma omp critical
job->opcnt+=ljob.opcnt; job->opcnt+=ljob.opcnt;
} /* end of parallel region */ } /* end of parallel region */
alm2map_comm (job, &minfo); alm2map_comm (job, &minfo);
/* phase->map where necessary */ /* phase->map where necessary */
phase2map (job, minfo.mmax, 0, job->ginfo->npairs); phase2map (job, minfo.mmax, 0, job->ginfo->npairs);
DEALLOC(mlim); DEALLOC(mlim);
DEALLOC(cth); DEALLOC(cth);
DEALLOC(sth); DEALLOC(sth);
DEALLOC(job->norm_l); DEALLOC(job->norm_l);
dealloc_phase (job); dealloc_phase (job);
}
sharp_destroy_mpi_info(&minfo); sharp_destroy_mpi_info(&minfo);
job->time=wallTime()-timer; job->time=wallTime()-timer;
} }