Skip to content

Commit

Permalink
copied test_Dslash.c to test_DslashBSM.c.
Browse files Browse the repository at this point in the history
  • Loading branch information
@m-schroeck
m-schroeck committed Nov 14, 2014
1 parent cf0e947 commit 2937f79
Show file tree
Hide file tree
Showing 3 changed files with 322 additions and 10 deletions.
8 changes: 4 additions & 4 deletions Makefile.in
View file
Original file line number Diff line number Diff line change
Expand Up @@ -65,7 +65,7 @@ GPUOBJECTS := $(patsubst $(srcdir)/$(GPUDIR)/%.cu, $(GPUDIR)/%.o, $(GPUSOURCES))

NOOPTMOD = test/check_xchange test/check_geometry

PROGRAMS = hmc_tm benchmark invert gen_sources test_Dslash \
PROGRAMS = hmc_tm benchmark invert gen_sources test_DslashBSM \
check_locallity test_lemon hopping_test LapH_ev

ALLOBJ = ${MODULES} ${PROGRAMS} ${SMODULES}
Expand All @@ -74,7 +74,7 @@ SUBDIRS = ${USESUBDIRS}
# delete the default suffix rules
.SUFFIXES:

all: Makefile dep $(SUBDIRS) hmc_tm invert benchmark test_Dslash
all: Makefile dep $(SUBDIRS) hmc_tm invert benchmark test_DslashBSM

$(SUBDIRS):
$(MAKE) --directory=$@
Expand Down Expand Up @@ -144,10 +144,10 @@ compile-clean: compile-clean-recursive Makefile
rm -f *.o *.d test/*.o test/*.d tests/*.o tests/*.d

clean: clean-recursive Makefile
rm -f benchmark hmc_tm invert test_Dslash *.o *.d test/*.o test/*.d tests/*.o tests/*.d
rm -f benchmark hmc_tm invert test_DslashBSM *.o *.d test/*.o test/*.d tests/*.o tests/*.d

distclean: distclean-recursive Makefile
rm -f benchmark hmc_tm invert test_Dslash *.o *.d *~ Makefile config.log config.status fixed_volume.h
rm -f benchmark hmc_tm invert test_DslashBSM *.o *.d *~ Makefile config.log config.status fixed_volume.h
rm -f config.h

.PHONY: all ${SUBDIRS} ${top_srcdir}/git_hash.h clean compile-clean distclean dep install \
Expand Down
13 changes: 7 additions & 6 deletions test_Dslash.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -202,12 +202,12 @@ int main(int argc,char *argv[])
(int)(T*g_nproc_t), (int)(LX*g_nproc_x), (int)(LY*g_nproc_y), (int)(g_nproc_z*LZ));
printf("# The local lattice size is %d x %d x %d x %d\n",
(int)(T), (int)(LX), (int)(LY),(int) LZ);
if(even_odd_flag) {
printf("# benchmarking the even/odd preconditioned Dirac operator\n");
}
else {
printf("# benchmarking the standard Dirac operator\n");
}
// if(even_odd_flag) {
// printf("# benchmarking the even/odd preconditioned Dirac operator\n");
// }
// else {
// printf("# benchmarking the standard Dirac operator\n");
// }
fflush(stdout);
}

Expand Down Expand Up @@ -265,6 +265,7 @@ int main(int argc,char *argv[])
#endif
t1 = gettime();

/* here the actual Dslash */
D_psi(g_spinor_field[0], g_spinor_field[1]);

t2 = gettime();
Expand Down
311 changes: 311 additions & 0 deletions test_DslashBSM.c
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,311 @@
/*
* test_Dslash.c
*
* Created on: Nov 13, 2014
* Author: mario
*/

/*******************************************************************************
*
* test program for Dslash (D_psi)
*
*
*******************************************************************************/

#ifdef HAVE_CONFIG_H
# include<config.h>
#endif
#include <stdlib.h>
#include <stdio.h>
#include <math.h>
#include <time.h>
#include <string.h>
#if (defined BGL && !defined BGP)
# include <rts.h>
#endif
#ifdef MPI
# include <mpi.h>
# ifdef HAVE_LIBLEMON
# include <io/params.h>
# include <io/gauge.h>
# endif
#endif
#ifdef OMP
# include <omp.h>
# include "init/init_openmp.h"
#endif
#include "gettime.h"
#include "su3.h"
#include "su3adj.h"
#include "ranlxd.h"
#include "geometry_eo.h"
#include "read_input.h"
#include "start.h"
#include "boundary.h"
#include "operator/Hopping_Matrix.h"
#include "operator/Hopping_Matrix_nocom.h"
#include "operator/tm_operators.h"
#include "global.h"
#include "xchange/xchange.h"
#include "init/init.h"
#include "test/check_geometry.h"
#include "operator/D_psi.h"
//#include "phmc.h"
#include "mpi_init.h"

#ifdef PARALLELT
# define SLICE (LX*LY*LZ/2)
#elif defined PARALLELXT
# define SLICE ((LX*LY*LZ/2)+(T*LY*LZ/2))
#elif defined PARALLELXYT
# define SLICE ((LX*LY*LZ/2)+(T*LY*LZ/2) + (T*LX*LZ/2))
#elif defined PARALLELXYZT
# define SLICE ((LX*LY*LZ/2)+(T*LY*LZ/2) + (T*LX*LZ/2) + (T*LX*LY/2))
#elif defined PARALLELX
# define SLICE ((LY*LZ*T/2))
#elif defined PARALLELXY
# define SLICE ((LY*LZ*T/2) + (LX*LZ*T/2))
#elif defined PARALLELXYZ
# define SLICE ((LY*LZ*T/2) + (LX*LZ*T/2) + (LX*LY*T/2))
#endif

int check_xchange();

int main(int argc,char *argv[])
{
#ifdef _USE_HALFSPINOR
#undef _USE_HALFSPINOR
printf("# WARNING: USE_HALFSPINOR will be ignored (not supported here).\n");
#endif

if(even_odd_flag)
{
even_odd_flag=0;
printf("# WARNING: even_odd_flag will be ignored (not supported here).\n");
}
int j,j_max,k,k_max = 1;
#ifdef HAVE_LIBLEMON
paramsXlfInfo *xlfInfo;
#endif
int status = 0;

static double t1,t2,dt,sdt,dts,qdt,sqdt;
double antioptaway=0.0;

#ifdef MPI
static double dt2;

DUM_DERI = 6;
DUM_SOLVER = DUM_DERI+2;
DUM_MATRIX = DUM_SOLVER+6;
NO_OF_SPINORFIELDS = DUM_MATRIX+2;

# ifdef OMP
int mpi_thread_provided;
MPI_Init_thread(&argc, &argv, MPI_THREAD_SERIALIZED, &mpi_thread_provided);
# else
MPI_Init(&argc, &argv);
# endif
MPI_Comm_rank(MPI_COMM_WORLD, &g_proc_id);

#else
g_proc_id = 0;
#endif

g_rgi_C1 = 1.;

/* Read the input file */
if((status = read_input("test_Dslash.input")) != 0) {
fprintf(stderr, "Could not find input file: test_Dslash.input\nAborting...\n");
exit(-1);
}

#ifdef OMP
init_openmp();
#endif

tmlqcd_mpi_init(argc, argv);



if(g_proc_id==0) {
#ifdef SSE
printf("# The code was compiled with SSE instructions\n");
#endif
#ifdef SSE2
printf("# The code was compiled with SSE2 instructions\n");
#endif
#ifdef SSE3
printf("# The code was compiled with SSE3 instructions\n");
#endif
#ifdef P4
printf("# The code was compiled for Pentium4\n");
#endif
#ifdef OPTERON
printf("# The code was compiled for AMD Opteron\n");
#endif
#ifdef _GAUGE_COPY
printf("# The code was compiled with -D_GAUGE_COPY\n");
#endif
#ifdef BGL
printf("# The code was compiled for Blue Gene/L\n");
#endif
#ifdef BGP
printf("# The code was compiled for Blue Gene/P\n");
#endif
#ifdef _USE_HALFSPINOR
printf("# The code was compiled with -D_USE_HALFSPINOR\n");
#endif
#ifdef _USE_SHMEM
printf("# The code was compiled with -D_USE_SHMEM\n");
# ifdef _PERSISTENT
printf("# The code was compiled for persistent MPI calls (halfspinor only)\n");
# endif
#endif
#ifdef MPI
# ifdef _NON_BLOCKING
printf("# The code was compiled for non-blocking MPI calls (spinor and gauge)\n");
# endif
#endif
printf("\n");
fflush(stdout);
}


#ifdef _GAUGE_COPY
init_gauge_field(VOLUMEPLUSRAND + g_dbw2rand, 1);
#else
init_gauge_field(VOLUMEPLUSRAND + g_dbw2rand, 0);
#endif
init_geometry_indices(VOLUMEPLUSRAND + g_dbw2rand);

if(even_odd_flag) {
j = init_spinor_field(VOLUMEPLUSRAND/2, 2*k_max+1);
}
else {
j = init_spinor_field(VOLUMEPLUSRAND, 2*k_max);
}

if ( j!= 0) {
fprintf(stderr, "Not enough memory for spinor fields! Aborting...\n");
exit(0);
}
j = init_moment_field(VOLUME, VOLUMEPLUSRAND + g_dbw2rand);
if ( j!= 0) {
fprintf(stderr, "Not enough memory for moment fields! Aborting...\n");
exit(0);
}

if(g_proc_id == 0) {
fprintf(stdout,"# The number of processes is %d \n",g_nproc);
printf("# The lattice size is %d x %d x %d x %d\n",
(int)(T*g_nproc_t), (int)(LX*g_nproc_x), (int)(LY*g_nproc_y), (int)(g_nproc_z*LZ));
printf("# The local lattice size is %d x %d x %d x %d\n",
(int)(T), (int)(LX), (int)(LY),(int) LZ);
// if(even_odd_flag) {
// printf("# benchmarking the even/odd preconditioned Dirac operator\n");
// }
// else {
// printf("# benchmarking the standard Dirac operator\n");
// }
fflush(stdout);
}

/* define the geometry */
geometry();
/* define the boundary conditions for the fermion fields */
boundary(g_kappa);

#ifdef _USE_HALFSPINOR
j = init_dirac_halfspinor();
if ( j!= 0) {
fprintf(stderr, "Not enough memory for halfspinor fields! Aborting...\n");
exit(0);
}
if(g_sloppy_precision_flag == 1) {
g_sloppy_precision = 1;
j = init_dirac_halfspinor32();
if ( j!= 0) {
fprintf(stderr, "Not enough memory for 32-Bit halfspinor fields! Aborting...\n");
exit(0);
}
}
# if (defined _PERSISTENT)
init_xchange_halffield();
# endif
#endif

status = check_geometry();
if (status != 0) {
fprintf(stderr, "Checking of geometry failed. Unable to proceed.\nAborting....\n");
exit(1);
}
#if (defined MPI && !(defined _USE_SHMEM))
check_xchange();
#endif

start_ranlux(1, 123456);
random_gauge_field(reproduce_randomnumber_flag, g_gauge_field);

#ifdef MPI
/*For parallelization: exchange the gaugefield */
xchange_gauge(g_gauge_field);
#endif

/* the non even/odd case now */
/*initialize the pseudo-fermion fields*/
j_max=1;
sdt=0.;
for (k=0;k<k_max;k++) {
random_spinor_field_lexic(g_spinor_field[k], reproduce_randomnumber_flag, RN_GAUSS);
}

#ifdef MPI
MPI_Barrier(MPI_COMM_WORLD);
#endif
t1 = gettime();

/* here the actual Dslash */
D_psi(g_spinor_field[0], g_spinor_field[1]);

t2 = gettime();
dt=t2-t1;
#ifdef MPI
MPI_Allreduce (&dt, &sdt, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
#else
sdt = dt;
#endif

if(g_proc_id==0) {
printf("# Time for Dslash %e sec.\n", sdt);
printf("\n");
fflush(stdout);
}

#ifdef HAVE_LIBLEMON
if(g_proc_id==0) {
printf("# Performing parallel IO test ...\n");
}
xlfInfo = construct_paramsXlfInfo(0.5, 0);
write_gauge_field( "conf.test", 64, xlfInfo);
free(xlfInfo);
if(g_proc_id==0) {
printf("# done ...\n");
}
#endif


#ifdef OMP
free_omp_accumulators();
#endif
free_gauge_field();
free_geometry_indices();
free_spinor_field();
free_moment_field();
#ifdef MPI
MPI_Barrier(MPI_COMM_WORLD);
MPI_Finalize();
#endif
return(0);
}

0 comments on commit 2937f79

Please sign in to comment.