From 470768662fbcf20229656bbe4b4066b7b399ef77 Mon Sep 17 00:00:00 2001
From: Abdullah Alperen <alperena@msu.edu>
Date: Sat, 15 Sep 2018 14:12:27 -0400
Subject: [PATCH] PuReMD: minor changes
---
PuReMD/src/linear_solvers.c | 56 ++++++++++++++++---------------------
PuReMD/src/qEq.c | 2 +-
PuReMD/src/reax_types.h | 6 ++--
3 files changed, 28 insertions(+), 36 deletions(-)
diff --git a/PuReMD/src/linear_solvers.c b/PuReMD/src/linear_solvers.c
index 0bd9faeb..83ddaf8a 100644
--- a/PuReMD/src/linear_solvers.c
+++ b/PuReMD/src/linear_solvers.c
@@ -106,8 +106,8 @@ void setup_sparse_approx_inverse( reax_system *system, storage *workspace,
mpi_datatypes *mpi_data, sparse_matrix *A, sparse_matrix **A_spar_patt,
int nprocs, real filter )
{
- int i, bin, total, pos, push;
- int n, n_gather, m, s_local, s, n_local;
+ int i, bin, total, pos;
+ int n, n_gather, s_local, s, n_local;
int target_proc;
int k;
int pj, size;
@@ -118,7 +118,6 @@ void setup_sparse_approx_inverse( reax_system *system, storage *workspace,
real *samplelist_local, *samplelist;
real *pivotlist;
real *bucketlist_local, *bucketlist;
- real *local_entries;
int *srecv, *sdispls;
int *scounts_local, *scounts;
@@ -140,7 +139,6 @@ void setup_sparse_approx_inverse( reax_system *system, storage *workspace,
dspls_local = NULL;
dspls = NULL;
bin_elements = NULL;
- local_entries = NULL;
comm = mpi_data->world;
@@ -155,15 +153,13 @@ void setup_sparse_approx_inverse( reax_system *system, storage *workspace,
Allocate_Matrix2( A_spar_patt, A->n, system->local_cap, A->m, comm );
}
- m = 0;
+ n_local = 0;
for( i = 0; i < A->n; ++i )
{
- m += A->end[i] - A->start[i];
+ n_local += A->end[i] - A->start[i];
}
- /* the sample ratio is 10% */
- /*n_local = m/10; */
- n_local = m;
s_local = (int) (12.0 * log2(n_local*nprocs));
+
MPI_Allreduce(&n_local, &n, 1, MPI_INT, MPI_SUM, comm);
MPI_Reduce(&s_local, &s, 1, MPI_INT, MPI_SUM, MASTER_NODE, comm);
@@ -177,7 +173,6 @@ void setup_sparse_approx_inverse( reax_system *system, storage *workspace,
dspls = malloc( sizeof(int) * nprocs );
pivotlist = malloc( sizeof(real) * (nprocs - 1) );
samplelist_local = malloc( sizeof(real) * s_local );
- local_entries = malloc ( sizeof(real) * m );
if ( system->my_rank == MASTER_NODE )
{
samplelist = malloc( sizeof(real) * s );
@@ -185,22 +180,15 @@ void setup_sparse_approx_inverse( reax_system *system, storage *workspace,
sdispls = malloc( sizeof(int) * nprocs );
}
- push = 0;
+ n_local = 0;
for( i = 0; i < A->n; ++i )
{
for( pj = A->start[i]; pj < A->end[i]; ++pj )
{
- local_entries[push++] = A->entries[pj].val;
+ input_array[n_local++] = A->entries[pj].val;
}
}
- srand( time(NULL) + system->my_rank );
- for ( i = 0; i < n_local ; i++ )
- {
- /* input_array[i] = local_entries[rand( ) % m]; */
- input_array[i] = local_entries[ i ];
- }
-
for ( i = 0; i < s_local; i++)
{
/* samplelist_local[i] = input_array[rand( ) % n_local]; */
@@ -376,8 +364,9 @@ void setup_sparse_approx_inverse( reax_system *system, storage *workspace,
/*broadcast the filtering value*/
MPI_Bcast( &threshold, 1, MPI_DOUBLE, target_proc, comm );
- int nnz = 0;
- /*build entries of that pattern*/
+ // int nnz = 0; uncomment to check the nnz's in the sparsity pattern
+
+ /* build entries of that pattern*/
for ( i = 0; i < A->n; ++i )
{
(*A_spar_patt)->start[i] = A->start[i];
@@ -390,18 +379,18 @@ void setup_sparse_approx_inverse( reax_system *system, storage *workspace,
(*A_spar_patt)->entries[size].val = A->entries[pj].val;
(*A_spar_patt)->entries[size].j = A->entries[pj].j;
size++;
- nnz++;
+ //nnz++;
}
}
(*A_spar_patt)->end[i] = size;
}
- MPI_Allreduce( MPI_IN_PLACE, &nnz, 1, MPI_INT, MPI_SUM, comm );
+ /*MPI_Allreduce( MPI_IN_PLACE, &nnz, 1, MPI_INT, MPI_SUM, comm );
if( system->my_rank == MASTER_NODE )
{
fprintf(stdout,"total nnz in all sparsity patterns = %d\nthreshold = %.15lf\n", nnz, threshold);
fflush(stdout);
- }
+ }*/
}
@@ -500,7 +489,7 @@ void sparse_approx_inverse(reax_system *system, storage *workspace, mpi_datatype
cnt = 0;
for( i = nbr1->atoms_str; i < (nbr1->atoms_str + nbr1->atoms_cnt); ++i )
{
- if( i >= A->n)
+ //if( i >= A->n)
{
cnt += row_nnz[i];
}
@@ -518,13 +507,13 @@ void sparse_approx_inverse(reax_system *system, storage *workspace, mpi_datatype
}
nbr2 = &system->my_nbrs[2 * d + 1];
- if ( nbr1->atoms_cnt )
+ if ( nbr2->atoms_cnt )
{
/* calculate the total data that will be received */
cnt = 0;
for( i = nbr2->atoms_str; i < (nbr2->atoms_str + nbr2->atoms_cnt); ++i )
{
- if( i >= A->n )
+ //if( i >= A->n )
{
cnt += row_nnz[i];
}
@@ -635,7 +624,7 @@ void sparse_approx_inverse(reax_system *system, storage *workspace, mpi_datatype
cnt = 0;
for( i = nbr1->atoms_str; i < (nbr1->atoms_str + nbr1->atoms_cnt); ++i )
{
- if( i >= A->n )
+ //if( i >= A->n )
{
j_list[i] = (int *) malloc( sizeof(int) * row_nnz[i] );
val_list[i] = (real *) malloc( sizeof(real) * row_nnz[i] );
@@ -659,7 +648,7 @@ void sparse_approx_inverse(reax_system *system, storage *workspace, mpi_datatype
cnt = 0;
for( i = nbr2->atoms_str; i < (nbr2->atoms_str + nbr2->atoms_cnt); ++i )
{
- if( i >= A->n )
+ //if( i >= A->n )
{
j_list[i] = (int *) malloc( sizeof(int) * row_nnz[i] );
val_list[i] = (real *) malloc( sizeof(real) * row_nnz[i] );
@@ -674,7 +663,7 @@ void sparse_approx_inverse(reax_system *system, storage *workspace, mpi_datatype
}
}
}
-
+
X = (int *) malloc( sizeof(int) * (system->bigN + 1) );
pos_x = (int *) malloc( sizeof(int) * (system->bigN + 1) );
@@ -1143,6 +1132,9 @@ int CG( reax_system *system, storage *workspace, sparse_matrix *H, real *b,
int i, scale;
real tmp, alpha, beta, b_norm;
real sig_old, sig_new;
+#if defined(HALF_LIST)
+ int j;
+#endif
#if defined(CG_PERFORMANCE)
if ( system->my_rank == MASTER_NODE )
@@ -1169,7 +1161,7 @@ int CG( reax_system *system, storage *workspace, sparse_matrix *H, real *b,
Vector_Sum( workspace->r , 1., b, -1., workspace->q, system->n );
/* pre-conditioning */
-#if defined(SAI_PRECONDITIONER)
+#if defined(SAI_PRECONDITIONER) && !defined(HALF_LIST)
Dist( system, mpi_data, workspace->r, MPI_DOUBLE, scale, real_packer );
Sparse_MatVec( workspace->H_app_inv, workspace->r, workspace->d, system->n );
#else
@@ -1210,7 +1202,7 @@ int CG( reax_system *system, storage *workspace, sparse_matrix *H, real *b,
Vector_Add( workspace->r, -alpha, workspace->q, system->n );
/* pre-conditioning */
-#if defined(SAI_PRECONDITIONER)
+#if defined(SAI_PRECONDITIONER) && !defined(HALF_LIST)
Dist( system, mpi_data, workspace->r, MPI_DOUBLE, scale, real_packer );
Sparse_MatVec( workspace->H_app_inv, workspace->r, workspace->p, system->n );
#else
diff --git a/PuReMD/src/qEq.c b/PuReMD/src/qEq.c
index 87b45bdd..9c5650f3 100644
--- a/PuReMD/src/qEq.c
+++ b/PuReMD/src/qEq.c
@@ -405,7 +405,7 @@ void QEq( reax_system *system, control_params *control, simulation_data *data,
// control->q_err, workspace->x, mpi_data, out_control->log);
//t_matvecs = 0;
-#if defined(SAI_PRECONDITIONER)
+#if defined(SAI_PRECONDITIONER) && !defined(HALF_LIST)
if( control->refactor > 0 && ((data->step - data->prev_steps) % control->refactor == 0))
{
Setup_Preconditioner_QEq( system, control, data, workspace, mpi_data );
diff --git a/PuReMD/src/reax_types.h b/PuReMD/src/reax_types.h
index 58ac8498..18cc8c49 100644
--- a/PuReMD/src/reax_types.h
+++ b/PuReMD/src/reax_types.h
@@ -41,13 +41,13 @@
#define PURE_REAX
//#define LAMMPS_REAX
-//#define SAI_PRECONDITIONER
-#define HALF_LIST
+#define SAI_PRECONDITIONER
+//#define HALF_LIST
//#define DEBUG
//#define DEBUG_FOCUS
//#define TEST_ENERGY
//#define TEST_FORCES
-//#define CG_PERFORMANCE
+#define CG_PERFORMANCE
#define LOG_PERFORMANCE
#define STANDARD_BOUNDARIES
//#define OLD_BOUNDARIES
--
GitLab