1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
|
# superlu_dist batch script for Perlmutter CPU-only compute nodes
# gnu compiler
# updated 2023/04/01
# please make sure the following module loads/unloads match your build script
module unload gpu
#module load PrgEnv-gnu
#module load gcc/11.2.0
#module load cmake/3.24.3
#module load cudatoolkit/11.7
if [[ $NERSC_HOST == edison ]]; then
CORES_PER_NODE=24
THREADS_PER_NODE=48
elif [[ $NERSC_HOST == cori ]]; then
CORES_PER_NODE=32
THREADS_PER_NODE=64
# This does not take hyperthreading into account
elif [[ $NERSC_HOST == perlmutter ]]; then
CORES_PER_NODE=128
THREADS_PER_NODE=256
else
# Host unknown; exiting
exit $EXIT_HOST
fi
# nprows=(1 2 4 8 8)
# npcols=(1 2 4 8 16)
nprows=(4)
npcols=(4)
NTH=1
NODE_VAL_TOT=1
for ((i = 0; i < ${#npcols[@]}; i++)); do
NROW=${nprows[i]}
NCOL=${npcols[i]}
CORE_VAL=`expr $NCOL \* $NROW`
NODE_VAL=`expr $CORE_VAL / $CORES_PER_NODE`
MOD_VAL=`expr $CORE_VAL % $CORES_PER_NODE`
if [[ $MOD_VAL -ne 0 ]]
then
NODE_VAL=`expr $NODE_VAL + 1`
fi
NCORE_VAL_TOT=`expr $NODE_VAL_TOT \* $CORES_PER_NODE / $NTH`
OMP_NUM_THREADS=$NTH
TH_PER_RANK=`expr $NTH \* 2`
export OMP_NUM_THREADS=$NTH
export OMP_PLACES=threads
export OMP_PROC_BIND=spread
export MPICH_MAX_THREAD_SAFETY=multiple
# export NSUP=5
# export NREL=5
# for MAT in big.rua
# for MAT in g4.rua
for MAT in s1_mat_0_126936.bin
# for MAT in s1_mat_0_507744.bin
# for MAT in StocF-1465.bin
# for MAT in s1_mat_0_126936.bin s1_mat_0_253872.bin s1_mat_0_507744.bin
# for MAT in matrix_ACTIVSg70k_AC_00.mtx matrix_ACTIVSg10k_AC_00.mtx
# for MAT in temp_13k.mtx temp_25k.mtx temp_75k.mtx
# for MAT in atmosmodj.bin StocF-1465.bin globalmat118_1536.bin
do
mkdir -p $MAT
echo "srun -n $NCORE_VAL_TOT -N $NODE_VAL -c $TH_PER_RANK --cpu_bind=cores ./EXAMPLE/pddrive -c $NCOL -r $NROW $CFS/m2957/liuyangz/my_research/matrix/$MAT | tee ./$MAT/SLU.o_mpi_${NROW}x${NCOL}_${NTH}_1rhs_2d"
srun -n $NCORE_VAL_TOT -N $NODE_VAL -c $TH_PER_RANK --cpu_bind=cores ./EXAMPLE/pddrive -c $NCOL -r $NROW $CFS/m2957/liuyangz/my_research/matrix/$MAT | tee ./$MAT/SLU.o_mpi_${NROW}x${NCOL}_${NTH}_1rhs_2d
# srun -n $NCORE_VAL_TOT -N $NODE_VAL -c $TH_PER_RANK --cpu_bind=cores ./EXAMPLE/pddrive3d -c $NCOL -r $NROW $CFS/m2957/liuyangz/my_research/matrix/$MAT | tee ./$MAT/SLU.o_mpi_${NROW}x${NCOL}_${NTH}_1rhs_3d
done
done
|
