File: test_matops.sh

package info (click to toggle)
libxsmm 1.17-4
  • links: PTS, VCS
  • area: main
  • in suites: sid, trixie
  • size: 14,976 kB
  • sloc: ansic: 119,587; cpp: 27,680; fortran: 9,179; sh: 5,765; makefile: 5,040; pascal: 2,312; python: 1,812; f90: 1,773
file content (138 lines) | stat: -rwxr-xr-x 3,649 bytes parent folder | download | duplicates (2)
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
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
#!/usr/bin/env bash
###############################################################################
# Copyright (c) Intel Corporation - All rights reserved.                      #
# This file is part of the LIBXSMM library.                                   #
#                                                                             #
# For information on the license, see the LICENSE file.                       #
# Further information: https://github.com/hfp/libxsmm/                        #
# SPDX-License-Identifier: BSD-3-Clause                                       #
###############################################################################

# sde can be downloaded here
SDE64_BIN=/swtools/sde/kits/latest/sde64
SDE64_ARCH="-skx"
SDE64_FLAGS="-ptr_check -null_check -ptr_raise"
SDE=${SDE64_BIN}" "${SDE64_FLAGS}" "${SDE64_ARCH}" -- "
GREP=$(command -v grep)

# iterations, order, precision, arch  and format
if [ $# -eq 5 ]
then
  REPS=$1
  PDEG=$2
  PREC=$3
  VLEN=$4
  FRMT=$5
else
  REPS=1000
  PDEG=5
  PREC=f64
  VLEN=64
  FRMT=csc
fi

if [[ $PDEG == "1" ]]
then
  K=4
  N=3
elif [[ $PDEG == "2" ]]
then
  K=10
  N=6
elif [[ $PDEG == "3" ]]
then
  K=20
  N=10
elif [[ $PDEG == "4" ]]
then
  K=35
  N=15
elif [[ $PDEG == "5" ]]
then
  K=56
  N=21
elif [[ $PDEG == "6" ]]
then
  K=84
  N=28
else
  echo "PDEG need to be in the range of 1 to 6"
  return -1
fi

if [[ $PREC == "f32" ]]
then
  if [[ $VLEN == "32" ]]
  then
    CRUN=8
  elif [[ $VLEN == "64" ]]
  then
    CRUN=16
  elif [[ $VLEN == "16" ]]
  then
    CRUN=4
  else
    echo "VLEN need to be either 16/32/64"
    return -3
  fi
elif [[ $PREC == "f64" ]]
then
  if [[ $VLEN == "32" ]]
  then
    CRUN=4
  elif [[ $VLEN == "64" ]]
  then
    CRUN=8
  elif [[ $VLEN == "16" ]]
  then
    CRUN=2
  else
    echo "VLEN need to be either 16/32/64"
    return -3
  fi
else
  echo "PREC needs to be either f32/f64"
  return -2
fi

if [[ $VLEN == "64" ]]
then
  #on an AVX512 platform we can run natively
  CPUFLAGS=$(if [ "${GREP}" ] && [ -e /proc/cpuinfo ]; then ${GREP} -m1 flags /proc/cpuinfo | cut -d: -f2-; fi)
  if [ "$(echo "${CPUFLAGS}" | ${GREP} -o avx512f)" ]; then
    SDE=
  fi
  if [ "$(echo "${CPUFLAGS}" | ${GREP} -o asimd)" ]; then
    SDE=
  fi
else
  SDE=
fi

# number of quantities is always 9
M=9

if [[ $FRMT == "csr" ]]
then
  # test flux matrices, CSR
  for i in `ls mats/tet4_${PDEG}_fluxN*_csr.mtx`; do ${SDE} ./bsparse_packed_csr_${PREC} ${M} ${N} ${K} ${CRUN} ${REPS} $i; done
  for i in `ls mats/tet4_${PDEG}_fluxT*_csr.mtx`; do ${SDE} ./bsparse_packed_csr_${PREC} ${M} ${K} ${N} ${CRUN} ${REPS} $i; done
  # test stiffness matrices, CSR
  for i in `ls mats/tet4_${PDEG}_stiff*_csr.mtx`; do ${SDE} ./bsparse_packed_csr_${PREC} ${M} ${K} ${K} ${CRUN} ${REPS} $i; done
elif [[ $FRMT == "csc" ]]
then
  # test flux matrices, CSC
  for i in `ls mats/tet4_${PDEG}_fluxN*_csc.mtx`; do ${SDE} ./bsparse_packed_csc_${PREC} ${M} ${N} ${K} ${CRUN} ${REPS} $i; done
  for i in `ls mats/tet4_${PDEG}_fluxT*_csc.mtx`; do ${SDE} ./bsparse_packed_csc_${PREC} ${M} ${K} ${N} ${CRUN} ${REPS} $i; done
  # test stiffness matrices, CSC
  for i in `ls mats/tet4_${PDEG}_stiff*_csc.mtx`; do ${SDE} ./bsparse_packed_csc_${PREC} ${M} ${K} ${K} ${CRUN} ${REPS} $i; done
else
  echo "FRMT need to be either csr/csc"
  return -4
fi
# test star matrices
${SDE} ./asparse_packed_csr_${PREC} ${M} ${K} ${M} ${CRUN} ${REPS} mats/tet4_starMatrix_csr.mtx
# test flux matrices
${SDE} ./asparse_packed_csr_${PREC} ${M} ${K} ${M} ${CRUN} ${REPS} mats/tet4_fluxMatrix_csr_sp.mtx
${SDE} ./asparse_packed_csr_${PREC} ${M} ${K} ${M} ${CRUN} ${REPS} mats/tet4_fluxMatrix_csr_de.mtx