forked from ChenhanYu/hmlp
-
Notifications
You must be signed in to change notification settings - Fork 0
/
set_env_tacc.sh
executable file
·243 lines (190 loc) · 7.14 KB
/
set_env_tacc.sh
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
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
#!/bin/bash
## TACC module
## ======================================
taccmachine=$(dnsdomainname)
module load TACC
module load intel
module load cmake
module load python
if [[ ${taccmachine} == *"maverick"* ]]; then
module load intel/15.0.3
module load cxx11
fi
if [[ ${taccmachine} == *"stampede."* ]]; then
module load cxx11
fi
## "REQUIRED" CONFIGURATION
## ======================================
## Make sure CC and CXX are set properly in your system.
export CC=icc
export CXX=icpc
##
## Ingore this flag if you are "not using" MacOS.
##
## If you are using MacOS, exporting CC and CXX are useless.
## CC and CXX are two executable files in /usr/bin, and you
## cannot overwrite them without super users.
##
## Notice that HMLP "does not" support clang compilers.
## You "must" set following flag to true to use Intel or
## flase to use GNU compilers.
##
export HMLP_USE_INTEL=true
## Whether use BLAS or not?
export HMLP_USE_BLAS=true
## Make sure MKLROOT is defined in your system. (icc/icpc)
export MKLROOT=${MKLROOT}
## Make sure OPENBLASROOT is defined. (gcc/g++)
export OPENBLASROOT=${OPENBLASROOT}
## Setup the maximum number of threads.
export OMP_NUM_THREADS=10
if [[ ${taccmachine} == *"ls5"* ]]; then
export OMP_NUM_THREADS=24
fi
if [[ ${taccmachine} == *"maverick"* ]]; then
export OMP_NUM_THREADS=20
fi
if [[ ${taccmachine} == *"stampede"* ]]; then
export OMP_NUM_THREADS=20
fi
if [[ ${taccmachine} == *"stampede2"* ]]; then
export OMP_NUM_THREADS=68
fi
## ARTIFACT FOR REPRODUCIABILITY
## ======================================
## If you also want to compile those artifact files, then specify the path.
export HMLP_ARTIFACT_PATH=sc17gofmm
## ADVANCE OPTIONS
## ======================================
## Manually set the target architecture.
export HMLP_GPU_ARCH_MAJOR=gpu
export HMLP_GPU_ARCH_MINOR=kepler
## (1) x86_64/sandybridge,
## (2) x86_64/haswell,
## (3) arm/armv8a
## (4) mic/knl
export HMLP_ARCH_MAJOR=x86_64
export HMLP_ARCH_MINOR=sandybridge
if [[ ${taccmachine} == *"ls5"* ]]; then
export HMLP_ARCH_MAJOR=x86_64
export HMLP_ARCH_MINOR=haswell
fi
if [[ ${taccmachine} == *"stampede2"* ]]; then
export HMLP_ARCH_MAJOR=mic
export HMLP_ARCH_MINOR=knl
fi
## Manually set the QSML path if you are using arm/armv8a architecture.
export QSMLROOT=/Users/chenhan/Documents/Projects/qsml/aarch64-linux-android
## Distributed environment poptions (if true, compile with MPI)
export HMLP_USE_MPI=false
## GPU compiler options (if true, compile the gpu library as well).
## Manually setup CUDA TOOLKIT path (otherwise cmake will try to find it).
export HMLP_USE_CUDA=false
export HMLP_CUDA_DIR=$TACC_CUDA_DIR
## MAGMA (GPU LAPACK support)
export HMLP_USE_MAGMA=false
export HMLP_MAGMA_DIR=/users/chenhan/Projects/magma-2.2.0
## Output google site data
export HMLP_ANALYSIS_DATA=false
## Advance OpenMP options
export OMP_NESTED=false
export OMP_PROC_BIND=spread
## HMLP communicator
export KS_JC_NT=1
export KS_IC_NT=$OMP_NUM_THREAD
export KS_JR_NT=1
## DO NOT CHANGE ANYTHING BELOW THIS LINE
## ======================================
## Check if CC and CXX are set
echo "===================================================================="
echo "Notice: HMLP and CMAKE use variables CC and CXX to decide compilers."
echo " If the following messages pop out:"
echo ""
echo " Variable CC is unset (REQUIRED) or"
echo " Variable CXX is unset (REQUIRED),"
echo ""
echo " then you must first export these two variables."
echo "===================================================================="
if [ -z ${CC+x} ];
then echo "Variable CC is unset (REQUIRED)";
else echo "Variable CC is set to '$CC'";
fi
if [ -z ${CXX+x} ];
then echo "Variable CXX is unset (REQUIRED)";
else echo "Variable CXX is set to '$CXX'";
fi
echo "===================================================================="
echo ""
## Check if MKLROOT is set
echo "===================================================================="
echo "Notice: HMLP and CMAKE use variables MKLROOT to find Intel MKL."
echo " If you are using intel compile and seeing the following:"
echo ""
echo " Variable MKLROOT is unset (REQUIRED by intel compilers)"
echo ""
echo " then you must first export MKLROOT=/path_to_mkl..."
echo "===================================================================="
if [ -z ${MKLROOT+x} ];
then echo "Variable MKLROOT is unset (REQUIRED by intel compilers)";
else echo "Variable MKLROOT is set to '$MKLROOT'";
fi
echo "===================================================================="
echo ""
## Check if OPENBLASROOT is set
echo "===================================================================="
echo "Notice: HMLP and CMAKE use variables OPENBLASROOT to find OpenBLAS."
echo " If you are using intel compile and seeing the following:"
echo ""
echo " Variable OPENBLASROOT is unset (REQUIRED by GNU compilers)"
echo ""
echo " then you must first export OPENBLASROOT=/path_to_OpenBLAS..."
echo "===================================================================="
if [ -z ${OPENBLASROOT+x} ];
then echo "Variable OPENBLASROOT is unset (REQUIRED by GNU compilers)";
else echo "Variable OPENBLASROOT is set to '$OPENBLASROOT'";
fi
echo "===================================================================="
echo ""
## setup project source directory
export HMLP_DIR=$PWD
echo "HMLP_DIR = $HMLP_DIR"
## Add our default building path
export DYLD_LIBRARY_PATH=${DYLD_LIBRARY_PATH}:${HMLP_DIR}/build/lib/
export DYLD_LIBRARY_PATH=${DYLD_LIBRARY_PATH}:${MKLROOT}/lib/
export DYLD_LIBRARY_PATH=${DYLD_LIBRARY_PATH}:${OPENBLASROOT}/:${OPENBLASROOT}/lib/
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:${HMLP_DIR}/build/lib/
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:${MKLROOT}/lib/
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:${OPENBLASROOT}/:${OPENBLASROOT}/lib/
export PYTHONPATH=${PYTHONPATH}:${HMLP_DIR}/build/python/
export PYTHONPATH=${PYTHONPATH}:${HMLP_DIR}/build/lib/
export PYTHONPATH=${PYTHONPATH}:${HMLP_DIR}/build/lib/python2.7/site-packages/
## architecture
export HMLP_GPU_ARCH=$HMLP_GPU_ARCH_MAJOR/$HMLP_GPU_ARCH_MINOR
export HMLP_ARCH=$HMLP_ARCH_MAJOR/$HMLP_ARCH_MINOR
echo "HMLP_GPU_ARCH = $HMLP_GPU_ARCH"
echo "HMLP_ARCH = $HMLP_ARCH"
## Compiler options (if false, then use GNU compilers)
echo "HMLP_USE_INTEL = $HMLP_USE_INTEL"
## use blas?
echo "HMLP_USE_BLAS = $HMLP_USE_BLAS"
## Manually set the QSML path if you are using arm/armv8a architecture.
echo "QSMLROOT = $QSMLROOT"
## Distributed environment poptions (if true, compile with MPI)
echo "HMLP_USE_MPI = $HMLP_USE_MPI"
## GPU compiler options (if true, compile the gpu library as well)
echo "HMLP_USE_CUDA = $HMLP_USE_CUDA"
## Manually setup CUDA TOOLKIT path (otherwise cmake will try to find it)
echo "HMLP_CUDA_DIR = $HMLP_CUDA_DIR"
## MAGMA (GPU LAPACK support)
echo "HMLP_USE_MAGMA = $HMLP_USE_MAGMA"
echo "HMLP_MAGMA_DIR = $HMLP_MAGMA_DIR"
## Output google site data
echo "HMLP_ANALYSIS_DATA = $HMLP_ANALYSIS_DATA"
## OpenMP options
echo "OMP_PROC_BIND = $OMP_PROC_BIND"
echo "OMP_NUM_THREADS = $OMP_NUM_THREADS"
echo "OMP_PLACES = $OMP_PLACES"
## HMLP communicator
echo "KS_JC_NT = $KS_JC_NT"
echo "KS_IC_NT = $KS_IC_NT"
echo "KS_JR_NT = $KS_JR_NT"