orca在超算中心安装求助

shaofeng

请问有没有高手在中科院超算中心安装并成功运行orca。我折腾了几天也没搞定。如果有请高手不吝赐教。

我本是个娃娃

http://bbs.keinsci.com/forum.php ... ypeid%26typeid%3D11

shaofeng

我本是个娃娃 发表于 2017-6-17 09:23
http://bbs.keinsci.com/forum.php?mod=viewthread&tid=5335&extra=page%3D1%26filter%3Dtypeid%26typeid%3 ...

感谢你的指点，不过我按照你的方法安装后，仍然不能正确运行。显示的错误信息如下：
[c5432:32369] *** Process received signal ***
[c5432:32369] Signal: Segmentation fault (11)
[c5432:32369] Signal code: Address not mapped (1)
[c5432:32369] Failing at address: 0x6b9410001bf
[c5432:32357] *** Process received signal ***
[c5432:32357] Signal: Segmentation fault (11)
[c5432:32357] Signal code: Address not mapped (1)
[c5432:32357] Failing at address: 0x6b9410001bf
[c5432:32359] *** Process received signal ***
[c5432:32359] Signal: Segmentation fault (11)
[c5432:32359] Signal code: Address not mapped (1)
[c5432:32359] Failing at address: 0x6b9410001bf
[c5432:32373] *** Process received signal ***
[c5432:32373] Signal: Segmentation fault (11)
[c5432:32373] Signal code: Address not mapped (1)
[c5432:32373] Failing at address: 0x6b9410001bf
[c5432:32356] *** Process received signal ***
[c5432:32356] Signal: Segmentation fault (11)
[c5432:32356] Signal code: Address not mapped (1)
[c5432:32356] Failing at address: 0x6b9410001bf
[c5432:32358] *** Process received signal ***
[c5432:32358] Signal: Segmentation fault (11)
[c5432:32358] Signal code: Address not mapped (1)
[c5432:32358] Failing at address: 0x6b9410001bf
[c5432:32361] *** Process received signal ***
[c5432:32361] Signal: Segmentation fault (11)
[c5432:32361] Signal code: Address not mapped (1)
[c5432:32361] Failing at address: 0x6b9410001bf
[c5432:32364] *** Process received signal ***
[c5432:32364] Signal: Segmentation fault (11)
[c5432:32364] Signal code: Address not mapped (1)
[c5432:32364] Failing at address: 0x6b9410001bf
[c5432:32370] *** Process received signal ***
[c5432:32370] Signal: Segmentation fault (11)
[c5432:32370] Signal code: Address not mapped (1)
[c5432:32370] Failing at address: 0x6b9410001bf
[c5432:32372] *** Process received signal ***
[c5432:32372] Signal: Segmentation fault (11)
[c5432:32372] Signal code: Address not mapped (1)
[c5432:32372] Failing at address: 0x6b9410001bf
[c5432:32374] *** Process received signal ***
[c5432:32374] Signal: Segmentation fault (11)
[c5432:32374] Signal code: Address not mapped (1)
[c5432:32374] Failing at address: 0x6b9410001bf
[c5432:32375] *** Process received signal ***
[c5432:32375] Signal: Segmentation fault (11)
[c5432:32375] Signal code: Address not mapped (1)
[c5432:32375] Failing at address: 0x6b9410001bf
[c5432:32363] *** Process received signal ***
[c5432:32376] *** Process received signal ***
[c5432:32376] Signal: Segmentation fault (11)
[c5432:32376] Signal code: Address not mapped (1)
[c5432:32376] Failing at address: 0x6b9410001bf
[c5432:32365] *** Process received signal ***
[c5432:32365] Signal: Segmentation fault (11)
[c5432:32365] Signal code: Address not mapped (1)
[c5432:32365] Failing at address: 0x6b9410001bf
[c5432:32363] Signal: Segmentation fault (11)
[c5432:32363] Signal code: Address not mapped (1)
[c5432:32363] Failing at address: 0x6b9410001bf
[c5432:32362] *** Process received signal ***
[c5432:32362] Signal: Segmentation fault (11)
[c5432:32362] Signal code: Address not mapped (1)
[c5432:32362] Failing at address: 0x6b9410001bf
[c5432:32366] *** Process received signal ***
[c5432:32366] Signal: Segmentation fault (11)
[c5432:32366] Signal code: Address not mapped (1)
[c5432:32366] Failing at address: 0x6b9410001bf
[c5432:32368] *** Process received signal ***
[c5432:32368] Signal: Segmentation fault (11)
[c5432:32368] Signal code: Address not mapped (1)
[c5432:32368] Failing at address: 0x6b9410001bf
[c5432:32360] *** Process received signal ***
[c5432:32360] Signal: Segmentation fault (11)
[c5432:32360] Signal code: Address not mapped (1)
[c5432:32360] Failing at address: 0x6b9410001bf
[c5432:32371] *** Process received signal ***
[c5432:32355] *** Process received signal ***
[c5432:32355] Signal: Segmentation fault (11)
[c5432:32355] Signal code: Address not mapped (1)
[c5432:32355] Failing at address: 0x6b9410001bf
[c5432:32367] *** Process received signal ***
[c5432:32367] Signal: Segmentation fault (11)
[c5432:32367] Signal code: Address not mapped (1)
[c5432:32367] Failing at address: 0x6b9410001bf
[c5432:32371] Signal: Segmentation fault (11)
[c5432:32371] Signal code: Address not mapped (1)
[c5432:32371] Failing at address: 0x6b9410001bf
[file orca_main/gtoint.cpp, line 139]: ORCA finished by error termination in ORCA_GTOInt

可能是openmpi没安装好。
用超算中心安装的openmpi-1.6.5运行orca3.3能够正常运行。
还请多指点一下。

liyuanhe211

shaofeng 发表于 2017-6-17 10:52
感谢你的指点，不过我按照你的方法安装后，仍然不能正确运行。显示的错误信息如下：
[c5432:32369] ***  ...

ORCA4必须用openmpi 2.0.2

shaofeng

liyuanhe211 发表于 2017-6-17 11:11
ORCA4必须用openmpi 2.0.2

是用的2.0.2，orca3.3用的是openmpi1.6.5。openmpi1.6.5是超算中心编译的。2.0.2是安装您介绍的方法编译的。

liyuanhe211

shaofeng 发表于 2017-6-17 11:22
是用的2.0.2，orca3.3用的是openmpi1.6.5。openmpi1.6.5是超算中心编译的。2.0.2是安装您介绍的方法编译 ...

运行mpirun --version查看输出

shaofeng

liyuanhe211 发表于 2017-6-17 12:10
运行mpirun --version查看输出

查看了一下，看来没装对。which mpirun显示是还/usr/bin/mpirun。不知道怎么安装好。

冰释之川

shaofeng 发表于 2017-6-17 19:59
查看了一下，看来没装对。which mpirun显示是还/usr/bin/mpirun。不知道怎么安装好。

建议作业提交脚本里直接定义2.0.2的环境变量

shaofeng

secondviolin 发表于 2017-6-17 21:42
在脚本中定义openmpi变量
另外在命令行中用全部路径

还是不能运行，不知道orca4.0 有没有openmpi1.6.5版本的

脚本内容如下：
#BSUB -W 0:05
#BSUB -n 18
#BSUB -R "span[ptile=18]"
#BSUB -q cpu_dbg
#BSUB -o %J.out
#BSUB -e %J.err
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/soft/compiler/intel/composer_xe_2013_sp1.0.080/compiler/lib/intel64
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/shaofeng/openmpi/lib
export PATH=$PATH:/home/shaofeng/openmpi/bin
export PATH=$PATH:/home/shaofeng/software/orca_4_0_0_linux_x86-64
/home/shaofeng/software/orca_4_0_0_linux_x86-64/orca fer_H2AsO4.inp > fer_H2AsO4.out

错误信息如下：
*** An error occurred in MPI_Init
*** on a NULL communicator
*** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
***    and potentially your MPI job)
[c5210:28549] Local abort before MPI_INIT completed completed successfully, but am not able to aggregate error messages, and not able to guarantee that all other processes were killed!
*** An error occurred in MPI_Init
*** on a NULL communicator
*** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
***    and potentially your MPI job)
[c5210:28545] Local abort before MPI_INIT completed completed successfully, but am not able to aggregate error messages, and not able to guarantee that all other processes were killed!
*** An error occurred in MPI_Init
*** on a NULL communicator
*** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
***    and potentially your MPI job)

fantasticqhl

我也是最近发现的，
export LD_LIBRARY_PATH=/soft/compiler/intel/composer_xe_2013_sp1.0.080/compiler/lib/intel64:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/home/shaofeng/openmpi/lib:$LD_LIBRARY_PATH
export PATH=/home/shaofeng/openmpi/bin:$PATH
export PATH=/home/shaofeng/software/orca_4_0_0_linux_x86-64:$PATH

这样你which mpirun的时候，你设定的openmpi就会是第一个被检索到的，而不是系统的

shaofeng

fantasticqhl 发表于 2017-6-18 00:34
我也是最近发现的，
export LD_LIBRARY_PATH=/soft/compiler/intel/composer_xe_2013_sp1.0.080/compiler ...

多谢指点。刚才试了下，提交任务仍然不行，但是直接在服务节点运行程序是没有问题的。不知道是不是提交的过程中没有把openmpi给提交过去。