Whoops, I didn’t realize this thread wasn’t publicly visible.
So I used this link to install psi4 in my MAC pc http://www.psicode.org/psi4manual/1.1/conda.html.
The details of my pc are :
System Software Overview:
System Version: OS X 10.9.2 (13C64)
Kernel Version: Darwin 13.1.0
Boot Volume: Macintosh HD
Boot Mode: Normal
Computer Name: cclear
User Name: Shreya Chandorkar (shreyac)
Secure Virtual Memory: Enabled
Time since boot: 24 days 23:30
I am not able to run jobs in multiple threads . I tried -n 4 command but still in the .out file it shows that it is running on a single thread .
What should I do now ?
-Thank you
You can try a conda binary of the the latest version conda create -n p4env python={x.x} psi4 -c psi4/label/dev. Or you can compile it yourself with an OpenMP-compatible compiler.
not AppleClang (XCode), as OpenMP is disabled
not GCC, which is just an old mess on Mac for modern software
not Intel, which has OpenMP but depends on Clang, as it’ll take you several days to build
plain Clang, should work
After doing this - conda create -n p4env python={x.x} psi4 -c psi4/label/dev.
What do I have to do next ?
Thank you
I trust you filled in a python version to that {x.x}.
If it didn’t print some instructions for you, look for the bin/.psi4-post-install file or follow Conda Binary Distribution
When I do source activate p4env it shows me “BASH_VERSION: Undefined variable.”
Is it because I am on the tcsh shell and not bash ?
Do I have to change my shell to bash then ?
So I changed my shell to bash , even activated the env to p4env , but even the when I submit my job with -n 4 , It shows me that it runs on a single thread .
You’ve got 1.2a1dev624, which by comparison to the main GitHub page, you’ll notice to be about 900 commits old (though still 600 commits ahead of where you started) and hence still non-openmp-compiled. This is probably because you’ve got sse41 for old mac hardware installed. You need to get rid of that ( probably with conda remove --features sse41) and recreate or update your env. You should be at around 1.2rc3dev100https://anaconda.org/psi4/psi4/files
Absolutely sure, no. But along with updating compilers, I’ve built packages differently now such that they should work on old hardware, so I’m reasonably confident. And if they don’t work, you can report so. Since you’re creating a new environment, you can always blow it away with no harm to your working 1.1/sse41.
Did you try the same steps as when you installed the sse41 psi4 or the same steps as when you installed the 1.2a1dev600 psi4 or the same steps I linked to the docs in the first post of this thread?
I am very glad to fill in gaps, provide up-to-date info, and suggest a path through the many resources, but while we’re not as stringent as stackoverflow guidelines, the same effort to solve your own problems and provide a detailed description of steps taken and diagnostic information is appreciated.
While you can install those files after downloading them from the website (search general conda info), those files are usually accessed through conda install, conda update, conda create -n <name> <pkg>, etc. So you need to do as advised and create a new env w/o sse41 that installs a new psi4 near the top of the list off anaconda.org
What output should I get after running the test_threading.py?
I just ran python test_threading.py and got this “Much of this script needs py3”.
Can you please tell me how to run it ?
Much of quantum chemistry involves interpreting text output and brief messages researchers have included to guide users.
There is plenty of good info and hints in the output of test_threading. If you post snippets of your output along with your analysis and the steps you have taken to support or counter those conclusions, I will confirm or correct your analysis.
