In the WIEN2k 14.2 userguide [ http://www.wien2k.at/reg_user/textbooks/usersguide.pdf ] on page 80, it says:

/lapw0_mpi is parallelized over the number of atoms and with a parallel FFT, which is important in case you have large FFT grids. This method leads to good scalability as long as there are more atoms than processors./

As the second sentence above says, you can use 128 cores or less, but you should NOT use more cores than that.

In the post at that link, the structure had 8 atoms, but the .machines file was set to use 48 cores. So the .machines file needed to be changed to not use more than 8 cores.

On 6/24/2016 7:53 AM, Wangwei Lan wrote:
Dear WIEN2K user,

I am now doing parallel calculation on lapw0, one thing I notice is that, lapw0 is only paralleled over atoms, as discussed here

http://www.mail-archive.com/wien%40zeus.theochem.tuwien.ac.at/msg06622.html


In my case, I have a 4x4x4 supercell, and in total 128 atoms, so that requires 128 cores. However, it's seems impossible for me to use so many cores at this time. Is my understanding correct on this issue?

Does anyone have a good idea on how to solve this problem? Thanks in advance.

Best Regards
Wangwei Lan
_______________________________________________
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Reply via email to