I’m new to EPW and willing to study electron-phonon coupling and superconductivity. I have performed all the required QE calculations and prepared an EPW input following the input documentation provided in EPW website. After submitting the job the code is running for several hours but no output is being generated. There must be something wrong in my input, and I’ll be thankful if someone can solve my problem. Here is my EPW input:
As I can realize this problem is related to parallelization, I was checking what happens if I vary [b:zf4l947v]-npool[/b:zf4l947v]. here is what I got :
Fisrt of all I performed all the QE calculations with 48 cores (8 nodes, 6 processors per node).
As instructed in EPW documentation, I used [b:zf4l947v]-np 48[/b:zf4l947v] and [b:zf4l947v]-npool 48[/b:zf4l947v], and the output file showed only one line :
[b:zf4l947v]APPLICATION TERMINATED WITH THE EXIT STRING: Hangup (signal 1)[/b:zf4l947v]
With [b:zf4l947v]-np 48[/b:zf4l947v] and [b:zf4l947v]-npool 8[/b:zf4l947v] the calculation started, but ended with the error :
[b:zf4l947v]stopped in DftiComputeBackward 5[/b:zf4l947v]
With [b:zf4l947v]-np 48[/b:zf4l947v] and [b:zf4l947v]-npool 6[/b:zf4l947v] the calculation started with more progress than the previous case, and ended with the error :
[b:zf4l947v]Error in routine setphases_wrap (1) :
only one proc per pool[/b:zf4l947v]
Do the test/Inputs/t01 in parallel with 4 cores or 48 as you prefer
Start modifying the t01 input so that you end up with want you want.
That way you will be able to understand which input variable is problematic.
For example filelph = ‘o_7.4t_p_1.elph’ is a depreciated variable and will be remove in the next release.