Error in routine readwfc (20)
Posted: Mon Mar 11, 2019 5:27 pm
Hello, I get the following error at random points:
By "random points" I mean that it can occur after any q-point. I found a dicussion between QE developers about this issue (https://github.com/easybuilders/easybuild-easyblocks/pull/1556). It was supposed to be fixed in QE 6.4 but apparently it is not. I use QE 6.4 and following compilers/libraries:
openmpi 3.1
inter compiler 17.0
mkl 2017
This bug is expecially annoying because it occurs randomly and one has sometimes to wait for several days. If it occurs, the calculation is useless and one has to start all over again. Is there a short fix for that? According to the linkes disucssion it has something to do with how EPW/QE writes/reads the wave function files and with DIRECT_IO_FACTOR in readwfc.f90.
PS: Just found a similar thread here: viewtopic.php?f=5&t=732
I will try to recompile QE with gfort and test.
Code: Select all
===================================================================
irreducible q point # 12
===================================================================
Symmetries of small group of q: 4
Number of q in the star = 2
List of q in the star:
1 0.250000000 0.000000000 -0.501658593
2 -0.250000000 0.000000000 0.501658593
q( 23 ) = ( 0.2500000 0.0000000 -0.5016586 )
q( 24 ) = ( -0.2500000 0.0000000 0.5016586 )
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Error in routine readwfc (20):
error opening wfc file
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
stopping ...
By "random points" I mean that it can occur after any q-point. I found a dicussion between QE developers about this issue (https://github.com/easybuilders/easybuild-easyblocks/pull/1556). It was supposed to be fixed in QE 6.4 but apparently it is not. I use QE 6.4 and following compilers/libraries:
openmpi 3.1
inter compiler 17.0
mkl 2017
This bug is expecially annoying because it occurs randomly and one has sometimes to wait for several days. If it occurs, the calculation is useless and one has to start all over again. Is there a short fix for that? According to the linkes disucssion it has something to do with how EPW/QE writes/reads the wave function files and with DIRECT_IO_FACTOR in readwfc.f90.
PS: Just found a similar thread here: viewtopic.php?f=5&t=732
I will try to recompile QE with gfort and test.