Hi all,
Recently our university installed a new computational cluster with
Intel Composer XE 2011 on board (featuring icc 12.0.0 and ifort 12.0.0).
Today I've tried to compile Siesta 3.0rc2 on the cluster using ifort
12 with arch.make made for ifort 10. The compilation ran smoothly and
successfully, both in serial and in parallel mode (using mvapich-1.2rc1
compiled with ifort 12). But when I run test calculation, either
serial or parallel, it terminates with the following error just before
calculating k-point distribution:
Total number of electrons: 8.000000
Total ionic charge: 8.000000
forrtl: severe (71): integer divide by zero
Image PC Routine Line Source
siesta 00000000012D0468 Unknown Unknown Unknown
siesta 000000000170AB38 Unknown Unknown Unknown
When tryng to build siesta using -g (debug) flag, the calculation goes
slightly further and falls with SIGSEGV:
Total number of electrons: 8.000000
Total ionic charge: 8.000000
Kpoints in: 1183 . Kpoints trimmed: 1099
siesta: k-grid: Number of k-points = 1099
siesta: k-grid: Cutoff (effective) = 16.156 Ang
siesta: k-grid: Supercell and displacements
siesta: k-grid: 0 13 0 0.000
siesta: k-grid: 0 0 13 0.000
siesta: k-grid: 13 0 0 0.000
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
siesta 0000000000772F2C Unknown Unknown Unknown
siesta 000000000076B193 Unknown Unknown Unknown
siesta 000000000077655F Unknown Unknown Unknown
siesta 0000000000726D53 Unknown Unknown Unknown
siesta 0000000000749D8E Unknown Unknown Unknown
siesta 000000000056D42E Unknown Unknown Unknown
siesta 000000000056B5FE Unknown Unknown Unknown
siesta 00000000005A7B67 Unknown Unknown Unknown
siesta 00000000005C2F5C Unknown Unknown Unknown
siesta 000000000040CF2C Unknown Unknown Unknown
libc.so.6 00007FFFF75F5C5D Unknown Unknown Unknown
siesta 000000000040CE29 Unknown Unknown Unknown
Arch.make is attached.
So, the question is: has anyone succeeded in building working siesta
(or transiesta) binary with ifort 12? If yes, then how?
Thank you in advance.
Andrey.
