Skip to Content.
Sympa Menu

forum - Jobs crashing on IBM Blue Gene/L

forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

Jobs crashing on IBM Blue Gene/L


Chronological Thread 
  • From: Markus Franke <Markus.Franke@informatik.tu-chemnitz.de>
  • To: forum@abinit.org
  • Subject: Jobs crashing on IBM Blue Gene/L
  • Date: Wed, 13 Sep 2006 13:06:10 +0200

Dear Abinit users,

I try to get the parallel version of Abinit working on an IBM Blue Gene/L system. Unfortunately, my jobs are crashing just after a few seconds working. The error file says something like this:

---snip---
1525-107
1525-107
<Sep 13 10:54:50.147100> BE_MPI (ERROR): The error message in the job record is as follows:
<Sep 13 10:54:50.147217> BE_MPI (ERROR): "killed by exit(1) on node 3"
<Sep 13 10:54:50.289831> BE_MPI (ERROR): The error message in the job record is as follows:
<Sep 13 10:54:50.289864> BE_MPI (ERROR): "killed by exit(1) on node 3"
---snap---


The output file looks like this:

---snip---
iofn1 : COMMENT -
Because of cpp option CHGSTDIO,
read file "ab.files" instead of standard input

ABINIT

Give name for formatted input file:
t30.in
Give name for formatted output file:
t30.out
Give root name for generic input files:
t30.i
Give root name for generic output files:
t30.o
Give root name for generic temporary files:
t30.status
---snap---

The error code "1525-107" comes from open ( ... status="NEW" ...) when the file already exists. An important fact is that this error occurs since a recompilation of Abinit with the new IBM XL Fortran Compiler V10.1. Version 9.1 didn't produce the problems.

Some facts about environment:

Abinit 5.1.4
BlueGene/L Version 1 Release 3
IBM XL C/C++ V8.0
IBM XL Fortran V10.1


Thanks for help,
Markus Franke




Archive powered by MHonArc 2.6.16.

Top of Page