The ABINIT Users Mailing List ( CLOSED )

[Alain Jacques <ajs.jacques@gmail.com>]

Hello Jim,

jimprtlnd01@gmail.com wrote:
> I am also attempting to run 5.6.4 on a Mac.  I had the same issue, and used 
> the
> EXPORT recommended above, and it worked.
>
> However, I cannot seem to get any of the tests to run successfully.  And 
> while
> my actual production runs will start, they all hang at some indeterminate 
> point
> during the run.  Sometimes after minutes, sometimes after hours.
>   

Let's go back to the basics ... run V1 test 01 by hand in order to check
your binaries

Make a clean work directory, copy the abinit-5.6.4/tests/v1/Input/t01.in
and abinit-5.6.4/tests/Psps_for_tests/13al.pspgth in it.
From a terminal window, invoque your abinis executable and fill the
answers to the questions ...

  ABINIT
 
  Give name for formatted input file:
t01.in

  Give name for formatted output file:
t01.out

  Give root name for generic input files:
t01i

  Give root name for generic output files:
t01o

  Give root name for generic temporary files:
t01tmp


.Version 5.6.4  of ABINIT
.(sequential version, prepared for a i386_darwin9.6.0_gcc4.2 computer)

....

.Starting date : Tue  6 Jan 2009.
 
- input  file    -> t01.in
- output file    -> t01.out
- root for input  files -> t01i
- root for output files -> t01o

 instrng :    53 lines of input have been read

 iofn2 : Please give name of formatted atomic psp file
13al.pspgth

.... (takes a couple of  seconds)

and now you should have in your work directory:
-rw-r--r--  1 myself  staff    323 Jan  6 22:06 13al.pspgth
-rw-r--r--  1 myself  staff   1398 Jan  6 21:53 t01.in
-rw-r--r--  1 myself  staff  24604 Jan  6 22:08 t01.out
-rw-r--r--  1 myself  staff   6001 Jan  6 22:08 t01o_DDB
-rw-r--r--  1 myself  staff  34974 Jan  6 22:08 t01o_DEN
-rw-r--r--  1 myself  staff    740 Jan  6 22:08 t01o_EIG
-rw-r--r--  1 myself  staff  27770 Jan  6 22:08 t01o_WFK

look at t01.out, compare it with abinit-5.6.4/tests/v1/Refs/t01.out and
there should be only tiny differences between the 2 results. Does it
work? If not what's the error message. I just checked with my own binary
... cf. the message (sequential version, prepared for a
i386_darwin9.6.0_gcc4.2 computer)

> So, I've done the following:
>
> ...
>
> And uploaded it as jimprtlnd01.report.tar.bz2.
>
> I could also get you a capture of the stdout as it "builds" the tests, if
> that's useful.
>
> Any information you can share about how to read these test results would be
> great.        I can't tell if it's a big failure or a success in parts, 
> because "file
> not found" doesn't always seem to be a bad thing (diffs).
>
>   

The idea - after the tests run "usefully" - is to look for files named
summaryfldiff in the subdirectories. They contain one line for each test
with a short report - success/fail/error - fast and convenient to spot
the problems. There can be plenty of reasons why the make test_acc
doesn't work - missing libs, moved build directories, lost execs, ...

> all hang at some indeterminate point
> during the run.  Sometimes after minutes, sometimes after hours.

I need more info to help you. Can you show a reproducible case?


Kind regards,

Alain