Hi all,

I am trying to run a fortran code which does a time evolution for
10^6 steps. The output is written for every 10 years or so. When I
try to direct the output to a file (and keep the process in the
background, as shown below), it runs up to around 420,000 steps and
THEN gives me the following error. I'd Appreciate any suggestions/
comments.
**************************************

chandra$ ./mercury6 >>planet_10E6 &

chandra$ forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line
Source
mercury6 08062C4B Unknown Unknown Unknown
mercury6 08061A94 Unknown Unknown Unknown
mercury6 08053C82 Unknown Unknown Unknown
mercury6 0804A958 Unknown Unknown Unknown
mercury6 08049D61 Unknown Unknown Unknown
libc.so.6 00126DEC Unknown Unknown Unknown
mercury6 08049C71 Unknown Unknown Unknown

On 3 août, 15:57, habzone2...@gmail.com wrote:

On 2008-08-03 10:57:48 -0300, habzone2007@gmail.com said:

On Aug 3, 10:33 am, Gordon Sande worldnet.att.net> wrote:

I found something strange...When I *don't* direct the output to a
file, and write the output on the screen, it runs fine without any
errors.

But when I compile with "ifort -debug all" and direct the executable
output to a file, it writes the output in that file but stops the
process after sometime (~ 800,000 out of the maximum 10^6 iterations).
The "-debug all" option does not display any error info. The process
just stops writing the output in the file and quits.

habzone2007@gmail.com wrote:

How big is the file at that time

it's around 8 MB....it's not a disk space problem because I have much
more disk space available.
When I give ulimit -a to see the allowable usage limit, this is what I
get.

On Aug 3, 9:58 am, habzone2...@gmail.com wrote:

Try commenting out most of your statements writing to the output
file. Just leave enough writes in there to help you know "when" the
program bombs. This may at least speed up each run you do while
looking for the problem. Try deleting chunks of your code a bit at a
time and see if the program still runs and bombs in the same
fashion. I find this technique sometimes helps me at least narrow
things down a bit.

Dan

On 2008-08-03 11:55:03 -0300, habzone2007@gmail.com said:

On Aug 3, 4:07 pm, Gordon Sande worldnet.att.net> wrote: