forrtl: severe (157): SIGSEGV, segmentation fault occurred

[gforney]

Hi,

I am running a number of FDS simulations and continually encounter the following error:
forrtl: severe (157): SIGSEGV, segmentation fault occurred.

------------------------------------------------
Compilation Date: Fri, 29 Oct 2010
Version: 5.5.3; MPI Disabled; OpenMP Disabled;
SVN Revision No. : 7031

image                     PC                                   Routine           Line
          Source
fds_win_64.exe       0000000140027458       Unknown        Unknown   Unknown
fds_win_64.exe       0000000140258168       Unknown        Unknown   Unknown
fds_win_64.exe       00000001404A7D38       Unknown       Unknown   Unknown
fds_win_64.exe       000000014058DDDC       Unknown       Unknown   Unknown
fds_win_64.exe       000000014056CE2B      Unknown        Unknown   Unknown
kernel32.dll            00000000766A652D       Unknown        Unknown   Unknown
ntdll.dll                   0000000076D9C521      Unknown        Unknown   Unknown

------------------------------------------------

I am running my simulation on different kind of OS and machine (win7, osx, ubuntu)
with the following specs:
Windows 7 on:
AMD Turionx2 64
4Gb RAM

Ubuntu 11.10 on:
AMD Turionx2 64
4Gb RAM

Snow Leopard on:
Intel Core 2 Duo 2,16 GHz
3Gb RAM

If I use the 32 bit version there is no problem but on 64 bit version the error is
always the same.
The version of FDS I am using is fds5_win64.exe on Win7, fds5_osx_64 on OSX, FDS_5.5.3-SMV_5.6_linux_64.tar.gz
on Ubuntu.

For the simulation I have 2 meshes.

I have attached the fds file. PLEASE help me :)
Thanks.

Original issue reported on code.google.com by lorenzo.dedonato on 2011-11-20 18:30:14

---

• _Attachment: SPADONI_2mesh.fds_

[gforney]

I'll take a look at it.

Original issue reported on code.google.com by mcgratta on 2011-11-21 16:23:14

[gforney]

All of your RAMP functions have only a single point. A RAMP should have at least 2 points.
Tell me if this fixes the problem.

Original issue reported on code.google.com by mcgratta on 2011-11-21 16:43:18

[gforney]

OK I will try! Thanks you VERY MUCH for your consideration to my problem!

Original issue reported on code.google.com by lorenzo.dedonato on 2011-11-21 17:39:21

[gforney]

Hi,
I've tried to understand the meaning of RAMP but I can't find where to modify the code.
We want all Function with constant value so we don't need RAMP.
We don't understand where to put the 2 points you wrote us to input. Can you make an
example of what do you mean about RAMP in our case?

------------------------------------------------------------------------------------------------------------
Do you mean this part of code? This code is about EFC (windows). The windows are close
or open and we don't have other values.
This windows could open thanks to "smoke detector".

&CTRL ID='C_53010_T', FUNCTION_TYPE='CUSTOM', RAMP_ID='RAMP_53010', INPUT_ID='TIMER'/
&CTRL ID='C_53010' , FUNCTION_TYPE='ANY', INPUT_ID='C_53010_T', LATCH = .FALSE., INITIAL_STATE=
.TRUE., DELAY = 0.00/

&CTRL ID='C_53012_T', FUNCTION_TYPE='CUSTOM', RAMP_ID='RAMP_53012', INPUT_ID='TIMER'/
&CTRL ID='C_53012' , FUNCTION_TYPE='ANY', INPUT_ID='C_53012_T', LATCH = .FALSE., INITIAL_STATE=
.TRUE., DELAY = 0.00/

&CTRL ID='C_53071_T', FUNCTION_TYPE='CUSTOM', RAMP_ID='RAMP_53071', INPUT_ID='TIMER'/
&CTRL ID='C_53071' , FUNCTION_TYPE='ANY', INPUT_ID='C_53071_T', LATCH = .FALSE., INITIAL_STATE=
.TRUE., DELAY = 0.00/

&CTRL ID='C_53074_T', FUNCTION_TYPE='CUSTOM', RAMP_ID='RAMP_53074', INPUT_ID='TIMER'/
&CTRL ID='C_53074' , FUNCTION_TYPE='ANY', INPUT_ID='C_53074_T', LATCH = .FALSE., INITIAL_STATE=
.TRUE., DELAY = 0.00/

&CTRL ID='C_53076_T', FUNCTION_TYPE='CUSTOM', RAMP_ID='RAMP_53076', INPUT_ID='TIMER'/
&CTRL ID='C_53076' , FUNCTION_TYPE='ANY', INPUT_ID='C_53076_T', LATCH = .FALSE., INITIAL_STATE=
.TRUE., DELAY = 0.00/

&CTRL ID='C_66636_T', FUNCTION_TYPE='CUSTOM', RAMP_ID='RAMP_66636', INPUT_ID='TIMER'/
&CTRL ID='C_66636' , FUNCTION_TYPE='ANY', INPUT_ID='C_66636_T', LATCH = .FALSE., INITIAL_STATE=
.TRUE., DELAY = 0.00/

&RAMP ID='RAMP_53010' T=10.25 F=-1/

&RAMP ID='RAMP_53012' T=10.25 F=-1/

&RAMP ID='RAMP_53071' T=40.25 F=-1/

&RAMP ID='RAMP_53074' T=40.25 F=-1/

&RAMP ID='RAMP_53076' T=40.25 F=-1/

&RAMP ID='RAMP_66636' T=40.25 F=-1/
------------------------------------------------------------------------------------------------------------

Thank you a lot, we wait for an answer! :)

Original issue reported on code.google.com by lorenzo.dedonato on 2011-11-23 16:20:54

[gforney]

&RAMP ID='RAMP_53010' T=0. F=-1/
&RAMP ID='RAMP_53010' T=10.25 F=-1/

Do this for each RAMP.

Original issue reported on code.google.com by mcgratta on 2011-11-23 16:41:05

[gforney]

OK'll try! Thanks!

Original issue reported on code.google.com by lorenzo.dedonato on 2011-11-23 17:08:43

[gforney]

It works!!!!!!! Thank you a lot!!!!!!

Original issue reported on code.google.com by lorenzo.dedonato on 2011-11-23 17:53:35

[gforney]

FDS was dividing by Delta T, which was 0 because there is only one T.

Original issue reported on code.google.com by mcgratta on 2011-11-23 17:54:57

[gforney]

Can I ask you something about multi processor calculation?
If I have a dual core cpu I must use 2 meshes .. it's right? And the same it's about
quad core? Or I need four meshes? Thanks.

Original issue reported on code.google.com by lorenzo.dedonato on 2011-11-23 17:59:48

[gforney]

If you use the MPI (Message Passing Interface) version of FDS, you need to have an equal
number of meshes to your processors. If you use the OpenMP version, you can exploit
multiple processors on the same machine. See the User's Guide for information.

Original issue reported on code.google.com by mcgratta on 2011-11-23 18:37:24

[gforney]

Ok I understand that MPI needs an equal number of meshes to my processors but if I use
multiple meshes with openMP what happens? Could you explain to me? Because I can't
run the calculation on a cluster ... I don't have it! Thank you.

Original issue reported on code.google.com by lorenzo.dedonato on 2011-11-24 18:32:09

[gforney]

http://groups.google.com/group/fds-smv/browse_thread/thread/a2e9d0803bad46a

Original issue reported on code.google.com by mcgratta on 2011-11-26 17:11:22