XANSYS Message: 237 [Go back to message list]
[bookmark on del.icio.us]
No rating yet
Rate item:

Subject: ANSYS/DYNA DEC configuration problem
Author: Heinz.Hempowitz@
Date: 1999-01-15 07:49:00

Hi XANSYS / ANSYS / LS-DYNA / FEM folks,

we need some support on a mixed software/hardware configuration problem.
I will explain the problem to you as good as possible.

We work with LS-DYNA 3D as a solver and with some programs as
Pre/Post-processors.

PreProcessors:

ANSYS/DYNA 5.51 Pre/Post
LS-INGRID (will come)

Post Processors:

ANSYS/DYNA 5.51 Pre/Post
LS-TAURUS

SOLVER

LS-DYNA MPP (as ordered from ERAB) ver. 9.402
LS-DYNA normal ver. 9.402

Our standard combination is ANSYS/DYNA PRE/Post as Pre- and Postprocessor
and LS-DYNA 3D solver. Most of the time we use LS-Taurus for post-support.

The Main-Computer is a DEC ALPHASERVER 8400 with 10 EV56 Processors.
We build the model in ANSYS which prepares a *.k-file (LS-DYNA3D keyword
file).

The calculation in LS-DYNA3D produces the d3plot files (LS-Dyna3d results)
and also the *.his, *.db and *.rst files.
Later, we use ANSYS 5.51 for postprocessing. Like we said before, sometimes
also LS-TAURUS.

We used DEC UNIX 4.0d as OP-SYSTEM.

IT WORKED.

Now we should get the EV6 processor upgrade for the DEC ALPHA.
Therefore, our data team had to change the OS to DEC UNIX 4.0e.

On our machine, the "sizer -v" command gives the following output:

Digital UNIX V4.0E (Rev. 1091); Fri Jan 8 16:52:41 MET 1999

After the upgrade, the calculation and preprocessing worked normal, until
we execute a plot or list command in /post1.

- we had ALLWAYS a CRASH with SEGMENTATION FAULT ERROR .
it was possible to resume, to ask for a set,list and a set,1,20.
CRASH on prns,***,*** and also on plns,***,***.

prns,epto

PRINT EPTO NODAL SOLUTION PER NODE

*** ERROR *** CP= 7.215 TIME= 17:48:44
An unexpected error ( Segmentation Violation ) has occurred... Memory
resident data has been corrupted. ANSYS is unable to recover and will
terminate.

- ANSYS had the same error also in using the PC-version and reading the
data from the DEC.
Same Crash, same behaviour, but with a little bit more detailed error
description.

prns,epto

PRINT EPTO NODAL SOLUTION PER NODE

*** ERROR *** CP= 7.215 TIME= 17:48:44
An unexpected error ( Segmentation Violation ) has occurred... Memory
resident data has been corrupted. ANSYS is unable to recover and will
terminate.

Previously saved files are unaffected.

Current ANSYS Traceback:
panavg
paprns
paprnt
post1
ANSYS

DATA INPUT WRITTEN ON test.log

- also LS-Taurus had an error in plotting stress or strain vectors.
Crash with the folowing error:

forrtl: error (65): floating invalid
Abort process

Because LS-DYNA3D and ANSYS/DYNA Pre/Post should run on the DEC ALPHA with
the new EV6 processors, it must be some solvable conflict or problem in the
configuration of the machine or LS-DYNA or ANSYS.

ALL of the desired procedures have worked very good on the old EV56
processors with DEC UNIX 4.0d

The errors are independent on the input file.

- the installed software list:

setld -i: (See attached file: setld-i.log)

We have some serious problems in finding the real error behind the
segmentation fault.
Maybe someone at ANSYS INC. or LSTC or COMPAQ/DEC can help us.

best regards,

Heinz

Heinz Hempowitz
NORSK HYDRO ASA
Technology & Projects
LME
FEA-Team
P.O. Box 190
N-1321 Stabbek
NORWAY

phone ++47 2273 7763
fax ++47 2273 9535
mobile ++47 950 83 885
email heinz.hempowitz@h...