Page 1 of 1
Error using more than 1 processor
Posted: Mon Oct 13, 2008 4:32 pm
by macarojaime
Hello,
I have installed and configured all the files so I can make use of OpenSeesSP. I followed very carefully all the steps in doing this.
Whenever I try to run (for instance):
mpiexec -np 2 OpenSeesSP.exe C:\model.tcl
I get the following error text:
Slave Process Running 1
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
Now, when I change the np to 1 (instead of 2), the interpreter works perfectly, giving me the same results as if I were using single computing, and taking the same time to process my model.
My system is Windows Vista, and I am working in a laptop computer with AMD Turion X2 processor. I am assuming I have 2 CPUs (duo-processor). Am I wrong?
Thank you.
MARIA JAIME
Doctoral Student
University of Pittsburgh
Posted: Mon Oct 13, 2008 4:41 pm
by macarojaime
To be more precise, my processor is an AMD Turion™ 64 X2 Dual-Core.
MARIA
Posted: Tue Oct 14, 2008 11:13 am
by fmk
the element you are using could not send/recvSelf .. i have updated the element .. can you send me the materials you are using so i can ensure they have the needed routines before i post a new .exe
Posted: Tue Oct 14, 2008 11:37 am
by macarojaime
I appreciate it Frank. Does it mean I have to download a newer version of the binary and try it again?
I am using nDMaterial NewTemplate3Dep (from UCDavis). Please let me know if I should change something.
Thank you.
MARIA JAIME
Posted: Tue Oct 14, 2008 11:59 am
by fmk
what options are you using to build the template material with.
New error
Posted: Tue Oct 14, 2008 12:02 pm
by macarojaime
Frank, I did download the binary again and tried to run my model, obtaining the following
NEW error log in the DOS shell:
C:\>mpiexec -np 2 OpenSeesSP.exe C:\model.tcl
END of Stage 0: SOIL COLUMN CREATED ----------------
Stage 1 begins: CREATE SELF WEIGHT ----------------
Slave Process Running 1
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
FEM_ObjectBrokerAllClasses::getNewElement - - no Element type exists for class tag 7003
Master Process Running OpenSees Interpreter 0
OpenSees -- Open System For Earthquake Engineering Simulation
Pacific Earthquake Engineering Research Center -- 2.0.0
(c) Copyright 1999,2000 The Regents of the University of California
All Rights Reserved
(Copyright and Disclaimer @
http://www.berkeley.edu/OpenSees/copyright.html
DOF_Group::getTangForce() - not yet implemented
Posted: Tue Oct 14, 2008 12:04 pm
by fmk
i have not updated the working element yet .. it looks like i have to add some code to the NewTemplate as well .. just want to get the options you pass into the material so i get them all.
Posted: Tue Oct 14, 2008 12:12 pm
by macarojaime
This is what I am using for my material, followed by the elements definition:
Code: Select all
#~~~~~ Material Parameters ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
set MatCons "$rho $e0 $G0 $poisson $P_at $k_c $M_csl $c $lambda_c $xi $e_r $m_low $h0 $c_h $n_b $A0 $n_d $z_max $c_z"
set intTens "$alpha_T $z_T"
set MP "MaterialConstant 19 $MatCons InternalTensor 2 $intTens"
#~~~~~ Elastic State ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
set ES "Dafalias-Manzari 3 4 5 6 2 $initS"
#~~~~~ Yield Function ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
set YF "Dafalias-Manzari 0 12 2 1"
#~~~~~ Plastic Flow ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
set PF "Dafalias-Manzari 0 2 0 11 0 9 0 10 0 5 0 12 0 7 0 8 0 16 0 17 2 1 2 2"
#~~~~~ Tensor Evolution Law ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
set tel "Dafalias-Manzari 2 11 9 10 5 12 7 8 15 13 14 3 1 2"
set ftel "Dafalias-Manzari-fabric 12 18 19 1 2"
set TE "$tel $ftel"
nDMaterial NewTemplate3Dep 1 -MaterialParameter $MP \
-ElasticState $ES \
-YieldFunction $YF \
-PlasticFlow $PF \
-TensorEvolution $TE \
-Algorithm 0
##############################################################
for {set i 1} {$i <= $Num_Elem} {incr i} {
set N1 [expr 4*$i+1] ; set N2 [expr 4*$i+2]
set N3 [expr 4*$i+3] ; set N4 [expr 4*$i+4]
set N5 [expr 4*$i-3] ; set N6 [expr 4*$i-2]
set N7 [expr 4*$i-1] ; set N8 [expr 4*$i-0]
element Brick8N_u_p_U $i $N1 $N2 $N3 $N4 $N5 $N6 $N7 $N8 1 0.0 0.0 $g $poro $alpha $rho_s $rho_f \
$perm_x $perm_y $perm_z $B_s $B_f $PRESS
}
[/size]
Posted: Wed Oct 22, 2008 2:31 pm
by fmk
can you send me the file .. i don't have any example to test and your variables are not defined fmckenna AT berkeley DOT edu
Posted: Mon Oct 27, 2008 7:42 pm
by macarojaime
Frank,
I am running into some inconsistencies in the results (stress calculations) after trying modelling a looser soil layer on top of a denser one. I believe there might be a bug in the code. I am still running with 1 processor, eventhough the commands are setup for parallel computing. This was working fine for a uniform soil mass but not anymore for non-uniform soils. I will try to solve this by contacting the author and then I'll get back to you.
Thanks for your time.
Maria
Posted: Mon Dec 08, 2008 10:10 am
by macarojaime
Hi Frank,
I'm going to email you the file of the script I'm using right now.
I am using Brick20N_u_p_U elements and nDMaterial NewTemplate3Dep material model.
This is the error message I recently got:
WARNING MumpsParallelSolver::solve(void)- Error -10 returned in substitution dmumps()
WARNING KrylovNewton::solveCurrentStep() -the LinearSysOfEqn failed in solve()
ERROR 1 during ass_niv1 140 223
WARNING MumpsParallelSolver::solve(void)- Error -8 returned in substitution dmumps()
WARNING KrylovNewton::solveCurrentStep() -the LinearSysOfEqn failed in solve()
ERROR 1 during ass_niv1 140 223
The 3rd and 4th warnings are repeated over and over again.
I will appreciate you helping me with this one.
Thanks,
Maria
Sequential does not work either
Posted: Mon Dec 08, 2008 10:53 am
by macarojaime
I changed the system from Mumps to UmfPack, thinking that my initial problem was the parallel interpreter, so I ran it in the sequential OpenSees.exe.
I got similar error messages with the sequential interpreter as well, so now I believe my problem is in the actual solving of each element. By the way, I am only applying a gravity force by means of BrickW element load.
Before, I was using 8-node elements without this type of problem in the solution. These error messages appeared since I tried to implement 20-node elements.
These are the error messages running in the sequential interpreter:
UMD2FA: matrix is singular. Only 908 pivots found.
WARNING UmfpackGenLinSolver::solve(void)- 4 returned in factorization UMD2FA()
WARNING KrylovNewton::solveCurrentStep() -the LinearSysOfEqn failed in solve()
Please help.
Re: Error using more than 1 processor
Posted: Thu Nov 03, 2011 7:29 pm
by jskabanda
I am trying to run OpenseesSP on my intel laptop which am sure has more than 1 processor, it seems to run when I use mpiexec -np 2 Opensees.exe ex1.tcl but it will not run when I use mpiexec -np 2 Opensees.exe ex1.tcl