Error using more than 1 processor

This forum is for issues related to parallel processing
and OpenSees using the new interpreters OpenSeesSP and OpenSeesMP

Moderator: selimgunay

Post Reply
macarojaime
Posts: 21
Joined: Sat Jan 05, 2008 8:40 pm
Location: University of Pittsburgh

Error using more than 1 processor

Post by macarojaime » Mon Oct 13, 2008 4:32 pm

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

macarojaime
Posts: 21
Joined: Sat Jan 05, 2008 8:40 pm
Location: University of Pittsburgh

Post by macarojaime » Mon Oct 13, 2008 4:41 pm

To be more precise, my processor is an AMD Turion™ 64 X2 Dual-Core.

MARIA

fmk
Site Admin
Posts: 5866
Joined: Fri Jun 11, 2004 2:33 pm
Location: UC Berkeley
Contact:

Post by fmk » Tue Oct 14, 2008 11:13 am

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

macarojaime
Posts: 21
Joined: Sat Jan 05, 2008 8:40 pm
Location: University of Pittsburgh

Post by macarojaime » Tue Oct 14, 2008 11:37 am

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 Does it mean I have to download a newer version of the binary and try it again?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

fmk
Site Admin
Posts: 5866
Joined: Fri Jun 11, 2004 2:33 pm
Location: UC Berkeley
Contact:

Post by fmk » Tue Oct 14, 2008 11:59 am

what options are you using to build the template material with.

macarojaime
Posts: 21
Joined: Sat Jan 05, 2008 8:40 pm
Location: University of Pittsburgh

New error

Post by macarojaime » Tue Oct 14, 2008 12:02 pm

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

fmk
Site Admin
Posts: 5866
Joined: Fri Jun 11, 2004 2:33 pm
Location: UC Berkeley
Contact:

Post by fmk » Tue Oct 14, 2008 12:04 pm

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.

macarojaime
Posts: 21
Joined: Sat Jan 05, 2008 8:40 pm
Location: University of Pittsburgh

Post by macarojaime » Tue Oct 14, 2008 12:12 pm

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]

fmk
Site Admin
Posts: 5866
Joined: Fri Jun 11, 2004 2:33 pm
Location: UC Berkeley
Contact:

Post by fmk » Wed Oct 22, 2008 2:31 pm

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

macarojaime
Posts: 21
Joined: Sat Jan 05, 2008 8:40 pm
Location: University of Pittsburgh

Post by macarojaime » Mon Oct 27, 2008 7:42 pm

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

macarojaime
Posts: 21
Joined: Sat Jan 05, 2008 8:40 pm
Location: University of Pittsburgh

Post by macarojaime » Mon Dec 08, 2008 10:10 am

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

macarojaime
Posts: 21
Joined: Sat Jan 05, 2008 8:40 pm
Location: University of Pittsburgh

Sequential does not work either

Post by macarojaime » Mon Dec 08, 2008 10:53 am

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.

jskabanda
Posts: 3
Joined: Tue Jul 05, 2011 6:28 pm
Location: University of Toronto

Re: Error using more than 1 processor

Post by jskabanda » Thu Nov 03, 2011 7:29 pm

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

Post Reply