Civil Engineering Reference
In-Depth Information
nels nxe nze nip
125000 50 50 8
aa bb cc kx ky kz
0.02 0.02 0.02 1.0 1.0 1.0
dtim nsteps
0.0002 5000
npri val0
500 100.0
Figure 12.23
Data for Program 12.5 example
This job ran on 16 processors
There are 132651 nodes 7651 restrained and 125000 equations
Time after setup is : 0.209999999999126885
Time after element integration is : 8.25999999999476131
Time Pressure
0.1000E+00 0.8551E+02
0.2000E+00 0.4599E+02
0.3000E+00 0.2228E+02
.
.
0.9000E+00 0.2595E+00
0.1000E+01 0.1231E+00
This analysis took : 1006.36000000000058
Figure 12.24
Results from Program 12.5 example
Mesh No of Processors Analysis Time(secs)
50x50x50 16 1006
32 516
Figure 12.25
Performance statistics: Program 12.5 (IBM SP2)
As is often the case in the parallel programs, an extra loop
elements_0
is used
to organise geometry and distribution of the freedoms via steering vector
g
. Thereafter
loop
elements_1
with its embedded
gauss_pts
carries over from serial to parallel
versions, as does loop
elements_2
. In parallel, results at the appropriate freedom,
is
,
are printed from processor
it
. Data are listed as Figure 12.23 with results as Figure 12.24
and performance statistics as Figure 12.25.
Program 12.6 Three dimensional steady state Navier-Stokes analysis. Compare
Program 9.2.
PROGRAM p126
!-------------------------------------------------------------------------
! Program 9.2 steady state 3-d Navier-Stokes equation
! using 20-node velocity hexahedral elements ; ns_cube
! coupled to 8-node pressure hexahedral elements ; u-p-v-w order
! element by element solution using BiCGSTAB(L) ; parallel version
!-------------------------------------------------------------------------
USE new_library; USE geometry_lib; USE precision; USE utility
USE mp_module ; USE timing; USE global_variables1
USE gather_scatter6; IMPLICIT NONE
! ndof,nels,neq,ntot are now global variables - not declared
