| |
Abaqus Version 6.5 Performance Data
The Abaqus benchmark problems are designed to provide an estimate of
the performance that can be expected when running representative Abaqus
jobs on different computer platforms. With Version 6.5, Abaqus has
introduced three new timing problems (S6, S7, and E8) that are referred to
below as "Application Benchmarks". The Application Benchmarks are much
more indicative of "industrial strength problems" (larger model size and
longer run times) than the remaining problems in the suite, and so may
provide a better indication of scalability for "real world" problems.
If you are an Abaqus Customer, please refer to Answer 2342 for instructions on obtaining the input files
associated with these benchmark problems.
NOTE: The benchmark test problems may change between
releases. Therefore the timing data contained on this page should not be
directly compared with benchmark data obtained with other versions of
Abaqus.
Contents
Abaqus/Standard Timing Problem Descriptions
The problems described below provide an estimate of the performance
that can be expected when running Abaqus/Standard on different computers.
The jobs are representative of typical Abaqus/Standard applications:
linear and nonlinear statics and dynamics, eigenvalue analysis, and crack
propagation analysis.
S1: Plate with gravity load.
This is a linear static analysis of a plate with gravity loading. Most
of the time is spent in the inner loop of the solver.
| Input file name: |
s1-plate-with-gravity-load.inp |
| Total increments: |
1 |
| Total iterations: |
1 |
| Degrees of freedom: |
1,085,406 |
| Max. Floating point operations per iteration: |
1.9E+11 |
| Nodes: |
135,901 |
| Elements: |
45,000 |
| Element types: |
S8R5 |
| Materials: |
Linear elastic |
| Procedures: |
Static |
| Loads: |
Gravity |
| Interactions: |
None |
| Constraints: |
Fixed displacements |
| Min. Memory requirement (approximate): |
256 MB |
| Disk space requirement (approximate): |
2 GB |
S2: Natural frequencies of stiffened plate.
This is an eigenvalue analysis of a stiffened plate that is solved
using the Lanczos eigensolver. A total of 1000 eigenvectors are requested
with the number of parallel intervals set to 16. This problem is dominated
by I/O since each Lanczos iteration requires backward passes on the
decomposed and shifted stiffness matrix. The memory and disk space
requirements listed below apply for a single cpu analysis.
| Input file name: |
s2-natural-frequencies-of-a-stiffened-plate.inp |
| Total increments: |
N/A |
| Total iterations: |
N/A |
| Degrees of freedom: |
290,826 |
| Max. Floating point operations per iteration: |
1.5E+11 |
| Nodes: |
96,942 |
| Elements: |
64,800 |
| Element types: |
C3D8 |
| Materials: |
Linear elastic |
| Procedures: |
Frequency |
| Loads: |
None |
| Interactions: |
None |
| Constraints: |
Fixed displacements |
| Min. Memory requirement (approximate): |
260 MB |
| Disk space requirement (approximate): |
4.75 GB |
S3: Flywheel with centrifugal loads.
This is a static analysis of a flywheel with centrifugal loading. Most
of the time is spent in the inner loop of the solver.
| Input file name: |
s3-flywheel-with-centrifugal-load.inp |
| Total increments: |
6 |
| Total iterations: |
12 |
| Degrees of freedom: |
474,744 |
| Max. Floating point operations per iteration: |
1.86E+12 |
| Nodes: |
158,248 |
| Elements: |
145,480 |
| Element types: |
C3D8R |
| Materials: |
Linear elastic, Mises plasticity |
| Procedures: |
Static |
| Loads: |
Centrifugal |
| Interactions: |
None |
| Constraints: |
Fixed displacements |
| Min. Memory requirement (approximate): |
600 MB |
| Disk space requirement (approximate): |
4.5 GB |
S4: Dynamic crush of a car frame member.
This is a highly nonlinear dynamic analysis of a car frame member that
buckles under compression loading. The scalar speed of the computer is
important because many vector operations are on short vectors.
| Input file name: |
s4-dynamic-crush-of-a-car-frame-member.inp |
| Total increments: |
109 |
| Total iterations: |
398 |
| Degrees of freedom: |
4,686 |
| Max. Floating point operations per iteration: |
5.01E+07 |
| Nodes: |
781 |
| Elements: |
711 |
| Element types: |
S4R, MASS |
| Materials: |
Rate dependent Mises plasticity |
| Procedures: |
Dynamic |
| Loads: |
Initial velocity |
| Interactions: |
None |
| Constraints: |
Fixed displacements, linear constraint
equations |
| Min. Memory requirement (approximate): |
20 MB |
| Disk space requirement (approximate): |
8 MB |
S5: J-Integral analysis of an elbow with a crack.
This is an analysis to compute J-integrals for an elbow with a crack.
| Input file name: |
s5-j-integral-analysis-of-an-elbow-with-a-crack.inp |
| Total increments: |
1 |
| Total iterations: |
1 |
| Degrees of freedom: |
108,057 |
| Max. Floating point operations per iteration: |
1.35E+11 |
| Nodes: |
34,657 |
| Elements: |
4,937 |
| Element types: |
C3D20R, C3D27 |
| Materials: |
Mises plasticity |
| Procedures: |
Static |
| Loads: |
Pressure |
| Interactions: |
Tied contact pairs |
| Constraints: |
Fixed displacements |
| Min. Memory requirement (approximate): |
130 MB |
| Disk space requirement (approximate): |
750 MB |
Application Benchmarks
S6: Mounting, inflation, and footprint of a tire.
This is a three-dimensional nonlinear static analysis that simulates
the mounting, inflation, and footprint of a tire.
| Input file name: |
s6-mounting-inflation-and-footprint-of-a-tire.inp |
| Total increments: |
54 |
| Total iterations: |
243 |
| Degrees of freedom: |
738,820 |
| Max. Floating point operations per iteration: |
7.33E+10 |
| Nodes: |
170,025 |
| Elements: |
105,847 |
| Element types: |
C3D8H, C3D6H, SFM3D4R |
| Materials: |
Linear elastic, hyperelastic |
| Procedures: |
Static |
| Loads: |
Pressure, point force, prescribed
displacements |
| Interactions: |
Finite-sliding contact pairs |
| Constraints: |
Rigid bodies, surface-based mesh ties,
embedded elements |
| Min. Memory requirement (approximate): |
500 MB |
| Disk space requirement (approximate): |
1.5 GB |
S7: Cylinder head bolt-up.
This is a three-dimensional nonlinear static analysis that simulates
bolting a cylinder head onto an engine block. There are three versions of
this timing problem: a 700,000 DOF version that is suitable for use with
the direct sparse solver on 32-bit systems, a 5,000,000 DOF version that
is suitable for use with the direct sparse solver on 64-bit systems, and a
5,000,000 DOF version that is suitable for use with the iterative solver
on 64-bit systems.
| 700,000 DOF Version |
| Input file name: |
s7a-cylinder-head-bolt-up-700kdof.inp |
| Total increments: |
1 |
| Total iterations: |
5 |
| Degrees of freedom: |
720,059 |
| Max. Floating point operations per iteration: |
5.86E+11 |
| Nodes: |
213,221 |
| Elements: |
890,390 |
| Element types: |
C3D4, GK3D8, C3D8I |
| Materials: |
Linear elastic, elastic-plastic
gasket |
| Procedures: |
Static |
| Loads: |
Bolt load (pre-tension section) |
| Interactions: |
Small-sliding contact pairs |
| Constraints: |
Surface-based mesh ties, fixed
displacements |
| Min. Memory requirement (approximate): |
1.2 GB |
| Disk space requirement (approximate): |
3 GB |
| 5,000,000 DOF Version |
| Input file name: |
s7b-cylinder-head-bolt-up-5mdof.inp |
| Total increments: |
1 |
| Total iterations: |
5 |
| Degrees of freedom: |
5,236,958 |
| Max. Floating point operations per iteration: |
1.2E+13 |
| Nodes: |
1,060,953 |
| Elements: |
670,145 |
| Element types: |
C3D10M, GK3D8, C3D8I |
| Materials: |
Linear elastic, elastic-plastic
gasket |
| Procedures: |
Static |
| Loads: |
Bolt load (pre-tension section) |
| Interactions: |
Small-sliding contact pairs |
| Constraints: |
Surface-based mesh ties, fixed
displacements |
| Min. Memory requirement (approximate): |
5 GB |
| Disk space requirement (approximate): |
24 GB |
| 5,000,000 DOF Version (Iterative solver) |
| Input file name: |
s7c-cylinder-head-bolt-up-5mdof-ddm.inp |
| Total increments: |
1 |
| Total iterations: |
3 |
| Degrees of freedom: |
5,248,154 |
| Max. Floating point operations per iteration: |
3.76E+11 |
| Nodes: |
1,060,953 |
| Elements: |
665,813 |
| Element types: |
C3D10M, GK3D8, C3D8I |
| Materials: |
Linear elastic, elastic-plastic
gasket |
| Procedures: |
Static |
| Loads: |
Bolt load (pre-tension section) |
| Interactions: |
Small-sliding contact pairs |
| Constraints: |
Surface-based mesh ties, tied contact
pairs, fixed displacements |
| Min. Memory requirement (approximate): |
14.5 GB |
| Disk space requirement (approximate): |
3.5 GB |
Abaqus/Explicit Timing Problem Descriptions
The problems described below provide an estimate of the performance
that can be expected when running Abaqus/Explicit on different computers.
The jobs are representative of typical Abaqus/Explicit applications. The
value given for total increments may vary slightly from one platform to
another.
E1: Pipe whip simulation.
This analysis simulates a pipe-on-pipe impact resulting from the
rupture of a high-pressure line in a power plant. It is assumed that a
sudden release of fluid could cause one segment of the pipe to rotate
about its support and strike a neighboring pipe. The computer time is
balanced between the element routines and the contact routines.
| Input file name: |
e1-pipe-whip-simulation.inp |
| Total increments: |
18,416 |
| Degrees of freedom: |
19,260 |
| Nodes: |
3,210 |
| Elements: |
3,074 |
| Element types: |
S4R, MASS, ROTARYI |
| Materials: |
Mises plasticity |
| Procedures: |
Explicit dynamic |
| Loads: |
Initial velocity |
| Interactions: |
General contact |
| Constraints: |
Fixed displacements |
| Kinetic Energy: |
2.439E+03 |
| Estimated initial stable time increment |
8.336E-07 |
| Memory requirement (approximate): |
13.6 MB |
E2: Impact of a copper rod.
This analysis simulates the high velocity impact of a copper rod onto a
rigid wall. Extremely high plastic strains develop at the crushed end of
the rod, resulting in severe local mesh distortion. The performance in
this problem is a direct measure of the performance of the elements and
the three-dimensional Mises plasticity. There is no contact in this
problem.
| Input file name: |
e2-impact-of-a-copper-rod.inp |
| Total increments: |
16,872 |
| Degrees of freedom: |
14,859 |
| Nodes: |
4,953 |
| Elements: |
4,104 |
| Element types: |
C3D8R |
| Materials: |
Mises plasticity |
| Procedures: |
Explicit dynamic |
| Loads: |
Initial velocity |
| Interactions: |
None |
| Constraints: |
Fixed displacements |
| Kinetic Energy: |
2.124E+00 |
| Estimated initial stable time increment |
3.352E-08 |
| Memory requirement (approximate): |
6.5 MB |
E3: Explosively loaded cylindrical panel.
This analysis simulates a 120° cylindrical shell panel, firmly clamped
on all four sides, that is exposed to the detonation of a high explosive
layer. The problem illustrates the use of initial velocity conditions to
model sudden, impulsive loadings arising from the detonation of high
explosives. In the course of the analysis a strong plastic hinge will form
along the edge of the detonation area. The performance in this problem is
a direct measure of the performance of the elements and the plane stress
Mises plasticity. There is no contact in this problem.
| Input file name: |
e3-explosively-loaded-cylindrical-panel.inp |
| Total increments: |
5,689 |
| Degrees of freedom: |
50,310 |
| Nodes: |
8,385 |
| Elements: |
8,192 |
| Element types: |
S4RS |
| Materials: |
Mises plasticity |
| Procedures: |
Explicit dynamic |
| Loads: |
Initial velocity |
| Interactions: |
None |
| Constraints: |
Fixed displacements |
| Kinetic Energy: |
5.488E+01 |
| Estimated initial stable time increment |
1.87E-07 |
| Memory requirement (approximate): |
19.1 MB |
E4: Rail crush simulation.
This analysis simulates the impact of a rectangular, box-section rail
against a rigid wall. The analysis accounts for self contact including the
effects of changing shell thickness.
| Input file name: |
e4-rail-crush-simulation.inp |
| Total increments: |
8,689 |
| Degrees of freedom: |
11,382 |
| Nodes: |
1,901 |
| Elements: |
1,802 |
| Element types: |
S4R, R3D4 |
| Materials: |
Mises plasticity |
| Procedures: |
Explicit dynamic |
| Loads: |
Initial velocity |
| Interactions: |
General contact |
| Constraints: |
Rigid bodies, surface-based mesh
ties |
| Kinetic Energy: |
5.849E+02 |
| Estimated initial stable time increment |
8.11E-07 |
| Memory requirement (approximate): |
9.1 MB |
E5: Axisymmetric billet forming.
This analysis illustrates the forming of a circular billet of metal
that is reduced in length by 60%. The specimen is compressed between flat,
rough, rigid dies. The model is axisymmetric, and symmetry is utilized to
include only the top half of the billet. An analytical rigid surface is
used to model the die.
| Input file name: |
e5-axisymmetric-billet-forming.inp |
| Total increments: |
43,433 |
| Degrees of freedom: |
13,125 |
| Nodes: |
6,562 |
| Elements: |
6,400 |
| Element types: |
CAX4R |
| Materials: |
Mises plasticity |
| Procedures: |
Explicit dynamic |
| Loads: |
Prescribed velocity |
| Interactions: |
Finite-sliding contact pair |
| Constraints: |
Rigid bodies, fixed displacements |
| Kinetic Energy: |
8.855E+00 |
| Estimated initial stable time increment |
1.272-08 |
| Memory requirement (approximate): |
5.9 MB |
E6: Rolling of a thick plate.
This analysis simulates the hot rolling process which is used to
transform preformed shapes into a form suitable for further processing. An
analytical rigid surface is used to model the roller.
| Input file name: |
e6-rolling-of-a-thick-plate.inp |
| Total increments: |
9,122 |
| Degrees of freedom: |
41,181 |
| Nodes: |
13,726 |
| Elements: |
11,760 |
| Element types: |
C3D8R |
| Materials: |
Mises plasticity |
| Procedures: |
Explicit dynamic |
| Loads: |
Initial velocity, prescribed
velocity |
| Interactions: |
Finite-sliding contact pair |
| Constraints: |
Rigid bodies, fixed displacements |
| Kinetic Energy: |
4.640E+02 |
| Estimated initial stable time increment |
1.566E-07 |
| Memory requirement (approximate): |
19.9 MB |
E7: Deep drawing of a square box.
This analysis illustrates the forming of a three-dimensional shape by a
deep drawing process. Since the forming process is essentially a
quasi-static problem the computations are performed over a sufficiently
long time period to render inertial effects negligible. The performance of
this problem is a direct measure of the performance of the
three-dimensional general contact algorithm.
| Input file name: |
e7-deep-drawing-of-a-square-box.inp |
| Total increments: |
5,649 |
| Degrees of freedom: |
39,429 |
| Nodes: |
8,099 |
| Elements: |
7,781 |
| Element types: |
S4R, R3D4, MASS |
| Materials: |
Mises plasticity |
| Procedures: |
Explicit dynamic |
| Loads: |
Prescribed velocity |
| Interactions: |
General contact |
| Constraints: |
Rigid bodies, fixed displacements |
| Kinetic Energy: |
8.492E+00 |
| Estimated initial stable time increment |
1.749E-07 |
| Memory requirement (approximate): |
31.5 MB |
Application Benchmarks
E8: Concentric spheres.
This analysis consists of a large number of concentric spheres with
clearance between each sphere. The outer sphere is violently shaken which
results in complex contact interactions between the contained spheres.
| Input file name: |
e8-concentric-spheres.inp |
| Total increments: |
23,296 |
| Degrees of freedom: |
1,095,954 |
| Nodes: |
365,318 |
| Elements: |
244,124 |
| Element types: |
C3D8R |
| Materials: |
Mises plasticity |
| Procedures: |
Explicit dynamic |
| Loads: |
Prescribed displacements |
| Interactions: |
General contact |
| Constraints: |
None |
| Kinetic Energy: |
2.035E+06 |
| Estimated initial stable time increment |
2.116E-07 |
| Memory requirement (approximate): |
900.1 MB |
Abaqus Timing Comparisons
All times are given in hours:minutes:seconds.
Sequential Execution
For sequential execution, the times given for each problem are elapsed
time (wall-clock time) when the problem is running stand-alone on the
computer. For Abaqus/Standard, this time may vary for the same computer
depending on the amount of memory that is assigned to the Abaqus/Standard
job and the type of disks that are used. For Abaqus/Explicit, the memory
configuration on the machine may not significantly affect the run times as
long as there is sufficient memory available (the maximum memory required
for any of the Abaqus/Explicit problems is approximately 900.1 MB). All
Abaqus/Explicit timing data is obtained using the single precision
executable.
All of the parallel runs are made with the default parallel settings in
Abaqus:
Abaqus/Standard jobs with direct solver: Only Solver is run in
parallel (STANDARD_PARALLEL=SOLVER)
Abaqus/Standard jobs with iterative solver: All of Standard is run in
parallel (STANDARD_PARALLEL=ALL, MP_MODE=MPI)
Abaqus/Explicit jobs: MPI-based domain parallel, with number of
domains equal to number of CPUs (MP_MODE=MPI) on all platforms except
Windows/x86-32. On Windows/x86-32 the default is THREAD-based domain
parallel, with number of domains equal to number of CPUs
(MP_MODE=THREADS).
The column headed "Total" gives the total wall-clock (elapsed) time
from the start of the first job to the finish of the last job when the
benchmark problems are run sequentially on a dedicated system. This time
provides an indication of the performance of Abaqus when only one problem
is run on a dedicated machine.
The Total time should only be used for general, overall performance
comparisons. Due to the variation in the individual times, comparisons
between platforms for individual problems of interest may be more
indicative of the relative performances. These times should only be used
as guidelines. Times change due to modifications within Abaqus, and these
timings should not be used to compare platforms unless the same version of
Abaqus has been used for both machines. The times also depend on the
actual configuration of the computer. If a detailed comparison between
computers is important, then the timing problems should be rerun using the
same version of Abaqus and the actual configuration of the computer that
is of interest.
Simultaneous Execution
Times are also reported for situations when the machine may be heavily
loaded with jobs as might occur in a multi-user environment. These times
are obtained by running multiple jobs simultaneously.
For Abaqus/Standard, the set of jobs used for simultaneous execution
comprise S1 through S5. For Abaqus/Explicit, the set of jobs used for
simultaneous execution comprise E1 through E7. The Total Time per Set is
the sum of the times required to run multiple versions of every job in a
set simultaneously. For example, the Total Time per Set running 3 jobs
simultaneously would be obtained by adding the times required to run 3 S1
jobs at the same time, followed by 3 S2 jobs at the same time, and so on
until all the jobs in the set have been run. The data for "1 Simultaneous"
jobs is populated from the timings made available for the sequential runs.
Each job is run using only a single processor.
The column headed "Average Time per Set" is the Total Time per Set
divided by the number of simultaneous jobs. These times provide a basis
for estimating the relative performance of computer systems heavily loaded
with Abaqus jobs.
Abaqus/Standard Performance Data
NOTE: Performance data will be published as it becomes
available from the hardware vendors.
| |
| Linux/x86-32 |
| Machine Details: 2x3.0 GHz Intel Xeon CPUs, 2048 MB
Mem, 73 GB HD, 512KB Cache, Linux SuSE 8.2 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MODERATE |
0:03:56 |
2:17:32 |
2:01:40 |
0:01:57 |
0:01:35 |
8:41:48 |
0:36:40 |
|
|
|
| 2 |
MODERATE |
0:03:23 |
6:30:15 |
1:24:48 |
0:01:48 |
0:01:16 |
8:20:09 |
0:32:47 |
|
|
|
| 1 |
MAXIMUM |
0:03:42 |
2:11:10 |
1:55:18 |
0:01:55 |
0:01:33 |
8:41:48 |
0:36:40 |
|
|
|
| 2 |
MAXIMUM |
0:03:11 |
4:50:21 |
1:21:12 |
0:01:48 |
0:01:16 |
8:17:24 |
0:32:47 |
|
|
|
| Simultaneous Execution |
| Number of Simultaneous Jobs |
Total Time per Set |
Average Time per Set |
| 1 |
4:36:40 |
4:36:40 |
| 2 |
12:37:22 |
6:17:39 |
| |
| |
| Windows/x86-32 |
| Machine Details: 3.6 GHz Intel Xeon, 2 CPUs, 4096 MB
Mem, 73 GB HD, Windows XP, Abaqus V6.5-6 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MODERATE |
0:03:19 |
|
1:42:31 |
0:01:39 |
0:01:45 |
8:04:57 |
0:31:56 |
|
|
|
| 2 |
MODERATE |
0:03:12 |
|
1:12:58 |
0:01:58 |
0:01:30 |
7:15:23 |
0:30:34 |
|
|
|
| 1 |
MAXIMUM |
0:03:29 |
|
1:44:09 |
0:01:37 |
0:01:42 |
|
0:32:33 |
|
|
|
| 2 |
MAXIMUM |
0:03:31 |
|
1:14:34 |
0:01:50 |
0:01:41 |
|
0:29:44 |
|
|
|
| |
| |
| Linux/Itanium |
| Machine Details: Bull Novascale 5160 16x1.6 GHz Intel Itanium
CPUs with 9MB L3 cache, 64GB Mem, Bull Advanced Server 4 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MAXIMUM |
0:02:46 |
1:52:01 |
1:01:59 |
0:01:36 |
0:01:11 |
5:31:51 |
0:24:54 |
3:47:17 |
3:14:49 |
15:58:24 |
| 2 |
MAXIMUM |
0:02:29 |
1:50:59 |
0:34:56 |
0:01:27 |
0:00:58 |
4:39:19 |
0:20:39 |
2:18:37 |
2:07:51 |
11:57:15 |
| 4 |
MAXIMUM |
0:02:00 |
1:30:46 |
0:24:32 |
0:01:25 |
0:00:48 |
4:20:09 |
0:18:15 |
1:49:34 |
1:33:18 |
10:00:47 |
| 8 |
MAXIMUM |
0:01:54 |
1:39:48 |
0:24:41 |
0:01:29 |
0:00:27 |
4:28:00 |
0:18:03 |
1:56:34 |
1:11:42 |
10:02:38 |
| Simultaneous Execution |
| Number of Simultaneous Jobs |
Total Time per Set |
Average Time per Set |
| 1 |
2:44:08 |
2:44:08 |
| 2 |
3:11:43 |
1:35:52 |
| |
| Machine Details: HP RX4640 4x1.5 GHz Intel Itanium
CPUs, 16GB Mem, 10 DS2300 disks/2 SCSI controller, RH 3.0 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MAXIMUM |
0:02:43 |
1:10:57 |
1:04:24 |
0:01:43 |
0:01:11 |
5:10:10 |
0:23:39 |
|
|
|
| 2 |
MAXIMUM |
0:02:17 |
1:15:11 |
0:36:27 |
0:01:34 |
0:00:55 |
4:22:20 |
0:19:15 |
|
|
|
| 4 |
MAXIMUM |
0:02:02 |
1:05:47 |
0:22:48 |
0:01:30 |
0:00:46 |
4:01:14 |
0:16:59 |
|
|
|
| Simultaneous Execution |
| Number of Simultaneous Jobs |
Total Time per Set |
Average Time per Set |
| 1 |
2:20:58 |
2:20:58 |
| 2 |
2:25:29 |
1:12:45 |
| 4 |
3:31:46 |
0:52:56 |
| |
| Machine Details: HP RX1620 2x1.6 GHz Intel Itanium
CPUs, 16GB Mem, 14 DS2300 disks in a MSA30/2 SCSI controller,
RH3.0 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MAXIMUM |
0:02:31 |
1:04:13 |
1:00:14 |
0:01:38 |
0:01:07 |
5:03:38 |
0:25:13 |
|
|
|
| 2 |
MAXIMUM |
0:02:08 |
1:09:22 |
0:34:34 |
0:01:29 |
0:00:52 |
4:16:56 |
0:17:57 |
|
|
|
| Simultaneous Execution |
| Number of Simultaneous Jobs |
Total Time per Set |
Average Time per Set |
| 1 |
2:09:43 |
2:09:43 |
| 2 |
2:20:15 |
1:10:07 |
| |
| Machine Details: SGI Altix A330, 1.6 GHz Itanium2
with 6MB L3 cache, 16 CPUs, 4GB memory per CPU, Propack 3 SP6, MPT,
Abaqus V6.5-5 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1* |
S2 |
S3* |
S4 |
S5* |
S6* |
S7a* |
S7b* |
S7c |
Total |
| 1 |
MAXIMUM |
0:02:38 |
1:06:47 |
1:00:56 |
0:01:46 |
0:01:20 |
4:54:25 |
0:22:20 |
3:53:51 |
3:18:44 |
|
| 2 |
MAXIMUM |
0:02:24 |
1:08:57 |
0:33:09 |
0:01:35 |
0:01:17 |
3:05:31 |
0:16:33 |
2:18:35 |
2:10:27 |
|
| 4 |
MAXIMUM |
0:01:33 |
0:54:01 |
0:19:45 |
0:01:35 |
0:01:16 |
2:51:20 |
0:13:08 |
1:24:43 |
1:14:58 |
|
| 8 |
MAXIMUM |
0:01:38 |
0:47:39 |
0:12:19 |
0:01:40 |
0:01:02 |
1:48:52 |
0:10:55 |
1:00:01 |
0:55:43 |
|
| 12 |
MAXIMUM |
0:01:50 |
0:45:44 |
0:10:30 |
0:01:48 |
0:01:17 |
1:51:15 |
0:10:55 |
0:49:26 |
0:52:41 |
|
| * Indicates that jobs were run using non-default
parallel element operations.
| |
| |
| Machine Details: SGI Altix A350, 1.6 GHz Itanium2
with 6MB L3 cache, 16 CPUs, 4GB memory per CPU, Propack 3 SP6, MPT,
Abaqus V6.5-5 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1* |
S2 |
S3* |
S4 |
S5* |
S6* |
S7a* |
S7b* |
S7c |
Total |
| 1 |
MAXIMUM |
0:02:35 |
1:07:28 |
1:00:53 |
0:01:41 |
0:01:16 |
4:57:40 |
0:22:20 |
3:55:29 |
3:18:41 |
|
| 2 |
MAXIMUM |
0:02:17 |
1:09:06 |
0:32:34 |
0:01:32 |
0:01:01 |
3:04:56 |
0:16:02 |
2:18:30 |
2:12:41 |
|
| 4 |
MAXIMUM |
0:01:48 |
0:57:31 |
0:18:44 |
0:01:34 |
0:00:53 |
2:53:12 |
0:12:13 |
1:24:08 |
1:16:02 |
|
| 8 |
MAXIMUM |
0:01:39 |
0:45:51 |
0:12:09 |
0:01:37 |
0:00:45 |
1:42:14 |
0:10:30 |
0:57:53 |
0:48:18 |
|
| 12 |
MAXIMUM |
0:01:35 |
0:47:30 |
0:11:29 |
0:01:40 |
0:00:43 |
1:40:45 |
0:10:28 |
0:45:57 |
0:42:05 |
|
| * Indicates that jobs were run using non-default
parallel element operations.
| |
| |
| Machine Details: SGI Altix A3700, 1.6 GHz Itanium2
with 9MB L3 cache, 64 CPUs, 4GB memory per CPU, Propack 3 SP6, MPT,
Abaqus V6.5-5 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1* |
S2 |
S3* |
S4 |
S5* |
S6* |
S7a* |
S7b* |
S7c |
Total |
| 1 |
MAXIMUM |
0:02:38 |
|
1:00:55 |
0:01:41 |
0:01:19 |
4:51:37 |
0:22:16 |
3:50:14 |
3:02:32 |
|
| 2 |
MAXIMUM |
0:01:56 |
1:18:19 |
0:32:53 |
0:01:38 |
0:01:04 |
3:13:43 |
0:16:58 |
2:15:15 |
1:45:31 |
|
| 4 |
MAXIMUM |
0:01:41 |
0:57:36 |
0:18:44 |
0:01:33 |
0:00:52 |
2:51:10 |
0:12:10 |
1:21:22 |
1:09:22 |
|
| 8 |
MAXIMUM |
0:01:20 |
0:43:60 |
0:11:37 |
0:01:34 |
0:00:44 |
1:44:47 |
0:10:19 |
0:52:09 |
0:48:22 |
|
| 12 |
MAXIMUM |
0:01:16 |
0:44:37 |
0:09:49 |
0:01:39 |
0:00:42 |
1:42:19 |
0:10:06 |
0:40:29 |
0:42:35 |
|
| * Indicates that jobs were run using non-default
parallel element operations.
| |
| |
| |
| Linux/x86-64 |
| Machine Details: 4x1.8 GHz AMD Opteron CPUs, 8GB Mem,
36GB & 144GB HDs, 1024KB Cache, Linux SuSE 9.0 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MODERATE |
0:03:56 |
2:00:29 |
3:19:06 |
0:01:37 |
0:02:24 |
7:54:37 |
0:38:02 |
12:08:29 |
|
|
| 2 |
MODERATE |
0:02:42 |
2:02:29 |
1:44:10 |
0:01:30 |
0:01:34 |
5:43:06 |
0:25:02 |
6:53:13 |
|
|
| 4 |
MODERATE |
0:02:08 |
1:31:29 |
1:09:08 |
0:01:25 |
0:01:09 |
4:50:53 |
0:18:27 |
5:11:32 |
|
|
| 1 |
MAXIMUM |
0:03:49 |
1:46:53 |
3:21:39 |
0:01:44 |
0:02:23 |
7:45:11 |
0:38:35 |
12:07:57 |
|
|
| 2 |
MAXIMUM |
0:02:41 |
1:47:02 |
1:45:08 |
0:01:28 |
0:01:33 |
5:38:37 |
0:24:53 |
6:50:33 |
|
|
| 4 |
MAXIMUM |
0:02:05 |
1:24:28 |
0:57:22 |
0:01:25 |
0:01:07 |
4:38:56 |
0:18:52 |
4:15:18 |
|
|
| Simultaneous Execution |
| Number of Simultaneous Jobs |
Total Time per Set |
Average Time per Set |
| 1 |
5:27:32 |
5:27:32 |
| 2 |
6:18:58 |
3:09:29 |
| |
| Machine Details: 2x3.4 GHz Intel EM64T CPUs, 4GB Mem,
72GB HDs, 1024KB Cache, Linux SuSE 9.1 |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MODERATE |
0:02:13 |
1:30:28 |
1:37:33 |
0:01:19 |
0:01:16 |
4:56:23 |
0:28:24 |
|
|
|
| 2 |
MODERATE |
0:01:45 |
1:51:59 |
1:03:43 |
0:01:12 |
0:00:56 |
4:03:55 |
0:22:21 |
|
|
|
| 1 |
MAXIMUM |
0:02:25 |
1:19:10 |
1:36:28 |
0:01:19 |
0:01:14 |
4:52:50 |
0:27:01 |
|
|
|
| 2 |
MAXIMUM |
0:02:17 |
1:49:11 |
1:07:41 |
0:01:11 |
0:00:54 |
4:14:05 |
0:15:45 |
|
|
|
| Simultaneous Execution |
| Number of Simultaneous Jobs |
Total Time per Set |
Average Time per Set |
| 1 |
3:12:49 |
3:12:49 |
| 2 |
5:21:54 |
2:40:57 |
| |
| Machine Details: HP DL585 4x2.2GHz AMD Opteron
dual-core CPUs, 32GB Mem, 16 disks in 2 MSA30/2 SCSI
controllers, SLES9 SP1 (Abaqus V6.5-4) |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MAXIMUM |
0:02:40 |
1:15:23 |
1:57:33 |
0:01:20 |
0:01:37 |
5:16:28 |
0:25:08 |
6:44:16 |
2:57:30 |
18:41:55 |
| 2 |
MAXIMUM |
0:01:53 |
1:11:18 |
1:02:03 |
0:01:12 |
0:01:07 |
4:03:55 |
0:17:02 |
3:43:32 |
1:47:06 |
12:09:08 |
| 4 |
MAXIMUM |
0:01:32 |
0:53:22 |
0:34:08 |
0:01:07 |
0:00:52 |
3:26:47 |
0:13:02 |
2:14:15 |
1:09:00 |
8:34:05 |
| 8 |
MAXIMUM |
0:01:21 |
0:41:21 |
0:20:37 |
0:01:09 |
0:00:46 |
3:20:15 |
0:11:10 |
1:29:52 |
0:58:07 |
7:04:38 |
| Simultaneous Execution |
| Number of Simultaneous Jobs |
Total Time per Set |
Average Time per Set |
| 1 |
3:18:33 |
|
| 2 |
3:19:47 |
|
| 4 |
3:23:26 |
|
| 8 |
3:45:09 |
|
| |
| Machine Details: Fujitsu 4x2.4 GHz AMD Opteron CPUs,
16GB Mem, Linux Cluster |
| Sequential Execution |
| CPUs |
STANDARD_MEMORY_POLICY |
S1 |
S2 |
S3 |
S4 |
S5 |
S6 |
S7a |
S7b |
S7c |
Total |
| 1 |
MODERATE |
|
|
|
|
|
6:11:28 |
0:26:44 |
9:03:14 |
3:21:34 |
|
| 2 |
MODERATE |
|
|
|
|
|
4:33:04 |
0:16:55 |
4:55:05 |
1:57:51 |
|
| 4 |
MODERATE |
|
|
|
|
|
3:46:48 |
0:22:56 |
4:59:14 |
1:27:23 |
|
| | |
