StandAloneBSSN (40x40x20)
The StandAloneBSSN benchmark
application runs the same computational kernal as the
BenchBSSN benchmark, but runs outside of Cactus. This makes
it convenient for study, for example by compiler developers.
This benchmark assigns a load of 40x40x20 grid points
on a single processor.
The results shown here are for version 1.0.
| Machine ID |
Architecture |
Fortran Compiler
& flags |
Run Time
(secs) [1] |
MFlops [2][3]
(%Peak) |
| Compaq |
|
| Hitachi |
|
| IA32 Linux |
|
| Xeon |
Dual Pentium Xeon IV 1.80 GHz |
Intel ifc v 7.0 beta
-align -O3 -tpp7 -ip -xW -parallel -unroll -i_dynamic |
62.5 |
395
() |
| IA64 Linux |
|
| Titan |
Dual IA64 800 MHz |
efc v 7.0 beta
-O1 |
5681 |
4.34
(0.1%) |
| Titan |
Dual IA64 800 MHz |
efc v 7.0 beta
-O1 -override_limits |
192 |
129
(4.0%) |
| Titan |
Dual IA64 800 MHz |
efc v 7.0 beta
-O2 -override_limits |
232 |
106
(3.3%) |
| Titan |
Dual IA64 800 MHz |
efc v 7.0 beta
-O3 -override_limits |
234 |
105
(3.3%) |
| IBM |
|
| Psi |
Power 4 |
xlc |
|
() |
|
| Seaborg |
SP3 |
Native |
|
(%) |
|
| SGI |
|
| AEI |
Origin 2000 |
MIPSpro 7.3.1.2m
-O3 -mips4 -OPT:roundoff=3:IEEE_arithmetic=3:const_copy_limit=100000:Olimit=0 -r10000 -64 |
312 |
79.1
(20.3%) |
| AEI |
Origin 2000 |
MIPSpro 7.3.1.2m
-O3 -mips4 -OPT:roundoff=3:IEEE_arithmetic=3:const_copy_limit=100000:Olimit=0 -r10000 -64 -TARG:madd=no |
279 |
88.45
(22.7%) |
Notes
- Measured using total time of gettimeofday clock.
- MFlops: This is calculated using the number of floating
point operations calculated on the Origin 2000 using perfex
including standard optimisation but
switching off multadds
(compiling with -TARG:madd=OFF). The number of floating
point operations for the different cases are listed on
the pages describing the individual benchmarks.
- MFlops here means 1,000,000 Flops.
|
![[MAINIMAGE]](../../images/development.jpg)
