Some benchmark results

Wed Feb 9 07:13:00 GMT 2005

Hi Terry,
SciMark 2 has a commandline option "-large" enlarge the problem sets.
You should try it unless you want to measure JVM start-up time.
Yes, start-up does also matter but throughput is more important for
HPC applications.
There are results of SciMark 2, SPEC JVM98 and Linpack benchmark on
the following page:
 Performance Comparison of Java/.NET Runtimes (Oct 2004)
 http://www.shudo.net/jit/perf/
GCJ 3.4.2 could not beat Sun's JVMs in SciMark 2 with "-large" option.
Thanks for your efforts.
I have been hoping Java and GCJ is a feasible and a good language
and runtime even for HPC, working on it.
> I did some benchmarking, and things look quite good for gcj and high
> performance computing.

> and the other is "Scimark" from NIST. Gcj beat JDK 1.5 very
> handily in both.

> ---benchmark results follow---

> For Scimark, the numbers are:
>> gcj:
> SciMark 2.0a
>> Composite Score: 397.56724916413043
> FFT (1024): 412.54507384895
> SOR (100x100): 468.38963398340786
> Monte Carlo : 26.556732682957968
> Sparse matmult (N=1000, nz=5000): 412.0465286948107
> LU (100x100): 668.2982766105255

You will see the problem sizes as 1048576, 1000x1000, N=100000,
nz=1000000 and 1000x1000 when you specify the "-large" option.
 Kazuyuki Shudo	shudo@computer.org	http://www.shudo.net/