This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: gcj Linpack performance
- To: Florian Schintke <schintke at zib dot de>
- Subject: Re: gcj Linpack performance
- From: Jeff Sturm <jsturm at one-point dot com>
- Date: Tue, 10 Jul 2001 15:39:20 -0400 (EDT)
- cc: Manfred Hollstein <manfred dot h at gmx dot net>, Tim Prince <tprince at computer dot org>, gcc at gcc dot gnu dot org
On Tue, 10 Jul 2001, Florian Schintke wrote:
> [Manfred Hollstein]
> > IIRC, -fno-bounds-check is having the biggest effect on performance.
>
> Also this flag with some combination with others don't gave me
> significantly better performance. The best was:
>
> with -O3 -fno-bounds-check
>
> Mflops/s: 24.719 Time: 1.74 secs Norm Res: 4.7
> Precision: 2.220446049250313E-16
Where did you get Linpack.java? My results are very different than yours:
[jsturm@toronto bench]$ gcj -O2 Linpack.java --main=Linpack -o linpack
[jsturm@toronto bench]$ ./linpack
Mflops/s: 76.296 Time: 0.01 secs Norm Res: 1.14 Precision:
2.220446049250313E-16
[jsturm@toronto bench]$ java Linpack
Mflops/s: 13.205 Time: 0.05 secs Norm Res: 1.43 Precision:
2.220446049250313E-16
[jsturm@toronto bench]$ java -version
java version "1.3.0"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.3.0)
Classic VM (build 1.3.0, J2RE 1.3.0 IBM build cx130-20000815 (JIT enabled:
jitc))
Jeff