You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@xalan.apache.org by Sebastien Sahuc <ss...@imediation.com> on 2000/10/20 16:48:37 UTC
RE: Benchmark of XSLT performance
> -----Original Message-----
> From: Stuart Roebuck [mailto:sr@adolos.com]
> Sent: Friday, October 20, 2000 2:53 PM
> To: cocoon-dev@xml.apache.org
> Subject: RE: Benchmark of XSLT performance
>
>
> Sebastian,
>
> On Friday, October 20, 2000, at 10:23 AM, Sebastien Sahuc wrote:
>
> > > Sebastien, how's it compare with the XalanJ1 version?
> >
> > Faster (twice as fast) when it comes to short XML document
> (1kb) but almost
> > the same when dealing with longer source (100kb). So still
> 3 times longer
> > than XT :-( in this case.
>
> Do you mean XalanJ2 is faster than 1 or 1 faster than 2.
See below.
> Running your benchmarks on MacOS X Beta (Java 1.2.2 +
> Hotspot) I got (readings from second 100 only):
With Hotspot 100 loops is definitely not enough. see below
>
> 1K Source 100K Source
> Xalan2 + Aelfred 30 2384
> Xalan2 + Xerces 40 2389
> Xalan1.2 + Xerces 23 355
> XT + Xerces 24 270
> XT + Sun parser 9 228
> XT + Aelfred 9 233
> Xalan2 default parser 39 2416
I believe it depend on the JVM use and on the number of loops. Below some
figures for 2 JVMs and 2*200 loops ...
1K source:
----------
Sun 1.2.2 with JIT Sun1.2.2 Hotspot2
Xalan2 + Aelfred 14 11
Xalan2 + Xerces 23 16
Xalan1.2 + Xerces 14 10
XT + Xerces 24 13
XT + SUN parser 12 14
XT + Aelfred 13 15
100K source:
----------
Sun 1.2.2 with JIT Sun1.2.2 Hotspot2
Xalan2 + Aelfred 668 462
Xalan2 + Xerces 617 464
Xalan1.2 + Xerces 360 260
XT + Xerces 208 172
XT + SUN parser 151 124
XT + Aelfred 195 133
All the best,
Sebastien
>
> Stuart.
>
>
> > Please download my benchmark at
> >
> > http://sebsahuc.citeweb.net/Xslbench.zip
> >
> > It's a autosufficient bench (every lib included), just need
> to launch
> >
> > run 100
> >
> > for testing the 1kb sample two times 100 loops (to make
> sure the code is
> > compiled when we reached the second set of 100 loops)
> >
> > and launch
> >
> > run2 100
> >
> > for benching the 100kb sample two times 100 loops.
> >
> >
> > The bench show different combinations of XSL processor and
> XML parser.
> >
> > Let's me know what are your thoughts on this. Thanks
> >
> > Sebastien
> >
>
> --------------------------------------------------------------
> -----------
> Stuart Roebuck, BSc, MBA Tel.: 0131 228 4853 / Fax.:
> 0870 054 8322
> Managing Director Alpha Numeric Pager:
> 07654 588898
> ADOLOS
http://www.adolos.com/