Re: XSLT engine performance

[Paul Tchistopolskii <paul at qub dot com>]

----- Original Message ----- 
From: Matt Sergeant 

> On Wed, 26 Jul 2000, Sebastian Rahtz wrote:
> 
> > What is it with me and Oracle? 
> > 
> > I just ran a test to see how different XSLT processors coped with
> > reading a 3mbyte XML file and going over the tree twice:
> > 
> >                  real      user       sys
> > XT: 0m23.131s 0m19.030s 0m0.510s
> > Saxon: 0m35.125s 0m28.900s 0m0.730s
> > Sablotron: 0m47.187s 0m39.320s 0m1.300s
> > Xalan: 0m55.277s 0m50.540s 0m1.280s
> > Oracle: 6m56.811s 1m15.790s 0m10.920s
> > 
> > the interesting results here are
> > 
> >  - Oracle has some real problem in working well  for me :-}
> >  - Sablotron is slower than Saxon, despite being compiled C++
> > 
> > bizarrely, Xalan produced the results even though it (rightly) said my
> > XML file was not valid against the DTD. Is this to be expected?
> > 
> > does anyone have reason to doubt the above ranking, apart from the odd
> > Oracle result? 

To me this seems very close to reality.
 
> I have no reason to doubt it, but I'd like to see some version numbers
> next to everything. I'd also like to see some more realistic tests, such
> as transforming smaller documents lots of times (like a dynamic XML web
> server might do).

It will take relatively big effort to produce  'the more 
realistic tests'  you are talking about. Each engine should 
be executed  in servlet mode. Engines,  which allow 
pre-compilation of the stylesheet should be invoked in 
this 'pre-compiled stylesheet mode'. ( And I think the results 
will be that XT + XP will be still twice as fast as any other 
engine . he-he. ).

If you are ready to prepare such 'realistic testcase'  
for Sablotron and then you will provide me with the 
detailed installation instructions - I could produce the 
similar benchmark for XT and run both benchmarks on 
windows box. Or I can provide you with detailed 
installation instructions so that you could run 
XT stuff on some UNIX box.  

The funny problem is that  benchmarking Linux solution 
on Windows has no sense ( lack of 'good' Apache ). 
On another hand benchmarking Java solution on Linux 
also has no sense ( lack of  HotSpot ).

Whatever - if you like to spend your time preparing the 
'realistic test' to 'compare'  XT and Salbotron - feel free 
to write me and I will 'prepare'  XT part. If you are not 
interested - forget it, please - I think we both have some 
stuff to do other than preparing complex real-life 
benchmarks.

Rgds.Paul.



 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list