RE: [xml] Benchmark...

["Ron Ohmer" <rohmer aisconsulting net>]

Of course that is the problem with benchmarks..

In order to make them @ all valid, you have to call everything the same
way..

Which of course does not take into account API differences in versions and
different API families..

So basically all it tells you is:
If you call an API this way: X; in the following APIs: y; you get Z..

Which is great, if you plan on only using said API the way they use it in
the benchmark.  

You see the same thing with Databases..  Oracle states that it is the
fastest, yet IBM states that DB2 is the fastest...  They are both right.
When you contrive the test in a certain way to allow for optimizations of
each product.

Well, enough ranting on Benchmark data..  The important things are:
1. Looks like Daniel showed below, that 2.6.5 is faster that 2.4.3 using
valid LibXML testing.  Which means LibXML is only getting better.
2. In *EVERY* benchmark I have seen, LibXML kicks the competition's butt.

Therefore in my educated, and enlightened opinion, LibXML Rocks! ;-)

Thanks Daniel!
Ron

-----Original Message-----
From: xml-admin gnome org [mailto:xml-admin gnome org] On Behalf Of Daniel
Veillard
Sent: Thursday, February 12, 2004 5:47 AM
To: Fabrice Desré FTR&D/DTL/TAL
Cc: xml gnome org
Subject: Re: [xml] Benchmark...

On Wed, Feb 11, 2004 at 08:20:54AM +0100, Fabrice Desr? FTR&D/DTL/TAL wrote:
> Of course benchmarks are only lies, but this one is not that bad :
> http://xmlbench.sourceforge.net/
>
> libxml/libxslt performs very well, and one thing to note is that libxml 
> 2.6 seems to be slower than libxml 2.4.

  So far, I can't reproduce this with just xmllint, I will have to
look more closely at their code and data to see if it exhibit some
funky behaviour. This may be due to the fact that they use the interfaces
based on SAX v1 instead on SAX v2 with 2.6, getting the most out of 
the new code requires to change the API used slighly, I need to write
docs about this.

server:~/libxml2-2.4.30 ->  ls -l ../XML/db10000.xml
-rw-r--r--    1 veillard vcsa      2009238 Oct  6 18:06 ../XML/db10000.xml
server:~/libxml2-2.4.30 -> ./xmllint --noout --timing ../XML/db10000.xml
Parsing took 536 ms
Freeing took 161 ms
server:~/libxml2-2.4.30 -> ./xmllint --noout --memory --timing
../XML/db10000.xml
Parsing took 537 ms
Freeing took 166 ms
server:~/libxml2-2.4.30 ->
 

 same box latest version compiled the same way:
 

erver:~/libxml2-2.6.5 -> ls -l ../XML/db10000.xml
-rw-r--r--    1 veillard vcsa      2009238 Oct  6 18:06 ../XML/db10000.xml
server:~/libxml2-2.6.5 -> ./xmllint --noout --timing ../XML/db10000.xml
Parsing took 432 ms
Freeing took 143 ms
server:~/libxml2-2.6.5 -> ./xmllint --noout --memory --timing
../XML/db10000.xml

Parsing took 431 ms
Freeing took 144 ms
server:~/libxml2-2.6.5 ->

  Further examination needed :-)

Daniel

-- 
Daniel Veillard      | Red Hat Network https://rhn.redhat.com/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/
_______________________________________________
xml mailing list, project page  http://xmlsoft.org/
xml gnome org
http://mail.gnome.org/mailman/listinfo/xml