RE: performance of copying all source to result

Unfortunately all my benchmarks disagree with you. 

I measure only the time needed for the transformation itself (the time
needed to complete the line 

transformer.transform(source, result) ;

And I do it enough times in a loop so that it won't be dominated by the Java
VM start up.

Also, I checked this on Windows 2000, Solaris and HP unix (The Unix machines
are only a little faster). I checked with several XSLT processors and
several XML parsers. The length of the transformation is between 1500 ms and
4000 ms.

The only major improvement I saw was on Windows 2000, when I switched to JDK
1.3. The time drops to around 500-600 ms. But it is still not the 50-100 ms
you'd expect.

Therefore, I still suspect my stylesheet could be changed to provide the
same results faster.

Any suggestions?

	Evyatar

-----Original Message-----
From: Kay Michael [mailto:Michael.Kay@xxxxxxx]
Sent: Saturday, December 23, 2000 2:20 AM
To: 'xsl-list@xxxxxxxxxxxxxxxxxxxxxx'
Subject: RE:  performance of copying all source to result


> My source documents are relatively small (1 KB - 4 KB).
> 
> The question is - how long will it take (at run-time) to do this
> transformation?
> 
> My measurements show it will be more then 1.6 seconds (1600 
> milliseconds).
> 
Your measurements are probably dominated by the time taken to start up the
Java VM and the XSLT processor.

If you do lots of such transformations, I would expect them to take about
100ms each; if you do lots using the same stylesheet, about 50ms each.
Depending on many factors, such as your processor, your Java VM, the amount
of memory, etc.

Mike Kay 


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