Source document load time: 0 milliseconds
Stylesheet document load time: 27.01 milliseconds
Stylesheet compile time: 8.515 milliseconds
Stylesheet execution time: 0 milliseconds
The timing figures you obtained look bogus. Loading a 6MB
document in less than 1 ms is not possible with current
parsers/hardware, equally performing a transfrom is in less than
1ms for such a document is highly unlikely.
I would suggest the problem is the MSXML command line -t option.
Normally to obtain reliable figures you need to perform a
transform multiple times and average the results. Even in this
case disk cache issues mean document loading times can vary
significantly. If you are interested I have some code that
should enable you to benchmark your performance using MSXML.
Mail me off list for this.
I suspect your problem however is an algorithmic issue in your
stylesheet that you are not seeing. One way to identify this
type of issue is to employ a profiler on different input sizes
and look for non-proportional changes for individual templates
or specific lines. I don't think you can do this directly
against MSXML but other processors support this. See catchXSL
for one example or if you can package your stylesheets and a
sample input document I can do it against the Sarvega XSLT
processor for you.
Kevin Jones
Sarvega Inc.
On Monday 22 September 2003 21:33, Dipesh Khakhkhar wrote:
Hi,
Thanks a lot for replying.
Well as I said it as a discrepancy, i mean even if the timing
shown on the Windows 2000 server was less, it took longer than
on the windows XP even if server was having more memory.
The CPU Speed of the two machine are
Windows XP : Single Processor of 800 Mhz.
Windows 2000 server: Dual Processor each of 500 Mhz.
So even everything is more configuration wise why it took long
time (whereas it didn't showed the correct time with -t
command in MSXSL).
Thats why i found discrepancy.
Thanks once again for replying.
Regards
Dipesh
Date: Sat, 20 Sep 2003 09:45:39 +0200
From: "Dimitre Novatchev" <dnovatchev(_at_)yahoo(_dot_)com>
Subject: [xsl] Re: Relation between Memory /Time Problem and
OS ??
My input file is of 6.09 MB and I ran the xsl on two
different OS and was surprised with the result. Here are the
result.
Why should there be anything surprising?
Your two platforms were:
Time on Windows XP Desktop System with P3 Processor and 512
RAM
and
Time on Windows 2000 Server System with P3 Processor and 1.5
GB RAM
But you missed to provide very important data -- the CPU speed
of the two P3-s -- it can be quite different.
Also, the second platform has thrice the memory of platform 1.
Most probably on platform one the RAM was insufficient,
therefore swapping and thrashing occured.
On platform 2 the memory was three times more, there was no
swapping, (the CPU speed was probably faster) so it took
dramatically less time to complete.
Anyone who has encountered such discrepancy (atleast for me)
or know the reason for such behavior please throw some light
on this issue.
As explained above, this is not discrepancy, but a logical
fact.
=====
Cheers,
Dimitre Novatchev.
http://fxsl.sourceforge.net/ -- the home of FXSL
XSL-List info and archive:
http://www.mulberrytech.com/xsl/xsl-list
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list