You are viewing a plain text version of this content. The canonical link for it is here.
Posted to j-users@xerces.apache.org by Rob Outar <ro...@ideorlando.org> on 2003/01/23 16:19:36 UTC
Curious benchmarking results
I recently used JProbe to do some benchmarking using Xerces 2, the results I
got were sort of confusing.
Results:
Description
Parsing time Memory usage
Description
Parsing Time
Memory Usage
Xerces 1.4.3
Validation and Namespace turned off
1:40
14879 kilobytes
(14.5 megabytes)
Xerces 1.4.3
Validation and Namespace turned on
2:56
34138 kilobytes
(33.3 megabytes)
Xerces 2
Validation and Namespace turned off
1:48
13987 kilobytes
(13.6 megabytes)
Xerces 2
Validation and Namespace turned on
3:20
14028 kilobytes
(13.7 megabytes)
There was not much memory difference with validation and namespaces on/off
for Xerces 2. What I did notice is when validation and namespaces were on
for Xerces 2, the garbage collector was invoked MANY, MANY, MANY, times..
this accounted for the low memory usage. Do these results seem valid?
Let me know your thoughts,
Rob
Re: Curious benchmarking results
Posted by Andy Clark <an...@apache.org>.
Rob Outar wrote:
> I recently used JProbe to do some benchmarking using Xerces 2, the
> results I got were sort of confusing.
Running a program from within JProbe is a good way
to find memory leaks and extraneous object use but
is not a good way to measure performance. I think
it can show very different results when run outside
of this kind of tool.
--
Andy Clark * andyc@apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: xerces-j-user-unsubscribe@xml.apache.org
For additional commands, e-mail: xerces-j-user-help@xml.apache.org
RE: Curious benchmarking results
Posted by "Dr. Robert Spiske" <sp...@chemie.uni-halle.de>.
At 09:19 24.01.2003 -0500, you wrote:
>no comments anyone? I would like to know if my results make sense...
Ok, i'll try one.
We expect that the validation takes time, so using validation should take
longer, as we see.
Since you need to read the DTD/XSD to resolve the entities, the memory for
the DTD is needed anyway. Savin no memory there.
It is a basic optimisation principle that performance and memory
consumption correlate.
So reducing the memory usage will result in lower performance or better
performance can be reached if you use more memory (e.g. by caching). We see
that with your results too.
So there is nothing confusing about it.
I guess, the different behavior could have something to do with using DOM
or SAX for validation.
Hope that helps
Robert
---------------------------------------------------------------------
To unsubscribe, e-mail: xerces-j-user-unsubscribe@xml.apache.org
For additional commands, e-mail: xerces-j-user-help@xml.apache.org
RE: Curious benchmarking results
Posted by Rob Outar <ro...@ideorlando.org>.
no comments anyone? I would like to know if my results make sense...
Thanks,
Rob -----Original Message-----
From: Rob Outar [mailto:routar@ideorlando.org]
Sent: Thursday, January 23, 2003 10:20 AM
To: xerces-j-user@xml.apache.org
Subject: Curious benchmarking results
I recently used JProbe to do some benchmarking using Xerces 2, the results
I got were sort of confusing.
Results:
Description
Parsing time Memory usage
Description
Parsing Time
Memory Usage
Xerces 1.4.3
Validation and Namespace turned off
1:40
14879 kilobytes
(14.5 megabytes)
Xerces 1.4.3
Validation and Namespace turned on
2:56
34138 kilobytes
(33.3 megabytes)
Xerces 2
Validation and Namespace turned off
1:48
13987 kilobytes
(13.6 megabytes)
Xerces 2
Validation and Namespace turned on
3:20
14028 kilobytes
(13.7 megabytes)
There was not much memory difference with validation and namespaces on/off
for Xerces 2. What I did notice is when validation and namespaces were on
for Xerces 2, the garbage collector was invoked MANY, MANY, MANY, times..
this accounted for the low memory usage. Do these results seem valid?
Let me know your thoughts,
Rob