Re: XPP3 optimization

Ville Skyttä wrote:

>for some time now, I wanted to check if using final local variables in
>Java makes a difference performance-wise.  XPP3 seemed like a lib of
>suitable size, so I went ahead and tried it out.
>
>The result is that it indeed seems to be a slight performance
>improvement, see the attached patch against xpp3-1.1.2a and the saxbench
>results, comparing vanilla (marked w/MXP1 beta1) and patched (MXP1-OPT)
>1.1.2a measured on my box, a P4 @ 1.95GHz and Sun's J2SE 1.4.1_02.
>
hi Ville,

looking on results i have not seen big performance improvments (+/-1-5% 
right?) but i have applied some chnages:

did you have maybe chance to run pqrser under profiler and identify what 
are hot spots?

>Feel free to apply if you think it's ok.  Here's a summary of the patch:
>- final'ize local variables.
>
i think this has noimpact on performance aqs JITs are now smart enough 
to figure it out.

>- Don't instantiate Booleans.
>
this unfortunately can not be used as MIDP has no state Boolean.TRUE or 
Boolean.FALSE AFAIK.

>- Eliminate one unused variable.
>
hotspot probalbly could otimize it out during execution.

>- Add one initial size to a StringBuffer.
>
this is the most useful performance-wise optimizaton(!)

>Also attached is a similar patch against 1.1.2.1 xmlpull APIs.
>
i have finalized some of local variables in XmlPullParserFactory - it 
will make for safer code to maintain :-)

thanks,

alek

-- 
"Mr. Pauli, we in the back are all agreed that your theory is crazy. 
What divides us is whether it is crazy enough to be true." Niels H. D. Bohr



------------------------ Yahoo! Groups Sponsor ---------------------~-->
Get A Free Psychic Reading! Your Online Answer To Life's Important Questions.
http://us.click.yahoo.com/Lj3uPC/Me7FAA/uetFAA/2U_rlB/TM
---------------------------------------------------------------------~->

To unsubscribe from this group, send an email to:
[email protected]

 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/