Any major performance change with I/O from JDK 6 that one should be aware of?

[Tomas Hurka]

Hi Fabrizio,

On 14 May 2012, at 13:05, Fabrizio Giudici wrote:
>> Thanks, I think see what's going on. Can you try on exclude MappedByteBuffer.load from compilation and re-run your tests to see if that "fixes" the problem, ie:
>> 
>> cat > .hotspot_compiler << EOF
>> exclude java/nio/MappedByteBuffer load
>> EOF
>> 
>> You'll know if you've excluded it if you get a message such ### Excluding compile ... message.
> 
> I was going to ask you something similar, as after looking at the sources of MappedByteBuffer I think there could be a over-optimization. But I'm definitely not a HotSpot expert.
> 
> Bingo. See console log below, in any case I'm getting now a similar performance than Apple JDK 6 and I see disk reads in the system monitor.
> 
> Since this seems to workaround the problem for my app, how can I reliably set the .hotspot_compiler stuff? I mean, is there any -XX option that can be used as well?
Use -XX:CompileCommand. See <http://cleversoft.wordpress.com/2010/07/21/xxcompilecommandexclude-jvm-option/> for more details.

Bye,
--
Tomas Hurka   <mailto:tomas.hurka at oracle.com>
NetBeans Profiler http://profiler.netbeans.org
VisualVM http://visualvm.java.net
Software Developer
Oracle, Praha Czech Republic