[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-09-02 Thread Derek Munneke
BoD wrote: > Funny that today google also had something to say about javascript and VMs: > > http://blogoscoped.com/google-chrome/13 > > for those that missed this link in the Chrome thread: http://blog.beuchelt.org/2008/09/02/Googles+Path+To+World+Domination.aspx > it seems that Germany has al

[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-09-02 Thread Michael Neale
yeah - if there are calculations being performed in js - sure, this will help lots. But from my experience building large hairy ajax apps, its *still* the dom manipulation that holds things back for most everyday things. But yes, in future, that could well be a bottleneck too. On Sep 2, 8:36 pm,

[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-09-02 Thread Casper Bang
I understand what you mean but also seem to have noticed JavaScript/ Ajax really starting to push the envelope and deliver visuals ironically enough originally envisioned for the parent language Java. Case of point: http://www.classycode.co.uk/CoverFlow/ /Casper On Sep 2, 5:26 am, Michael Neale

[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-09-01 Thread Michael Neale
yeah their description of JIT may have made sense once, but any jit that goes beyond the initial stage of timing methods would go beyond it pretty quickly. This is also a good example of a straw man argument (if it was an argument). On the subject though, I don't think its JS performance that hu

[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-09-01 Thread BoD
Funny that today google also had something to say about javascript and VMs: http://blogoscoped.com/google-chrome/13 BoD sherod wrote: > Java gets the (usual) dishonorable mention with Mozilla implying that > their tracing JIT compiler is superior to the JIT solution in Java. > > Anybody with st

[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-09-01 Thread Tom Hawtin
On Aug 31, 11:44 am, sherod <[EMAIL PROTECTED]> wrote: > *Start quote* > the runtime of the method. A whole-method compiler, however, has to > always analyze and translate the entire method, even if parts of it > are not particularly “compilation-worthy”. So, a little more code has been compile

[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-08-31 Thread Casper Bang
> Hotspot is *not* a traditional JIT compiler. While that was true in   > the early days of HotSpot (many years ago) today it does lots of   > advanced realtime prediction and analysis.   Well I think that's kind of the point, nobody is dissing the performance of hotspot, they are (in the white p

[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-08-31 Thread Joshua Marinacci
Hotspot is *not* a traditional JIT compiler. While that was true in the early days of HotSpot (many years ago) today it does lots of advanced realtime prediction and analysis. I would say that HotSpot is the most advanced general purpose VM today, and it's constantly being improved. Where

[The Java Posse] Re: Tracing JIT compiler for Firefox

2008-08-31 Thread Casper Bang
Yeah Firefox 3.1 JavaScript performance is going to be ridiculously fast. There's a white paper on the technique here, surprisingly (and like JavaScript itself) derived from the Java world: http://www.usenix.org/events/vee06/full_papers/p144-gal.pdf It makes you wonder about whether this has para