John Resig on TraceMonkey
ejohn.org
John Resig on TraceMonkey
1–10 of 21 posts
Re: John Resig on TraceMonkey
#2John does a really job of discussion what implications this had as well as how it does it. Great article.
Re: John Resig on TraceMonkey
#3Install the update and do the image manipulation test. For best results, run the test before and after. The difference is astounding.
Re: John Resig on TraceMonkey
#4Has anyone run a comparison between this new engine and SquirrelFish yet? It seems to me like two major ideas are competing:
1. Better JIT for interpreters. 2. Faster interpreters, period.
I'm really curious to see which has the faster ramp-up and which has the best end-game results.
Re: John Resig on TraceMonkey
#5It's funny, 2 months after I give up on GameClay (partially because of performance issues in the prototypes), this comes out. Maybe I should go back to it. Well, that's why I chose not to open-source the code and my cofounder's hanging on to the domain names.
Though any application to games needs to get over the fact that 90% of the casual game market still uses IE...
Re: John Resig on TraceMonkey
#6Install the update and do the image manipulation test. For best results, run the test before and after. The difference is astounding.
I thought "astounding" would be an overstatement. Turns out you're right.
Re: John Resig on TraceMonkey
#7check out Steve Yegge's video/transcript about dynamic language optimization in case you missed it:
http://steve-yegge.blogspot.com/2008/05/dynamic-languages-st...
Re: John Resig on TraceMonkey
#8I think the key point of this article over the other related article (http://news.ycombinator.com/item?id=284192) is that John points out that tracing will be able to be applied to DOM manipulation/property access in the future.
"Being able to trace through a DOM method would successfully speed up, not only, math and object-intensive applications (as it does now) but also regular DOM manipulation and property access."
To me, this means that JavaScript will finally be integrated with the browser in a way that is required to advance the web. I've always wondered what would be the "big thing" to lead up to the move towards the web even more, and this is definitely the answer.
The only thing left now is making it so JavaScript can be compiled (the only way I can think of how to accomplish preventing stealing code).
Re: John Resig on TraceMonkey
#9I think the key point of this article over the other related article ( http://news.ycombinator.com/item?id=284192 ) is that John points out that tracing will be able to be applied to DOM manipulation/property access in the future. "Being able to trace through a DOM method would successfully speed up, not only, math and object-intensive applications (as it does now) but also regular DOM manipulation and property acces…
speed up, not only, math and object-intensive applications (as it does now) but also regular DOM manipulation
Agreed. The sooner they get to that, the better!
Edit: As an aside, the web arguably wouldn't have happened without "stealing code". I see this as a strength not a weakness.
Re: John Resig on TraceMonkey
#10check out Steve Yegge's video/transcript about dynamic language optimization in case you missed it: http://steve-yegge.blogspot.com/2008/05/dynamic-languages-st...
Things have been moving in this direction for quite some time, I think. Here's a master's thesis from March 2006: