Re: [v8-users] Digest for v8-users@googlegroups.com - 7 Messages in 2 Topics

[ahta trading]

  Today's Topic Summary

Group: http://groups.google.com/group/v8-users/topics

• What causes a function to be "optimized too many times"? How do I avoid it? [2 Updates]
• Does changing the type passed to a constructor change a hidden class (and other queries)? [5 Updates]

 What causes a function to be "optimized too many times"? How do I avoid it?

Vyacheslav Egorov <veg...@chromium.org> Nov 06 08:04AM -0800  

Hi Kevin,
 
Does it deoptimize?
 
I do not see any deoptimizations in the log you have attached, were
you running with --trace-deopt?
 
Vyacheslav Egorov
 
 


 

Kevin Gadd <kevin...@gmail.com> Nov 06 04:16PM -0800  

Hi Vyacheslav,
 
Yeah, as I said I ran with trace-opt, trace-bailout and trace-deopt turned
on. So 'disabled optimization for' doesn't mean the function is
deoptimized? That's really surprising to me, because I see a performance
hit for those functions, and I assume that optimization being turned off
would mean that the functions would have to run using unoptimized JIT
output. That's not the case then? Does that mean that this error message
doesn't matter, and it's intended that these functions keep getting
recompiled until they hit the limit?
 
It would be cool to know how to find out why the functions keep getting
marked for recompilation, since the compiles seem to be taking time, but I
guess that's less of an issue.
 
Thanks,
-kg
 
 
 
--
-kg


 

 Does changing the type passed to a constructor change a hidden class (and other queries)?

Wyatt <deltaba...@gmail.com> Nov 06 12:27PM -0800  

With the follow code one hidden class is created:
 
function Point(x, y) {
this.x = x;
this.y = y;
}
var p1 = new Point(11, 22);
var p2 = new Point(33, 44);
 
Will p2.x="aString"; change its hidden class?
 
Will p2.x=undefined; change its hidden class?
 
Will p3= new Point(42,"theAnswer"); create a new hidden class?
 
I'm inclined to think that the answer is yes for each case..?
Or at least each of these cases seems as if it could not be fully optimized.
 
Any help is much appreciated!


 

Vyacheslav Egorov <veg...@chromium.org> Nov 06 02:43PM -0800  

The answer is no for each case.
 
V8 does not track types of values assigned to a named properties.
 
Vyacheslav Egorov
 
 


 

Wyatt <deltaba...@gmail.com> Nov 06 03:44PM -0800  

Interesting! But wouldn't each of these cases nullify any assumptions
made by the type-specializing JIT?


 

Wyatt <deltaba...@gmail.com> Nov 06 03:48PM -0800  

Actually these cases would probably just trigger a recompilation
(if the type-specializing JIT was invoked in the first place).
 
I shall now go re-read your last five blog posts.


 

Vyacheslav Egorov <veg...@chromium.org> Nov 06 03:57PM -0800  

In V8 currently most assumptions are made and checked at uses, not at
definitions.
 
Consider for example:
 
var p = new Point(1, 2);
 
function add(p) {
return p.x + p.y
}
 
add(p);
 
Here the fact that p.x and p.y is numbers is checked at the +
operation. If add is optimized for these assumptions and you pass
point that contains strings in x and y then function add will
deoptimize. But this will happen when you execute add not when you
create point with string values in x and y.
Vyacheslav Egorov
 
 


 

You received this message because you are subscribed to the Google Group v8-users.
You can post via email.
To unsubscribe from this group, send an empty message.
For more options, visit this group.

--
v8-users mailing list
v8-u...@googlegroups.com
http://groups.google.com/group/v8-users