On Apr 20, 2008, at 2:23 PM, James Abley wrote:

> I think it's safe to say that there are some smart guys at
> Redmond working on those sorts of technologies.

Actually, since I was in Redmond last January for three days talking  
personally to those guys at Lang.NET, I feel safe to say they have  
shelved the JIT for several years, and their optimizations have not  
kept pace with those in the JVM.  (See my blog entry previously  
mentioned.)

Case in point:  Some of the CLR customers at Lang.NET were asking for  
vectorized loops.  Nobody could help them, because nobody was working  
on loops in the JIT.  Meanwhile, HotSpot recently improved its  
benchmark scores in part by vectorizing some common loops.

A significant number of HotSpot techniques have no CLR equivalent,  
especially those which depend on profiling and deoptimization.  The  
CLR JIT compiles at load time, and never looks back.

(For a long list, see http://wikis.sun.com/display/HotSpotInternals/ 
PerformanceTechniques )

According to people who use it in CLR, tailcall is uncomfortably  
slow.  Serrano's CLR version of BigLoo turns tailcalls off by default  
as a result.  Looks like a neglected stepchild feature to me.

As far as tailcall on the JVM goes, I know at least one researcher  
who is working on it; I wish we had it yesterday....  See my blog for  
how it will probably work.

-- John

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "JVM 
Languages" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at 
http://groups.google.com/group/jvm-languages?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to