The Great Startup Problem

[Marcus Lagergren]

Hi Per!

This is mostly invokedynamic related. Basically, an indy callsite requires a lot of implicit class and byte code generation, that is the source of the overhead we are mostly discussing. While tiered compilation adds non determinism, it is usually (IMHO) bearable…

/M


 On 23 Aug 2014, at 21:45, Per Bothner <per at bothner.com> wrote:

> 
> 
> On 08/23/2014 12:25 PM, Per Bothner wrote:
>> On 08/22/2014 01:08 PM, Charles Oliver Nutter wrote:
>>> What are the rest of you doing to deal with these issues?
>> 
>> Start-up does not appear to a problem for Kawa:
> 
> I should mention I'm not using invokedynamic, and have no
> concrete plans to do so.  However, I am working on pattern-matching, which
> may make multi-methods more attractive, in which case invokedynamic
> may be more of a benefit.  InDy would probably also be helpful for arithmetic
> (though for performance-critical code it is better to make use of
> optional type specifiers and type inference)
> 
> I will be using MethodHandles extensively in a re-write of the
> "apply" mechanism that I'm working on.  This will replace the "switch"
> pattern I'm using, and should avoid the need for generating a 'ModuleBody'
> class in some cases.
> -- 
> 	--Per Bothner
> per at bothner.com   http://per.bothner.com/
> _______________________________________________
> mlvm-dev mailing list
> mlvm-dev at openjdk.java.net
> http://mail.openjdk.java.net/mailman/listinfo/mlvm-dev

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://mail.openjdk.java.net/pipermail/mlvm-dev/attachments/20140824/5f6a2269/signature.asc>