Performance

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Performance

Srinivas JONNALAGADDA
James,

        Currently, there does not seem to be any in-line caching implemented, unless I have totally missed it somewhere.  The selector look-up mechanism may not provide many hints to the JVM to optimise that part.  Or, does it?  Do you plan on implementing explicit monomorphic/polymorphic in-line caching?

        I think that for Redline (Smalltalk, in general) to be accepted into the enterprise world, overall real-life performance in terms of time taken cannot exceed 4x-5x of that of Java -- at least for typical tasks.  Do you think that Redline will be able to target that bracket?  Thanks.

-- |0|0|
Reply | Threaded
Open this post in threaded view
|

Re: Performance

James Ladd
|0|0|,

You are correct in that there is no in-line caching implemented in Redline
at present. It is expected that optimization will happen at a later stage,
after the Runtime is complete.

There are many opportunities to improve the performance of Redline using
techniques like those you have suggested, as well as making use of
InvokeDynamic calls within the JVM.

Would you like to help with the optimization?

- James.

On Sun, Dec 16, 2012 at 6:07 PM, Srinivas JONNALAGADDA <[hidden email]> wrote:
James,

        Currently, there does not seem to be any in-line caching implemented, unless I have totally missed it somewhere.  The selector look-up mechanism may not provide many hints to the JVM to optimise that part.  Or, does it?  Do you plan on implementing explicit monomorphic/polymorphic in-line caching?

        I think that for Redline (Smalltalk, in general) to be accepted into the enterprise world, overall real-life performance in terms of time taken cannot exceed 4x-5x of that of Java -- at least for typical tasks.  Do you think that Redline will be able to target that bracket?  Thanks.

-- |0|0|