[Mono-list] JVM performance: JVM as a basis for CLR

[Miguel de Icaza]

> Have you investigated HP's Dynamo project?  They got
> very good performance through dynamic re-optimization
> rather than static "compile once, run many" techniques.
> They start with simple code translation, and work their
> way up as a profile of the program's runtime is built.
> I believe Sun was trying something similar with HotSpot.

I have not investigated it, but thanks for the information.  I found
it in google and will be reading on it.

A few weeks ago, I spoke to David Mosberger-Tang at Usenix.  He has
been recently working on the IA64 for HP and wanted to do something
interesting with what apparently is a very advanced profiling system
built into the chip.

He wanted to introduce the same concept of recompiling hot zones
dynamically, and bytecode languages were very appropiate for this kind
of task.

Funny, he is at HP labs, the same place that produced those Dynamo
documents. 

Miguel.