(I'm sorry about crossposting, but I think this forum is better suited for my particular question)
I'm currently doing some research regarding the adaptive optimization system in Java VMs.
I've looked through some documentation available on HotSpot, and searched this forum, but couldn't find any detailed technical information about the approach HotSpot takes in terms of sampling to identify hot methods, initial code generation and progressive optimization.
Does anyone know if and where such information is publicly available? Guidelines as to where to look in the available source code are also appreciated...
My main interest lies in the progressive optimization; does it use a number of predefined optimization levels like other VMs (such as JikesRVM) do, or does it use a more profile-guided approach which causes different optimizations to be applied depending on which method is being recompiled?
Any comments, pointers to related work and/or papers or suggestions are highly appreciated.