[virtmach] Threaded vs Bytecode VM

[Eric Marsden]

>>>>> "tf" == thaddaeus frogley <thaddaeus.frogley@creaturelabs.com> writes:

  tf> Could somebody clarify for me what a "threaded" VM actually is,
  tf> as I'm not familiar with the terminology. (I know what threading
  tf> is, but how does it apply in this context?)

Threaded code (nothing to do with lightweight processes) is an
implementation technique for virtual machines; see

   <URL:http://www.complang.tuwien.ac.at/forth/threaded-code.html>

There was also some implementation discussion in comp.compilers in
1991 (the archives are available at iecc.com).

      
>>>>> "mc" == Mats Carlsson <matsc@sics.se> writes:
      
  mc> Does anyone have pointers to papers reporting this kind of
  mc> analysis? I'm working with VM implementations of Prolog, and all
  mc> the Prolog VMs I know of do these optimizations, but in a more
  mc> or less ad-hoc way. I have never seen any papers reporting a
  mc> methodical analysis.

Ian Piumarta describes a technique for identifying common bytecode
sequences dynamically (which provides significantly better performance
than doing the analysis statically) at

   <URL:http://www-sor.inria.fr/publi/ODCSI_pldi98.html>

-- 
Warning: This article may be fatal if swallowed.