From: MikkelFJ <mikkelfj-anti-spam@...>
Date: 2002-09-04T10:03:32+09:00
Subject: Re: Best GC for Ruby?


> Out of curiosity, have you given this a shot to see how performance
> looks? I've been debating the merits of generational collection, but
> I've been worried about the performance impact (and the code hassle)
> of maintaining generation barriers.

Take a look at the papers posted for ocaml garbage collection. It is quite
efficient. It is impacted by a write barrier frequently making it faster to
write new data than updating references. New data is very cheap.
There are two generations: 50% space is young generation. It is
incrementally copied and compacted. The other half is a traditional
mark/sweep for old data.
collection is triggered when enough space is wasted and a small amount is
collected.
When comparing to various complex colouring principles the OCaml GC is both
simple and efficient.

see the gc section on the page.

http://caml.inria.fr/ocaml/papers.html

On the same page it is worth reading the Zinc experiment.
It is the foundation of the efficient bytecode interpreter (which in turn is
the foundation for the efficient native code compiler). The ZINC experiment,
an economical implementation of the ML language (10 years+ old, but still
interesting).

Mikkel