Team PLClub ICFP entry -- comparing the performance of OCAML and SML

[Matthias Blume]

From: Henry Cejtin <henry@sourcelight.com>
Subject: Re: Team PLClub ICFP entry -- comparing the performance of OCAML and SML
Date: Fri, 13 Oct 2000 04:04:40 -0500

> I wonder what in gods name is going on with SML/NJ paying such a  huge  price
> for  array  subscript  checking.  This is REALLY horrible since it encourages
> people to use unsafe subscripts.

AFAIK, no optimizations are even _attempted_ as of yet on array
subscript checks.  Worse, the infrastructure for better things to come
(aka "the new array representation") has been put into place without
also putting those better things into place.  As a result we currently
end up with a double-indirection through memory on every array access,
safe or no.  The missing subscript optimization therefore hurts us
twice.  (When using unsafe subscripts it still hurst us once.)

Matthias