This is the mail archive of the
guile@sources.redhat.com
mailing list for the Guile project.
Re: guile-vm-0.0
- To: Doug Evans <dje at transmeta dot com>
- Subject: Re: guile-vm-0.0
- From: cwitty at newtonlabs dot com (Carl R. Witty)
- Date: 01 Aug 2000 12:30:36 -0700
- Cc: Marius Vollmer <mvo at zagadka dot ping dot de>, Keisuke Nishida <kxn30 at po dot cwru dot edu>, guile at sourceware dot cygnus dot com
- References: <m3d7jxa9gv.fsf@indy.STUDENT.CWRU.Edu> <87wvi55di3.fsf@zagadka.ping.de> <m3zon0u2r7.fsf@indy.STUDENT.CWRU.Edu> <m3bszgth68.fsf@indy.STUDENT.CWRU.Edu> <87ittniykx.fsf@zagadka.ping.de> <m37la3fnbh.fsf@indy.STUDENT.CWRU.Edu> <87aeez8iu8.fsf@zagadka.ping.de> <v4jk8e1voke.fsf@bogomips.newtonlabs.com> <14726.5488.702952.806395@casey.transmeta.com>
Doug Evans <dje@transmeta.com> writes:
> Carl R. Witty writes:
> > Marius Vollmer <mvo@zagadka.ping.de> writes:
> >
> > > Wow, I didn't expect this. When I first read about GCC's support for
> > > computed gotos and that they canb be used to write very fast virtual
> > > machines, I thought, come on, how much can that be? I mean, a switch
> > > isn't exactly slow either. But...
> >
> > For one thing, you can't get rid of the bounds-check on a switch.
>
> And each switch case branches directly to the next
> switch case without having to go back up to the top
> and down again.
Of course, there's no reason a compiler couldn't do that with a
while (1) { switch (...) { ... } }
structure. (I've never seen a compiler do that particular
optimization, though.)
Carl Witty