This is the mail archive of the
guile@sourceware.cygnus.com
mailing list for the Guile project.
Re: librep's indirect threaded bytecode interpretter
- To: Michael Livshin <mlivshin at bigfoot dot com>
- Subject: Re: librep's indirect threaded bytecode interpretter
- From: karlheg at bittersweet dot inetarena dot com (Karl M. Hegbloom)
- Date: 22 May 2000 09:23:00 -0700
- Cc: guile at sourceware dot cygnus dot com
- References: <87g0rb6u4s.fsf@bittersweet.intra> <s3aehjyuan.fsf@bigfoot.com>
>>>>> "Michael" == Michael Livshin <mlivshin@bigfoot.com> writes:
Michael> karlheg@bittersweet.inetarena.com (Karl M. Hegbloom) writes:
>> Any chance that a similar byte code machine for Guile will ever be
>> implemented? I would really like Guile to be that quick.
Michael> would you like to volunteer? the only chance for something to be done
Michael> is somebody willing to do it.
That's a long way off, I'm afraid. I do not have the skills it would
require at this point in time. I don't fully understand Guile yet.
After a several months break from it, I'm back to my Lisp/Scheme
studies again... I'm reviewing right now, and soon will be moving
forward again with "Structure and Interpretation of Computer
Programs", "Lisp in Small Peices" and "Essentials of Computer
Languages" (MIT), amoung others. There's quite a lot left for me to
learn.
>> They also have a module system that is probably worth having a look
>> at.
Michael> it looks very much like the Scheme48 one, doesn't it?
That's what it says. I don't know how it works yet. I'm a newbie.
>>>>> "John" == John Harper <john@dcs.warwick.ac.uk> wrote:
Karl> I mentioned Rep on the Guile mailing list, and was told that Guile's
Karl> tree code interpretter should be faster than any byte code engine
Karl> there is. Ha! If they want to keep up, they'll need to implement a
Karl> threaded byte code machine also. It's a pretty neat trick, I'd say.
John> I agree, but thank Ceri Storey, who found the idea and did the
John> preliminary implementation :-)
John> actually, IIRC using the threaded interpreter only gets about a 10%
John> performance increase, there were some other optimizations as well
John> (and a 10% slowdown due to the module system; but it's well worth it!)