Return-Path: sinclair@uk.ac.gla.dcs Return-Path: Received: from hawaii.dcs.gla.ac.uk by goggins.dcs.gla.ac.uk with LOCAL SMTP (PP) id <22670-0@goggins.dcs.gla.ac.uk>; Wed, 30 Jun 1993 21:33:08 +0100 To: partain@dcs.gla.ac.uk Subject: Something for your NoFibs... Date: Wed, 30 Jun 93 21:33:05 +0100 Message-ID: <6210.741472385@dcs.gla.ac.uk> From: Duncan Sinclair Hi, I've taken a wee hack at my solid modelling program, integrating the hbc and glhc versions into one. And it all seems to work... I've now got a program that can run for half an hour in 96M of memory and produce a beautiful picture as output. Or else it can run for about quarter the time in half the memory, or a sixteenth the time.... The faster the program, the less beautiful the output. In it's default state, it runs for just less that a minute, in 8Meg of memory, on a SS-10. The hbc-compiled version, using -gc-gen, also 8Megs of memory takes about three minutes... The output is identical. I'm sure it's got some lurking space leaks in it, but I spend a good long while last year getting rid of some real whoppers. I think I cut the space*time numbers by a factor of well over a hundred, so I'm pretty happy. Anyhow, I thought it might make a nice curiosity for you... What should I do with it now? Is there anything I need to do to it to make it more commpatible with your benchmark stuff, etc..? You can check it out in ~sinclair/Work/fulsom . Duncan.