From news-rocq.inria.fr!univ-lyon1.fr!in2p3.fr!swidir.switch.ch!nntp.coast.net!howland.reston.ans.net!plug.news.pipex.net!pipex!tube.news.pipex.net!pipex!dish.news.pipex.net!pipex!soap.news.pipex.net!pipex!usenet Mon Jan 15 10:36:04 1996 Article: 3550 of rec.games.corewar Path: news-rocq.inria.fr!univ-lyon1.fr!in2p3.fr!swidir.switch.ch!nntp.coast.net!howland.reston.ans.net!plug.news.pipex.net!pipex!tube.news.pipex.net!pipex!dish.news.pipex.net!pipex!soap.news.pipex.net!pipex!usenet From: Robert Macrae Newsgroups: rec.games.corewar Subject: Re: Detecting imps? Date: 14 Jan 1996 13:02:50 GMT Organization: UnipalmPIPEX server (post doesn't reflect views of UnipalmPIPEX) Lines: 90 Message-ID: <4dautq$qoj@soap.news.pipex.net> References: NNTP-Posting-Host: ak216.du.pipex.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 1.1 (Windows; I; 16bit) akemi@netcom.com (David Boeren) wrote: >basic problem is that the trigger is unlikely to be pressed by an imp >spiral during the short (couple thousand) cycles which the bombing phase >lasts. I new I'd seen an anti-imp specialist somewhere :) James Layland entered Ivscan for one of the past tournaments (you still out there James?) and it has a scan followed by a custom Imp-stunner if appropriate. I am taking the liberty of posting his code since it is in the tournament archives. ;redcode-94 ;name ivscan ;author J.Layland ;strategy Kill imps and vamps. Otherwise die. org start ;macro ; Try very hard to kill imps and vamps. Die horribly when faced with ; anything else. Probably choke if I see a quick-cmp like Jazz. ; I had wanted to submit an anti-imp paper warrior in this round, and ; save this for the big coresize, but I couldn't tune my paper to ; reliably kill both imps&vamps at the same time, especially while trying ; to keep a slim startup to avoid a quick-cmp. I need another week... ; Steve Morrell's imps seem to be relatively immune to my anti-imp paper ; attempts. I'm not sure why. ; This scans for non-zero core, throws away small b-fields (+-1) and ; assumes it found either an imp or a fang. Wrong guesses will probably ; lead to stupid behavior. ; This could be slimmer, but I'm so big it probably wouldn't make ; much difference without radical recoding. And time. ; I've had major version-control problems developing this. Hopefully ; this is actually a version that works... test equ step+2000 step dat -7, -7 start add step, loc scan jmz.f -1, @loc mov @loc, test add #1, test slt #2, test jmp start loc seq @-33, -33 jmp avamp impsize mov @loc, 0 mov split, @loc add impsize, loc djn -2, #100 split spl #0, <-15 mov gate, <-12 jmp -1, <-17 gate dat <-17, <-16 avamp sub @loc, loc jmz.f @loc, start mov gate, @loc spl start spl 1 mov -1, 0 spl 1 l001 mov #6, 6 ;set source p001 mov