From corewar-ga-request@sunsite.auc.dk Sun Feb 15 00:09:57 1998 Received: from nez-perce.inria.fr (nez-perce.inria.fr [192.93.2.78]) by tobago.inria.fr (8.6.10/8.6.6) with ESMTP id AAA00947 for ; Sun, 15 Feb 1998 00:09:57 +0100 Received: from sunsite.auc.dk (qmailr@sunsite.auc.dk [130.225.51.30]) by nez-perce.inria.fr (8.8.7/8.8.5) with SMTP id AAA25241 for ; Sun, 15 Feb 1998 00:09:55 +0100 (MET) Received: (qmail 16171 invoked by uid 501); 14 Feb 1998 23:09:51 -0000 Resent-Date: 14 Feb 1998 23:09:51 -0000 Resent-Cc: recipient.list.not.shown:; Old-Return-Path: Delivered-To: corewar-ga@sunsite.auc.dk Message-Id: <3.0.1.16.19980214170959.10a7c9ec@nc5.infi.net> X-Sender: wtnewton@nc5.infi.net (Unverified) X-Mailer: Windows Eudora Light Version 3.0.1 (16) Date: Sat, 14 Feb 1998 17:09:59 To: corewar-ga@sunsite.auc.dk From: Terry Newton Subject: Re: System 4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Resent-Message-ID: <"l5xlNC.A.D8D.-Qi50"@sunsite.auc.dk> Resent-From: corewar-ga@sunsite.auc.dk X-Mailing-List: archive/latest/85 X-Loop: corewar-ga@sunsite.auc.dk Precedence: list Resent-Sender: corewar-ga-request@sunsite.auc.dk Status: RO Quoting George ... >Quoting Terry Newton (wtnewton@nc5.infi.net): >> At 09:21 AM 2/12/98 -0800, you wrote: >> >is anybody interested in my sys4? .. it's a ga system written in perl that >> >uses slightly different (and hopefully better) strategy then sys3 ... >> >> I'd like a copy. I probably won't be able to run it unless I get a >> perl up on this cranky dos/windoze thing but I'd like to study the >> code anyway. I ended up doing more than looking, found "Perl for Win32" and a little hacking later had it running. Didn't have to change much, mainly the pipe stuff for running pmars, forced exit on pmars error and no save on exit since so far the only way I've found to stop it is ctrl-crash. Nice idea. It begins with a random distribution of random warriors, then the stonger ones take over and spread out 2-dimensionally. Very A-Life-ish. I let it run all night and it made this... ;System 4 warrior ;name Rush (11,1) ;assert 1 L0 ADD.F > L3, # L7 L1 MOV.X $ L14, > L0 L2 CMP.BA $ L5, $ L6 L3 SUB.F $ L17, @ L7 L4 JMN.BA $ L11, * L19 L5 MUL.X $ L4, { L6 L6 CMP.AB * L16, # L2 L7 DJN.AB * 1177, @ L12 L8 ADD.B < 6494, $ L8 L9 MOD.F @ L0, > L5 L10 MUL.AB @ L15, * L18 L11 DIV.B { L5, $ L15 L12 ADD.A > L14, $ 5083 L13 DJN.F # 5066, < 6240 L14 MOV.B $ L15, $ 3341 L15 SUB.BA > L4, $ L2 L16 SPL.X # L11, < 4315 L17 DIV.X < L14, # L19 L18 DJN.BA < 7279, $ L10 L19 DJN.B * L15, < 7073 (Generation 58, coresize=8000, from a 12x12 grid running 30% change, 4% instruction change, 4% insertion/deletion. Lovely ain't it:) Ranks about 33% agaist a test set of 12 warriors, not great but it's a start. Barely beats FireStorm 1.1, Iron Gate and Tornado. This is an early sample, a few more days should bring improvement. >there are some programs that produce a bit of diagnostic output but not >very good .. good enough to see what's going on > it shouldn't be too hard to write some more diagnostic >software ... maybe graphical representations of the "earth" ... (yes in >this program the "earth" is flat and square:) I'll probably code a front-end or inspector for it in qbasic/batch, this is facinating. Later... Terry