[02:01] MSG: Quit: humhum [06:31] Join: fiveop joined #corewars [08:01] Join: Roy joined #corewars [10:30] Join: datagram joined #corewars [10:30] MSG: Client Quit [10:30] Join: datagram joined #corewars [11:16] Wow, a colleague just ate a bag of chips, a new change to win the ipod! [11:16] (didn't win the first time) :-( [12:07] Didn't win again...shoot.. :( [12:20] MSG: Ping timeout: 252 seconds [12:28] Join: Mizcu joined #corewars [12:33] 50000 randrys, 485 got to phase 4 [12:44] You are much to hard on them Miz :) [12:45] *Racist*.. [12:46] *bad-gene*.. [12:50] But, how is the process now? [12:50] Getting better versions? [12:50] probably over 140 in fshnop02 [12:59] fsh02? what? thats acient :) [13:00] 'cause i wont disclose current scores [13:02] Heh, we can't do much with a score [13:02] Just be amazed about its powers (or laught at it) [13:03] ... [13:03] 136 has been broken [13:05] though im running out of thing to optimize [13:36] Take my 20 constant warrior ;-) (with about 14 real constants) [13:37] Maybe even more constants (if you count decrements etc) [13:37] And even MORE if you include the qscan constants! [13:38] But hey, a warrior could max have 200 constants right, it isn't that much :) [13:56] ... [13:58] And it doesn't even have a lot of code.. [14:01] Join: fiveop_ joined #corewars [14:01] MSG: Client Quit [14:01] Join: fiveop_ joined #corewars [14:04] MSG: Ping timeout: 252 seconds [14:42] Join: elkauka joined #corewars [14:43] hi folks! happy new year (i konw its a little late...couldnt catch a computer earlier) [14:50] Join: Nenad joined #corewars [14:50] Hi [14:51] hi [14:58] hi all [14:58] elkauka: I see that vamps have died... do you have plans for some new warriors? [14:58] hi Roy [14:59] Nenad: had a few ideas about a "true" scanner ... and maybe add some extra anti imp power to the vamps and try again :P [15:00] Roy: what is Substratum? [15:01] Its...cool :) [15:01] And it kind of scans [15:01] it's the only thing on the hill that lost to my pathetic attempt of making a rezooom scanner [15:01] :) [15:05] heh bad :) [15:11] You are really hurting those scanners Nenad! [15:12] just wait until i send this scanner-killer *g* [15:14] well, DanceOfFallenAngels was designed as a scanner-killer - a proud descendant of Starfall... I was not wrong when I killed the old version [15:14] miz:stoney the fourth? [15:15] kauka: static rebuild [15:58] * Roy waves [15:58] MSG: [16:02] * Nenad waves to Roy [16:32] hrm. [16:32] greetings all. [16:35] Hi bvowk! [16:38] how's it going nenad? [16:39] nice. I'm using these few days to relax a little [16:40] lots of work to come in the next couple of weeks [16:40] you're a student right? [16:40] yes. [16:41] undergrad? [16:41] yes. math + cs. [16:42] which year? [16:42] 2nd year math + 3rd year cs [16:43] decided you weren't getting your RDA of numbers in CS? [16:43] ? RDA? [16:43] (sorry, symbols, there aren't numbers in math beyond the first year) [16:43] what is that? [16:43] heh.. [16:43] Recommended Daily Allotment.. [16:43] :) [16:44] how they term the dosages of vitamins for adults :) [16:44] as seen on the sides of cereal boxes [16:44] well, I always wanted to study math - but I didn't want to end up being a highschool math teacher... so I enrolled in cs, and when I saw that I had enough spare time for both... [16:45] I'm not sure yet what I'll be doing in a couple of years when I finish my studies [16:46] I hope that I'll get the chance to do some actual research, but we'll see... [16:46] well, I'm sure there's no shortage of demand for programmers who don't have bad math skillz [16:47] in research and out :) [16:47] I hope that it remains so until I graduate. :) [16:48] yeah.. [16:49] you ever noticed the windows installer's sliding bar isn't linear at the end.. you could fix that.. math is important.. [16:49] * bvowk ducks. [16:49] :) [16:49] how are you? what are you doing these days? [16:50] so when are you going to write an evolver? [16:50] right now I'm playing with circles.. [16:50] with he who must not be mentioned and pk [16:50] err... when I finish with the current exam stack. [16:50] but I want to try something better. [16:50] so 2008? [16:50] :) [16:51] I want to make a neural network and to teach it how to make redcode warriors. [16:51] in other words, a true "artificial redcoder" [16:51] heh [16:58] ... seems that I've got to go now. I'll be back tomorrow. [16:58] * Nenad waves [16:59] MSG: [17:03] /mw waves too [17:03] cya [17:03] Part: elkauka left #corewars [17:38] Join: Metcalf joined #corewars [17:38] Hi [17:39] hey met [17:39] how's it going? [17:40] Greetings bvowk. It's going fine for me thanks. How's it going for you? [17:40] pretty good [17:40] I'll go see at the library about your articles.. [17:41] sometime today.. [17:41] Thanks :-) [17:41] I think the issue has been resolved though [17:42] oh.. [17:42] well nevermind then! :) [17:42] http://en.wikipedia.org/wiki/Talk:Core_War [17:44] excellent [17:44] that page needs some information on evolvers.. [17:44] and tourneys [17:45] and spec [17:45] and interpreters [17:46] it needs alot [17:46] Yes [17:47] You're not about to edit it are you? [17:49] heh... [17:49] I'm not much of a write.. [17:49] writer.. [17:49] Okay, just wanted to check, I was just about to do some stuff to it [17:50] Or vaguely interestingly, see eg http://en.wikipedia.org/w/index.php?title=Talk:Core_War&oldid=33471019 before somebody decided to remove the comments making them look like an idiot :-) [17:51] lol... [17:51] Well, I suppose anyone can make a mistake. [17:52] That is William R Buckley, 3rd director of the ICWS and founding editor of TCWN [17:52] hey pack. [17:52] pak rather [17:54] It was the original edit summary calling people "unconscionable" for getting the date wrong which I found rather offensive... [17:54] Join: Roy joined #corewars [17:54] (Anyway, got to go home. Will read any further discussion sometime) [17:54] hi all [17:54] Hi Roy [17:55] * Metcalf waves to Phil [17:55] * Roy waves [17:57] I should finish my evolver faq [17:57] :-) [17:57] Still no new prime JM :-( [17:57] :-( [17:57] 48k the largest until now [17:57] Bvowk, is the progress so far online? [17:57] met: heck no.. [17:58] you know how much I hate to write :) [17:58] Roy: I have 3 computers running LLR and 3 computers running Proth and I haven't found one for a little while [17:58] it took fizmo 6 months of pestering to get the corewarrior article [17:58] I have one running LLR now, but I've sieved it very good, and its going to find something soon, I know it! [17:58] but I figure if someone collected a bunch of the scattered evolver wisdom, we might finally be able to get rid of those pesky humans [17:59] The longer it takes, the bigger the prime will be ;-) [17:59] roy: know any vhdl? [17:59] vhdl..? [17:59] he who must not be mentioned and I were toying with the idea of building a high performance bignum engine for speeding up big prime searches [18:00] hrm no.. I don't (intergrated circuit stuff :)) [18:01] I know just enough to get me in trouble :) [18:02] I sieve one of my ranges up to 330 billion [18:02] and all the people who could help me have "real" research going on :) [18:02] I've sieved it a little further, I have one machine that crashes using LLR so its continuesly sieving (up to two min per sieve) [18:03] Hmmm... have you tried one of the other prime search programs on it? [18:03] Now testen 2409*2^300.000 - 999.999 [18:03] Yeah it runs proth without problem [18:04] hey met.. [18:04] But one for sieving and one for LLR sounds like a good solution too [18:04] with your leet asm skillz, you should really play with optimizing some of the bignum libs [18:04] *-in redcode [18:04] a tiny bit of speedup there goes a long long long ways when doing primality tests.. [18:05] Have you seend Lux Aeterna on r.g.c? [18:06] Yes, but I haven't looked closely yet. [18:06] Looks a bit like Grendel? [18:06] Yeah, it uses about the same trick :) [18:06] Bvowk: the library used by LLR and Proth is optimized asm I believe [18:06] hey met.. did I tell you about my new computer? :) [18:06] I've discovered it before Grendel but now its known the Grendel-trick (dammit) :-) [18:06] CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ (2500.19-MHz 686-class CPU) [18:06] Origin = "AuthenticAMD" Id = 0x20fb1 Stepping = 1 [18:06] Features=0x178bfbff MOV,PAT,PSE36,CLFLUSH,MMX,FXSR,SSE,SSE2,HTT> [18:07] I should be able to get it running at 2800 I think.. but I don't want to do that until I'm not using it for the contest.. I need it to do math correctly :) [18:11] Roy: I have search K=1143, 2667 and 3039 from about 220000 to about 250000 with no luck. I am continuing [18:11] Thats just a fraction [18:12] 6557*2^161011 is a prime :-) (cw number!) [18:13] You're a year too late finding it :-( [18:13] But if LLR found it, test it with Proth too. [18:13] I did, no comment :) [18:14] Proth will test to see if it has any special properties [18:17] Have you figured out what Paul Kline's paper might be yet? [18:17] I think its just the one that was posted on r.g.c, only a bit improved [18:26] Okay, time for me to go [18:26] * Metcalf waves [18:26] MSG: Quit: mov.i #1,1 [18:47] MSG: [20:30] Join: Fluffy joined #corewars [20:31] :) [20:31] hey fluffer [20:32] *yawn* [20:32] tired but not tired [20:32] * Fluffy kicks bvowk very hard [20:32] Hi bvowk [20:32] heh [20:32] Hi Mizcu [20:32] don't kick me for your nick [20:33] Maybe you should spend some time reading my nick correctly before using it ;-) [20:33] I've just started another try to evolve sth. useful for '88 [20:34] but I get to giggle very time I read it my way.. [20:34] your way is giggle free. [20:34] Boys who wear fluffy clothes are.. [20:35] kept warm? [20:35] bvowk: Then continue to giggle. I love to spread fun :-) [20:36] The last time everybody giggled, when I was around, was last summer. [20:36] I have this black t-shirt [20:36] ... with "MILK" written on it with large white letters [20:36] black tshirt.. thats hilarious! [20:37] got milk [20:37] I really enjoyed it :) [20:37] heh [20:37] my wife noted the other day almost all my shirts are black [20:38] hmm ... just got my first result: average score of the soup: 6.144 [20:38] That seems to be a loooong way to go [20:39] heh [20:39] best warrior so far: 17.01 [20:39] I usually don't accept anything into the pool that scores less than 20 or so [20:40] Then I still have a problem with creating the initial set of warriors [20:40] gotta watch tho.. if you're not careful you'll end up with nothing but imps [20:41] I'm not even sure, whether I should really call it evolving [20:41] oh? [20:41] I have 100 warriors, benchmark then, replace the lower half of the "hill" with children/mutations of the upper half [20:42] but replacing is only performed, if new warrior is really better than the previous warrior [20:43] how does that sound? [20:43] evolving or not? [20:43] * Fluffy has no knowledge of nothing and he admits that! [20:43] yeah.. [20:43] that sounds like evolving to me.. [20:43] survival of the fittest right? [20:43] yes [20:44] unless they're more fir that the previous, they don't survive :) [20:44] any good tips for setting/changing the mutation rate/crossover probability? [20:48] keep it low.. [20:48] way lower than you think makes a difference.. [20:48] to high and you'll prevent anything good from coming about [20:49] I usually add a test to ensure that there's a change in the warrior.. [20:49] how low is low? [20:50] prob of mutating an instruction: 5 % [20:50] low enough that sometimes they don't get any mutations :) [20:50] prob of cross over: 10% [20:50] I usually run with somewhere between 3 and .5% [20:50] and there *HAS* to be a change between the parent and the child.. [20:51] its weighted as to the mutation as well, there's a tiny chance the instruction or imod will change.. [20:51] slightly greater chance of changing the a or b mods, and a greater chance of changing the operands [20:54] ok :) [20:56] I know that sounds odd.. but it keeps your mutations targeted towards not breaking warriors as often.. [20:56] especially as they improve.. [20:56] because once they stop sucking, you don't want it changing instructions or big changes to the code.. [20:56] you'd like it to work more on optimizing the steps and the like. [20:57] so mutations that break the code are a waste of a test. [20:57] so it is the search for better constants than completely new warriors? [20:58] I will keep that in mind [20:58] well.. once you've got something that doesn't suck right? [20:58] you want to avoid breaking up the complexity they develop [20:59] so if you've got a stone, you don't want your jumping instruction to suddenly turn into something useless. [20:59] hehe ... I've looked at the soup. All warriors suck. Apart from "jmp 0" I have: ... [20:59] ... "djn 0, x" or verrry slow djn-streams [21:00] heh [21:00] it takes a long time to bootstrap a random pool [21:00] suddenly one warrior will find soemthing that doesn't suck and *BOOM* the scores bounce up to that level as everything gets replaced.. [21:00] hmm ... the current soup took me about 6 hours (100 warriors) to generate [21:05] OMG [21:05] i changed the gaps between parts and +3 points [21:08] :) [21:09] bvowk: Another question ... do you know how to recognize when it is no longer worth to evolve with a given soup any longer? [21:09] Or do I have to simply wait until sth. new evolves? [21:10] thats kinda the art of it isn't it? :) [21:10] usually I stop when nothing new happens for long periods.. [21:10] then I mix things up [21:10] either by loading a new pool, shuffling species about.. [21:10] sometimes just resetting all the scores to zero and rescoring everything is enough to get it rolling again [21:11] ok ... then I will wait now a "little bit" [21:12] minimal score has increased from 0.8 to 3.4 :) [21:12] I think, that is is about time, that to become scared, bvowk ;-) [21:13] 3.4!!!! [21:13] That equals 4 for large values of 3.4! [21:14] only 296 more points until I have a nearly perfect warrior [21:14] Shouldn't be that hard [21:15] 139.8 , and im out of things to opt =/ [21:15] fsh94-0.2 or -0.3? [21:15] 0,3 [21:16] Try it on the hill! [21:18] ill send it once i break 140 [21:19] hehe ... do you really think, that is will make any difference? [21:19] you are probably using to few rounds to have an error margin lower than 0.2 [21:19] *too [21:23] MSG: [21:26] Join: Mizcu joined #corewars [21:27] Hi Mizcu :-) Already got another 0.2 points? [21:28] it seems optimax doesnt want to collaborate with me [21:28] How about redefining the value of zero to say ... 5.0? [21:29] -_- [21:30] what name should i use [21:30] restatic? static again? static 1.5? [21:31] one moment ... [21:31] static field? [21:31] no [21:32] How about sth. new? [21:32] nah [21:33] "Static changes" [21:33] static rebirth |) [21:36] :) [21:36] now i got the silliest idea [21:36] Static E1M2 [21:37] ? [21:37] E1M2 = Episode 1, Mission 2 [21:37] I like rebirth more [21:37] And you should start with E3! [21:37] err ... E4 [21:38] it came only to my mind because im now listening to the most famous song from Doom (the game), which just happens to be from E1M1 [21:38] and I was thinking about Star Wars [21:40] hmm ... "White Noise" would be a nice name, too, but unfortunately there is already a warrior for Tiny with that name [21:41] 139.93 =/ [21:41] No! You have to redefine the value of zero to 5.0 and not to sth. below 1.0 ;-) [21:42] Mizcu: "Fixed in place" [21:42] there exists Fixed [21:42] yes, but not ".... in place" [21:43] 139.95 X) [21:43] No! 5.0! [21:48] Mizcu: "Significant Change"? [21:49] hard to say significant when stone is the same, and impl+qs are changed by very little [21:49] Then it is a perfect name [21:49] hrm [21:49] reoptimizing the Qscan would be excellent, but not many has time to do that [21:49] and i dont really have an idea where to start with it [21:50] You only have to change the values of qy and qz [21:50] But of course you only have to use correct pairs [21:53] i've tired of optimizing any more [21:55] Then there is only one thing left to be done [21:55] 'yah [22:00] and that does mean what? [22:01] im not getting "good compile" -response.. [22:01] maybe a wrong ;redcode line? [22:02] no, some voodoo error [22:03] hmm last submission was today [22:04] Error in line 9: 'sboot equ 5162' [22:04] Unrecognized or improper placement of token: '5162' [22:04] Number of errors: 1 [22:05] focking great [22:05] then it must be sth. in line 1 - 9 [22:05] post the warrior here [22:06] I think, I can help ;-) [22:06] _|_ o_O _|_ [22:08] Join: Roy joined #corewars [22:08] Hi Roy! [22:08] Hiya [22:09] Did you see Nenads scanner? :) he changed my re-zoomer into dx42e-clone by inversed, but I like it :-) [22:09] actually ... no [22:09] r.g.c, Lux Aeterna is there too [22:10] wtf [22:10] 140 now? :) [22:10] not a damn thing wrong in the warrior and still errors [22:10] not using a label and constant both named sboot..? [22:10] (thats the only thing I can imagine that gives that error) [22:11] try just renaming it :-) [22:11] but that would cause error in pmars too? [22:12] I guess so.. [22:12] Maybe sboot is the secret label of the secret instructionset...! [22:12] JM will finally be exposed! [22:13] koth also had problems with : in using stone: blah blah [22:13] Good luck with your warrior, Mizcu [22:13] Heh yeah, but I hate using that too :) [22:13] * Fluffy waves [22:13] ;redcode-94nop [22:13] ;name Ultimate warrior [22:13] ;author Roy van Rijn [22:13] ;strategy Uses the secret instructions of sboot [22:13] ;assert 1 [22:13] jmp sboot [22:14] end [22:14] I'm going to send it ;-) [22:14] * Roy waves [22:14] Part: Fluffy left #corewars [22:14] Aww.. [22:14] Error in line 8: 'jmp sboot' [22:14] Undefined label or symbol: 'sboot' [22:14] Number of errors: 1 [22:15] changed to stoneboot and still errors at the magic number [22:15] Thats very odd.. [22:16] My warrior compiled fine (but it sucks) [22:16] So it has to be something in your warrior, the parser is working [22:17] ill answer to voodoo-problem with voodoo solution; switching e-mail [22:20] http://www.milliondollarhomepage.com/ <- I hate it when people get rich using so simple ideas, but at the same time, I admire the man :) [22:22] nope, didnt change [22:22] AAAAAARGHHHH [22:23] i moved the equ-lines lower, under first instructions and it compiled correctly [22:23] Well, I could have a peek at the code if you mail it or something, just strip it of all numbers and some code or something [22:23] Huh!/ [22:23] Thats silly.. [22:24] hi Inversed, i now you are reading logs [22:24] now -> know [22:26] umm.. right [22:26] Join: datagram joined #corewars [22:27] hrm [22:27] datagram [22:27] olla [22:27] koth still compiles it wrong [22:27] Hi [22:28] hi data [22:28] this is just crap [22:30] maybe its time someone reboots koth.. [22:31] Well, my warrior did fine, I really think its your code.. is the header correct etc? [22:32] ill smack it again at the morning [22:45] Probably a good idea, ok, I'll stop the prime searching and download fsh0.3 and optimise my obesive constant giant [22:46] woah, Fizmo wants everybody to kill my warriors!! I open the zip and all I see are my warriors [22:47] lol [22:50] Gargan/Eccentric/Gremlin/Hulla/Frantic/Yorba/Halcyon/Maelstrom/Deathstar/Dracula2003/Solo3/Leap/Cascade (13 warriors =>13.2% is mine!!) [22:50] But he also included a LOT of his warriors :-) [22:50] MSG: Quit: humhum [23:16] MSG: Read error: Connection reset by peer [23:21] MSG: