[01:45] MSG: Quit: Lost terminal [08:17] Join: Spikeman joined #corewars [08:18] Hello? [08:54] hi [08:54] how may i be of assistance? [08:54] Oh, I'm new to Core Wars. [08:54] Just wondering exactly how to get involved. [08:55] well, at first you need an redcode-simulator, what OS do you use? [08:55] Windows, I've actually been using nMars, should I use something else? [08:56] nMars should be fine, though i suggest Corewin for beginners [08:56] Alright. [08:57] I've been reading a bunch of documentation and source code, but I still don't quite understand how most of the more complicated programs work. [08:57] no worries about that [08:59] I did manage to make one semi-decent program. [08:59] if you want me/us to explain some warrior, or something in redcode, or other corewar-related, just ask [08:59] Can I copy paste it? (I'm new to IRC as well) [08:59] if its under ten lines, i dont mind [09:00] Alright, one second [09:00] the server has probably some flood-prevention active, so if it kicks you out, come again and paste it directly to me [09:00] mov 5,@4 [09:00] spl @3 [09:00] add #3,2 [09:00] jmp -3 [09:00] dat #0,#1337 [09:00] jmp 0 [09:01] I don't really remember how it works anymore.. [09:01] it throws jmp 0's around, and put processes on them [09:01] Yeah. [09:02] Is that a decent strategy? [09:02] It seems to tie most things I put it up against. [09:02] well, the problem is that it is not offensive, and scoring favours aggressiveness [09:03] Ahh. [09:04] I have this one program, a silk, that seems to copy incredibly fast, I don't understand how it goes so fast. [09:04] so technically, you spread yourself around to prevent death by single attacks, of if you get lucky, just jump on the opponent and assure an tie [09:04] (Silk Warrior 1.3 by J.Pohjalainen) [09:05] in silk processes work in paraller; eg. there are number of processes on top of each other, running one instruction X times, second X times etc [09:05] in start of the silk there is always something to the extent of spl 1 spl 1 spl 1 ; that turns one process into 8 paraller [09:06] Right, but I thought more processes slowed it down overall. [09:06] it slows down overall speed of _one_ copy [09:07] speed of warrior is always the same, regardless of the number of the processes [09:07] Hmmm.. [09:07] Say one warrior has a single process, and another has 8 processes, wouldn't the first warrior run eight times for each time the second is run? [09:07] Yes. [09:09] So wouldn't that make the second one slower overall? [09:10] both warriors run 8 instructions in 8 cycles [09:12] Well, when I run my Imp against Silk Warrior, Silk Warrior grows really fast and my Imp only starts growing when it is overwritten by the silk. [09:12] How does it grow so fast? [09:12] i was about to explain [09:12] lets assume the silk is starting at instruction 3, and we have 8 paraller processes on it [09:13] Right [09:13] so, the first instruction of the silk itself is a spl @0, somewhere [09:14] so running it 8 times will send 8 processes to [somewhere] [09:14] Yes [09:14] then the 8 processes that ran the spl will fall down to next instruction [09:14] the process queue will look like this then: [09:14] 4,somewhere,4,somewhere,4,s,4,s... [09:15] right [09:15] the instruction in 4 is a mov }-1, >-1 [09:16] so at first, it moves the spl -instruction at [somewhere], then increasing both a and b -values by one after moving is done [09:16] Does that increase the values of the mov instruction or the spl? [09:16] the spl [09:16] Ahh. [09:17] now, the process at somewhere runs, and it was just "given" an spl [09:17] Yeah [09:17] and the processing looks like 4, s, s*2, 4, s, 4, s ... [09:18] So there is two processes at s*2? [09:18] so the process at [4] copies the next 7 instructions to s+1, s+2 etc.. [09:18] .. at same time at s creates 8 new processes at s*2 [09:19] now the first two instructions at 3&4 are one, we have one copy of paper at s, and a non-running spl at s*2 [09:20] Okay [09:21] now, each spl/mov -pair will copy and start (number of processes) instructions including itself, and those below [09:22] the last mov/jmp -pair will do the same, but in reverse, and jump instead of starting new processes (because below the jmp is "empty", which will kill) [09:22] So it kills off the processes which already copied everything? [09:22] ah, no [09:23] Running it really slow, it looks like it just copies itself everywhere, right? [09:24] silks usually have mov-instructions in it to throw core around [09:24] Is that what the "gate" is? [09:24] either semi-random pieces of core (since "empty" kills), or intentional bombs [09:25] It has this: gate dat.f <2667, <2*2667 [09:25] the gate is a misnaming in that warrior [09:26] Really? [09:26] it is used as a specialist bomb thrown around [09:26] it should be named bomb or something [09:26] I think it uses it as part of a gate maybe. [09:26] The code is 10 lines, I'll just paste it [09:26] the instruction is not modified or run [09:27] i am watching at the warrior right now [09:27] spl 1, <-98 [09:27] mov -1, 0 [09:27] spl 1, <-99 [09:27] paper spl @paper, STEP [09:27] mov.i }paper, >paper [09:27] mov.i gate, >paper [09:27] mov.i gate, {-5*SIZE-1 [09:27] mov.i gate, {4000 [09:27] silk jmp @silk, {paper [09:27] gate dat.f <2667, <2*2667 [09:27] Doesn't the < modify it? [09:27] only if it is run [09:27] the warrior doesnt run the instruction [09:27] Okay [09:28] that bomb is specialized against certain structures, so it causes additional damage when hitting [09:29] Oh [09:29] the reason why silk's copy so fast is that they run only spl and mov -instructions, and only the amount necessary [09:30] Right [09:30] So any recommendations on what to try and make? [09:30] so far i dont have much to recommend [09:30] you could try making a warrior that scores 100% wins against a dwarf [09:31] Alright [09:31] and for a challenge, do it without spl [09:31] Okay, I'll try it [09:31] That would be a stone right? [09:31] dwarf is a simple stone, yes [09:32] but if you mean that stones dont have spl's , that is not correct [09:32] Oh so what I would be making is a paper, that's what I was wondering. [09:32] you dont need a paper to do it [09:33] Okay, but generally papers beat stones, if I am understanding the metaphor correctly. :P [09:33] yes [09:33] Okay [09:34] the dwarf-killer can be done in 3 instructions, but no need to do it fancy or so for now [09:35] Alright [09:36] So the dwarf stays in one place and drops bombs [09:36] I can think of how to tie it [09:37] 4 parallel imps, because the dwarf is modulo 4, one wouldn't get killed [09:38] but as one of the imps gets killed, there will be 3 left [09:39] The imps would travel 4 spaces at a time [09:39] Is that possible with one instruction? [09:39] each imp will move one instruction during 4 cycles. 4 instructions run in 4 cycles. [09:39] right [09:41] it is still possible for the dwarf to kill one of the imps, and then another [09:41] but all 4 would be quite impossible, yeah [09:41] But how would I go about killing the dwarf? [09:41] I would have to find it. [09:55] Join: fiveop joined #corewars [09:59] Well thanks for all your help Mizcu, I'll be back tomorrow hopefully with my dwarf-killer. [09:59] seeya [10:00] MSG: Quit: Chatzilla 0.9.77 [Firefox 2.0.0.2/2007021917] [15:59] Join: Fluffy joined #corewars [15:59] :) [16:04] . [17:01] Join: John joined #corewars [17:01] Hi :-) [17:05] h J [17:06] H i M i z c u [17:09] Join: sascha joined #corewars [17:09] Hi Sascha [17:09] Hello John [17:10] What are you up to at the moment? [17:10] Only 30 Minutes for a little chat, why? [17:10] I wondered if you are working on anything interesting? [17:11] Hmm, there ist something very interressting but i haven't found the right design [17:12] :-/ [17:12] The Challenge in the last week where the absoulte maximun after 1000 randrys...not enough [17:13] first a would try ther mainwarrioir it self before i do the "tricks" [17:14] but i can't find something successful - doesn't matter 6 liner, 8 liner or 9 liner.. [17:15] :-( [17:21] But there are some tests open,. i hope next week i will found something proper [17:24] there's always ideas waiting to be abused [17:24] yes, thats it, but the most only sounds good but scores poor. [17:25] thats what i thought about shubbery too, but inversed brought out the best of it [17:26] yay, yace made a multipass coreclear [17:26] always good to have such a friend with magic hands ;-) [17:27] too bad plain coreclear suck [17:27] oh,, time for lunch [17:28] * John waves [17:29] Just download nMars to play with [17:31] heya [17:33] hrm [17:34] oh, its you. [17:34] Who did you expect? [17:35] met. [17:35] what are you up to today fluffy? [17:35] you mentioned something about 94nop? [17:35] yes and I'm still working on it [17:35] Hi Bvowk, Fluffy [17:35] Hi JM [17:36] whatcha up to met? [17:37] I'm suffering from toothache [17:38] :( [17:38] Time to go otherwise I won't get anything done [17:38] * Fluffy waves [17:39] ugh, two spl/mov/djn's in a row [17:39] and? does it score well? [17:40] spl/mov/djn -coreclears dont score more than 120-130 in benchmark [17:40] and they are reaaallllly sticky endpoints [17:42] problem is, if i use laxer settings, that encourages those-that-i-hate [17:43] I'll upload a new benchmark after nw 2 [17:43] * John waves [17:44] MSG: Quit: mov.i #1,1 [17:49] yay, i got an imp-pump [17:50] so you're working on a 94nop evolver? optimizer? [17:50] just yace'ng for nano [17:51] and the one i am working for in my mind is a nano/tiny evolver [17:51] highest score 132, lowest 102 on the yace's internal hill.. [17:52] i smell imps [17:55] now i smell dogfart and nanopaper [17:55] but it works better as nonpaper ?=/ [18:03] and now it cant decide [18:06] what is to decide? [18:06] if papery is better that non-papery [18:07] Send to the hill and you'll see [18:07] benchmark says itll get swamped [18:08] It seems as if there wasn't anything to decide [18:09] fluffy: So you're optimizing for 94nop? evolving? [18:09] or just writing? [18:10] and now, new challenger comes with 40 point lead. It consists of a -- djn-train [18:11] bvowk: I've taken a warrior and gradually improved it by testing new ideas (and keeping the good ones). [18:11] So it is some kind of optimizing [18:12] But some ideas need a lot of time to test, so I estimate that it will take a lot more time until I declare it to be finished. [18:15] aand the djn-train is killed [18:16] number one, with eleven point lead is spl/mov/ djn hmm [18:18] Join: Core_old joined #corewars [18:20] Hi old Core! [18:29] Join: jon_ joined #corewars [18:29] Hi jon_! [18:30] good morning [18:30] hmm [18:47] MSG: Read error: Connection reset by peer [18:47] Join: Core_old joined #corewars [18:48] MSG: Client Quit [19:06] Join: Core_old joined #corewars [19:07] MSG: Quit: Lost terminal [19:18] MSG: Quit: Leaving. [19:32] bah. [19:32] cheches got knocked off [19:33] i have half a working warrior on my laptop. [19:33] i need more spare time or something. [19:34] greetings Jon [19:34] whats up? [19:48] monday :( [19:48] thats not really up ;) [19:48] maybe :) [20:22] Part: sascha left #corewars [21:01] Join: fiveop joined #corewars [22:02] MSG: Quit: Trillian (http://www.ceruleanstudios.com [22:20] Join: kaatil joined #corewars [22:26] hi there [22:26] greetings, earthling. [22:28] yo [22:30] darned, just when i am about to get get some sleep, a possible newb comes and i have to hold my duty as the official greeter [22:30] i am.. *ahem* kindof new to asm and codwar.... it is interesting machine language... i guess. :) [22:31] heh I see. [22:31] well, dont get to think that redcode is _that_ close to asm [22:31] ok :) [22:32] well, if you got questions right now in my mind, ill be happy to answer [22:32] kaatil: i'm also new. [22:32] good [22:32] after ill get to sleep, youll have to trust our night-shift to provide the answers (do we have one anymore?) [22:32] (dont take that personally, jon) [22:33] i am currently reading the guide to learn how to play codwar. [22:33] corewar rather [22:35] well, the best start is to take bunch of guides, and see how code runs in a core [22:36] ok [22:37] curious... can pMar program run under linux? [22:37] yep [22:37] ok [22:38] ill get to sleep n [22:38] ow [22:39] heh [22:40] have a good sleep, Miz [22:50] hrm. [22:50] never fails.. I submit winning warriors to infnano.. and now I've found something significantly better. [22:50] :( [22:53] oh no :( [23:02] heh [23:02] what did you found something SIGNIFCANTLY better? :) [23:11] better nano warrior.. [23:11] the evolver spit out something better [23:31] ah