[00:03] Join: jkw joined #corewars [00:04] MSG: Ping timeout: 255 seconds [00:20] Join: aasdgasd joined #corewars [00:52] MSG: Ping timeout: 255 seconds [00:54] MSG: Quit: -=SysReset 2.53=- [00:57] MSG: Quit: Trillian (http://www.ceruleanstudios.com [01:00] Join: johnkw joined #corewars [06:06] MSG: Read error: Operation timed out [06:14] Join: johnkw joined #corewars [06:38] MSG: Ping timeout: 255 seconds [06:39] MSG: Read error: Operation timed out [06:39] MSG: Read error: Operation timed out [06:39] Join: jkw joined #corewars [06:40] MSG: Ping timeout: 255 seconds [06:41] Join: asw joined #corewars [06:41] MSG: Ping timeout: 255 seconds [06:49] Join: Mizcu joined #corewars [09:00] Join: johnkw joined #corewars [09:00] Join: Miczu joined #corewars [09:00] MSG: Read error: Connection reset by peer [09:01] MSG: Read error: Operation timed out [09:01] Join: fiveop joined #corewars [09:34] MSG: Ping timeout: 255 seconds [09:41] Join: johnkw joined #corewars [09:53] MSG: Quit: Lost terminal [10:26] Nick Change: Miczu changed nick to Mizcu [10:44] Join: Core_old joined #corewars [12:49] Join: Roy joined #corewars [12:56] hR [12:57] hiya [13:39] MSG: Quit: Trillian (http://www.ceruleanstudios.com [14:01] Join: Core_old joined #corewars [14:32] Join: bvowk joined #corewars [15:09] MSG: Quit: Trillian (http://www.ceruleanstudios.com [16:00] Join: fiveop joined #corewars [16:11] MSG: Quit: dat [16:37] MSG: Quit: Lost terminal [16:38] hrm. [16:38] * bvowk dats roy [16:38] (until he returns) [17:02] * bvowk wonders if sf is goign to show up [17:11] Join: Fluffy joined #corewars [17:12] :) [17:13] hey fluf [17:14] you and sf have been quite active on the nano hill [17:15] sf's scores keep going down tho [17:18] yes, the hill doesn't like cookies [17:18] indeed. [17:18] I've raised my warriors on the flesh of human warriors.. [17:18] they're not into carbs.. ;) [17:20] carbs? [17:22] sorry, high protein diet reference [17:22] No need for the "sorry2 [17:23] What happend to all those ops? [17:23] I imagine they lost connection [17:24] I just thought, that you should never op me. I like kicking far too much and #corewar would become a very lonly place :) [17:25] heh [17:25] * Fluffy kicks bvowk [17:25] * bvowk verbally abuses fluffy [17:26] hrm [17:26] I shouldn't ever go near ebay [17:26] * Fluffy looks for his only existing brain cell and tries to taunt bvowk [17:27] Damn I shouldn't do that. [17:27] I've almost forgotten what my name is [17:32] I wish sf would come around so I could taunt her ;) [17:32] http://cgi.ebay.ca/Dell-PowerEdge-8450-Qty-8-700MHz-Xeon-2MB-Cache_W0QQitemZ200070276226QQihZ010QQcategoryZ51229QQrdZ1QQcmdZViewItem [17:32] 8 way xeon for you fluffy? [17:32] A little bit slow, isn't it? [17:33] and probably very noisy! [17:35] What I need is a shell account to a really fast computer. [17:35] Hey, I already have that :) [17:36] heh [17:36] I suppose I could give you access to mitzi [17:36] Don't you use mitzi for your evolver? [17:37] yeah [17:37] (so FYI, pycorewars runs on sparc64 ;)) [17:37] w/ SunCC no less [17:38] Now, that I have a little time to think about it. Do you have access to cpus, which have a really large L1 cache? [17:38] and same question for L2 cache [17:38] I'm talking about > 4 MiB L2 [17:39] the sparcs have 8MB L2 [17:39] That should be enough [17:40] one of my test versions of PyCorewar get a lot of speed by partially unrolling the loop [17:40] (where all instructions are executed) [17:41] But unsually the cache is too small to hold both the large code and the core + process queue [17:41] *usually [17:42] bvowk: Doesn't mitzi needs very much power? [17:42] *need [17:43] ... and who pays that bill? [18:26] Join: fiveop joined #corewars [18:26] Hi fiveop [18:44] Join: sf_ghoul joined #corewars [18:45] ha ha sf [18:45] having trouble with the new machine made hill? [18:45] Hello bvowky [18:45] Hi sf [18:45] Hello fluff [18:45] No [18:45] no trouble [18:45] I'm testing [18:45] I only need to switch an instruction for the final version [18:46] yeah? [18:46] I look forward to your clobbering ;) [18:46] sf: I hope that you'll push off tons of bvowk's warriors :) [18:47] I hope so too [18:47] I'm down to one :( [18:48] awww [18:48] hrm... [18:49] I wonder when you'll be able to get 320GB notebook disks [18:49] and what will you do with those 320 GiB? [18:49] fill them [18:49] with what? [18:49] data [18:49] ;) [18:49] images? videos? porn? [18:50] I cart about a huge library of materials on my notebook [18:50] huge swaths of interesting bits [18:50] so if I'm out somewhere without intarweb, I can always lookup whatever I might like [18:50] then why don't you download the internet? [18:50] :) [18:51] because I don't have enough disk ;) [18:51] one off :) [18:51] but wget -km www.foo.info is your friend. [18:51] hrm. [18:51] you pushed off one of my warriors! [18:51] you should put that into an infinite loop [18:51] to stay up to date [18:52] the interesting stuff on the web doesn't change that ofte [18:52] sf: Yay! Go on! [18:53] I've only got 2 in the nano hof top 50 now :( [18:54] sf: Yes, but bvowk has NONE! [18:55] bvowk: how does that compare to Fluff's 4 in the top 50? :P [18:55] Fluff: are you technically evolving or not? [18:55] No, technically I'm a hard-core-optimizer [18:56] Altough I think, that some evolver might be nothing but optimizers, too [18:57] *evolvers [18:57] eh? [18:57] bvowk, some evolvers, not all! [18:58] where do you draw the line? [18:58] I thanke a form a randomly test values and take those which result in better scores [18:58] *take [18:58] grr [18:58] I take a form and randomly ... [18:58] hm [18:58] ah [18:58] mines really evolutionary [18:59] the ccai one wasn't quite so much [18:59] Some evolvers start with randomly generated warriors, but often (d)evolve into a soup, which has only one form [18:59] from that point on the do technically nothing different than I do [18:59] two off :) [18:59] Yay [19:00] thats alright, you're pushing my old ones back up [19:00] old ones? With the current hill I'd call everything below 500 young :) [19:01] * bvowk shuffles through another 51,000 warriors [19:01] Any ideas what rumplestiltskin is? [19:01] fluffy: old for me:) [19:01] sf: pure evil! [19:01] three off :) [19:02] hey, I'm koth [19:03] :) [19:03] and I don't have any idea what rumpelstiltskin is [19:06] otherwise I'd flood the hill with optimized variants [19:06] :) [19:08] four off :) [19:10] Unfortunately your warriors won't live very long :( [19:11] Doesn't matter [19:11] Actually ... [19:11] ... I totally agree :) [19:11] The better one might stay on the hill once the others drop off [19:11] five off, this ones an old one ;) [19:14] I really wonder what happend to JM. [19:14] Oh, his warriors cause Bombus Sylvestris to drop down the hill and finally be pushed off [19:14] So I banned him :) [19:15] lol [19:20] Switched that split now [19:20] hmm ... free mp3s from Mercedes Benz [19:21] Bah, worse [19:21] http://71grad.de/2006/12/15/mercedes-benz-mixed-tapes-herunterladen/ [19:21] (ignore the German text and download the zips) [19:25] Join: Roy joined #corewars [19:25] Hi Roy [19:25] * sf_ghoul hides :( [19:26] Hi Roy [19:26] Hi [19:26] Why hide? [19:26] I'm hiding from fluff [19:26] 51 39.3 38.1 22.6 lord of the digital swamp Metcalf/Fluffy 140.5 296 [19:27] eeek! [19:27] Shhh! [19:27] :( [19:27] sf_ghoul: What have you done! [19:27] Whoops hehe... well [19:27] It was such a lovely warrior [19:27] scanner/imp [19:27] grmpf [19:27] yes, a really nice combination [19:28] John provided the warrior, I've optimized it [19:28] Post it post it! [19:28] yes, post :) [19:28] No, that's up to John [19:29] * sf_ghoul wishes microgp or zul would publish [19:30] No, that way their warriors won't stay very long on the hill. [19:30] I made that mistake too [19:30] me, too [19:31] I suppose me too [19:31] Yeah! thats the way, we suck! [19:35] * Fluffy is compiling XFce ... [19:35] XFce? [19:35] www.xfce.org [19:36] Nothing for Windows :) [19:36] * Roy has cygwin? :P [19:37] Then kill the rest and do it right :) [19:39] No.. I like windows in some odd way [19:39] Most apps are still made for windows, and so are the games.. [19:40] As long as you are fine with it :) [19:40] Yeah, everybody has its own taste [19:40] Lets play with my evolver again.. [19:40] * Roy fiddles around [19:40] :) [19:41] It still sucks very bad, but hey it evolves stuff and the warriors get better as it goes [19:41] anything on the hill yet? [19:42] Naaaahhh.. [19:42] It creates djn clears all the time [19:42] Its best warrior now (after 3 minutes) is this: [19:42] ;redcode [19:42] ;name Unnamed [19:42] ;author Origin [19:42] ;assert 1 [19:42] START mod.a #67 , >71 [19:42] spl.i >51 , #3 [19:42] spl.a #14 , #4 [19:42] djn.x #78 , {63 [19:42] sub.x >33 , <71 [19:43] END START [19:43] hmm ... [20:00] All my paperish ideas fail [20:00] hrm [20:00] sf_ghoul: Handcoded? [20:01] Yes [20:02] dammit. [20:02] you've pushed me back to under 2/5th of the hill again. [20:04] It seems to be very hard to be continouly koth on nano (with the same warrior) [20:08] yeah [20:08] nano is hard to stay on top of [20:20] * Fluffy is back in a couple of minutes ... [20:20] MSG: Quit: fluffy.i < 1, # 42 [20:22] Join: argiopewe joined #corewars [20:23] hi Argio [20:24] I'm trying to decide if a certain location in core is 0 or if it has been changed. I'm trying to do this with "jmn bomb,@data". Instead of only bombing when not zero, it bombs every time, just like the jmn wasn't there. Any clues on why? [20:24] Hi Mizcu. [20:24] noticed from the logs that you are new one - if you get any questions, feel free to ask me. I am a sort of newbie-guide here. [20:24] Hi Argio [20:24] could you give the whole code? [20:26] Oh, well thanks Mizcu. I'm actively working on this code, but yes, I can throw it on one of my servers. [20:28] Hold on here. It appears that the dats in core are initialized as "dat $0,$0" and not "dat #0,#0" Why is that? [20:29] in ye old times, you could not include dat $0, $0 in a warrior [20:29] and thus if you could find a dat #0, #0, that means that it was intentionally moved there [20:30] Join: Fluffy joined #corewars [20:30] in '94 standard it was enabled again [20:30] sf, after like 28 tries, I don't think its getting any closer ;) [20:30] no, enabled for the first time [20:31] think of mizcu like a spirit guide for corewars.. he seems to speak in riddles, but really he just needs more sleep ;) [20:31] hehe [20:31] Hi argiopewe [20:31] Hi fluffy. [20:32] * Fluffy is reading the logs ... [20:33] sleep is for the weak [20:33] Then why do you need so much, Mizcu [20:33] So, I do an "sne @data,data2" where @data points to a section of memory, and data2 is "dat $0,$0" would I be able to detect when an area of core had been modified? [20:33] ? [20:34] yes, but you can do better [20:34] you dont need the @ [20:34] ugh [20:34] yes, i need to catch some sleep [20:34] yes, thats a way to check it [20:34] I don't need the @? [20:35] sne $ a, $ b compares the positions "a" and "b" [20:35] But if I don't use the @, won't it be comparing data and data2, as compared to the address held in data to data2? [20:35] sne @ a, @ b compares the positions that are specified by the b-fields at position "a" and "b" [20:35] if you compare two positions in core, 98% chance says they will be both dat $0 [20:36] Right. [20:36] so you dont need to compare to some defined dat 0 [20:37] the rest 2%, well, then bomb the hell out of the place [20:37] Okay, so if I do that, how will I know where whatever it is that isn't equal is? [20:37] that's up to you [20:38] i.e. do a more specialized compare :) [20:38] Ah. [20:38] or bomb both positions [20:38] or make them close to each other [20:38] like "sne $ pos, $ pos + 5" [20:39] So what does $ mean as compared to @ (address at) and # (immediate?)? [20:40] $ usually means that an instruction is manipulated. # mean that a number in instruction is manipulated [20:40] "$ x" specifies the address x locations away [20:40] Okay. [20:40] "@ x" sth. like a table lookup [20:41] i.e. look at x and then look at the position specified by x [20:42] http://vyznev.net/corewar/guide.html explains at lot of theses things :) [20:43] Ah, must have missed that part... So the $ is the same as having nothing at all... [20:44] yes [20:45] What's the best method to use against an imp-type warrior? [20:45] a gate [20:45] Besides a gate, obviously. [20:45] jmp $ 0, < -10 [20:45] coreclear with gate [20:45] (that's the easiest form) [20:45] Coreclear with gate. Okay. [20:46] But generally there isn't a "best" method. A lot depends on you warrior and how its components play together [20:46] Gates will only kill imps, and they don't work if you're multi-threading, so I don't like them. [20:46] but gates come cheap [20:46] True. [20:46] so there is no reason not to use them if you can [20:46] Is there any good way to use them while threading? They don't seem to work well if they are skipping beats. [20:47] and you shouldn't think of killing your opponent 100% of the time, even 50% is a good values [20:47] *value [20:47] well, that are you threading? a paper? [20:47] a scanner? a stone? [20:47] I tried both a paper and a stone. [20:48] No scanners. [20:48] papers dont need a specific gate to defeat imps, only lots of bombing [20:48] a paper works well with bombing the core with "dat < 2667, < 2 * 2667" [20:48] (against B-imps) [20:49] or bombing with "mov # x, { 1" against A-imps [20:49] B-imp: mov.i $ 0, $ 1, A-imp: mov.i $ 1, * 0 [20:49] stones are a little different thing, but having a gated coreclear with a stone is one option (example: Son of Vain) [20:49] mov.i #1, *0 , silly Floff [20:50] I need to look more carefully at son of vain's code. [20:50] you wont understand Son of Vain's code very well [20:50] Mizcu: yes, but that's an A-imp, too [20:50] Probably not... [20:50] but that will come with time, argiopewe [20:51] Hmm, doesn't help that my copy doesn't have line breaks... [20:51] huh? [20:51] open in wordpad instead of notepad? [20:51] http://www.ociw.edu/~birk/COREWAR/94/HILL/sonofvain.red [20:52] That might work. [20:52] Much better. [20:52] :) [20:53] Hehe. It's commented... Makes it even easier. [20:53] woo! 104 against the nano hill, best score yet :P [20:53] Knowing a bit of assembly to begin with helps... [20:54] Roy: and what is it now? sill a simple clear? [20:54] *still [20:55] Yes, a clear thing.. it looks like it favours the # adress mode :P [20:55] hrm. [20:55] yes, it has plenty of ties [21:01] Sal seems to have stopped :( [21:01] eh? [21:02] Maybe your firewall has identified me as a spammer! [21:02] how long you been waiting? [21:02] nope, just checked that ;) [21:02] 15 mins [21:03] Okay, I double checked [21:03] 8 mins but it seems like 15! [21:04] Join: Core_old joined #corewars [21:05] Hi old Core! [21:05] Hi Core29 [21:05] hey! [21:05] I think it's Roy's warriors causing a delay [21:06] Every time he sends one, Sal slows down :( [21:06] But the queue is empty [21:06] hrm. [21:06] (did i sense argio having got stuck into a loop when analyzing SoV?) [21:07] I sent about 10 warriors which aren't there yet :( [21:07] Though art wize Mizcu. [21:08] argiopewe: Maybe you shouldn't start with SoV, but with "easier" warriors like Neverland:http://www.ociw.edu/~birk/COREWAR/94/HILL/neverland.red [21:08] sf: we're poking at it right now.. [21:08] Though what I'm doing currently is less looking at how it wins, and more looking at why it wins, and what it wins against. [21:08] sf: you didn't get any assembly notices? [21:09] "why it wins" is a really tough question [21:09] Hehe, that one is simple... [21:09] Fluffy: I know. That's why I'm going in circles. [21:09] No assembly notice :( [21:09] cause I don't see you in the queue either.. [21:10] and roy and I can both submit and they run normally.. I'm wondering if your side is the problem [21:10] Just send a test warrior to nano [21:10] in SoV the three components (stone, imp-spiral, and coreclear) are perfectly balanced and positioned [21:10] I can fight fine.. [21:11] ... and now it is on the hill [21:11] The stone kills scanners, and stone/imps are defeated by the combination of stone and coreclear [21:11] ... and got back the usual two mail [21:11] *s [21:12] papers have problems against SoV because its imp-spiral can often force an tie, and the stone has long enough bombing-run to cripple a paper with less than aggressive replication [21:14] Like the original imp. It has a hard time killing it because it doesn't ever sit still for more than one cycle. [21:14] but an imp cannot defeat an opponent (only make an opponent suicide in some rare cases) [21:15] .. and in some really rare cases even really good warriors are easily killed by an imp [21:15] Right. And that's pretty much useless. [21:16] actually, Toxic Spirit is only warrior good at its time that imp was capable of defeating [21:16] So, the moral of the story is, "Don't test stuff against imp" [21:16] toxic spigot! [21:17] there is probably other example, but i cant remember right now [21:18] So when did you all start playing with corewars+ [21:18] s/+/?/ [21:19] April 16, 2005 [21:19] around start of '03 [21:19] Well now. Must have been an earth changing date, eh? [21:19] ;) [21:20] no, but I've "wasted" much time since [21:20] Hehe [21:20] So you learned off of the available documentation and released warriors? [21:20] yes, kind of [21:21] Kind of? [21:21] when i started, this irc-channel was really, really empty, so there wasnt really other sources [21:21] As soon as you start to write warriors you learn really fast what each instruction does [21:23] I think part of my problem is that I'm shooting for my warriors to be to many things... How can I make a combination-does-everything-beats-everything warrior when I haven't even mastered the basic concept of the language? [21:23] * argiopewe things about some really simple bombers... [21:23] well ... you can't? [21:23] but actually you can! [21:23] I could, but it'll take me a couple more days. [21:24] Seeing as how I discovered that this even existed yesterday morning... [21:24] Take the code of other people and claim it to be your own :) [21:24] if you want to read about bombers, i have my stone-article's betaversion in the 'net [21:24] That's what those l4m3r n00bs do. I am not a l4m3r n00b. [21:24] I'd love to read it. Throw me a link Mizcu. [21:24] Join: gnik joined #corewars [21:24] hi gnik :) [21:25] THe king has come. [21:25] hello all! [21:25] argio: I have tried to write it to be easily understandable, but it might be a bit heavy [21:25] http://koti.mbnet.fi/mizcu/stonearticle.txt [21:26] Eh, I'll muddle through. I learned x86 assembly from the intel datasheets... [21:26] (I still havent recieved metcalf's fixed and suggestions) [21:26] Fluffy: are your warriors hand written or evolved ? [21:26] gnik: hard-core-optimized [21:26] :) [21:26] Hi Gnik [21:26] gnik: I take a basic form (like spl/mov/mov/mov/djn) and start to look for good values [21:26] ... until I can no longer improve the score [21:27] Sweet. I love guides on how to improve things. [21:27] sf: hi! I read in the logs that you're trying to create nano papers atm [21:27] he's lying argi, he's really only a softcore optimizer.. [21:28] bvowk: depends on what you call hardcore [21:28] but, I also don't have access to several hundreds of computers [21:29] Fluffy: Are you optimizing constants and adressing modes only ? [21:29] bvowk: Coming from you that could mean anything... [21:29] argi: I just like to tease fluffy. [21:29] gnik: Yes, but from time to time I optimize everything (including opcodes and modifiers) [21:29] gnik: I'm not doing very well :( [21:29] With name like that, who doesn't? [21:29] indeed, and with my allergy to cute.. well.. [21:29]