[00:30] MSG: Quit: Lost terminal [09:00] Join: fiveop joined #corewars [10:24] Join: lol joined #corewars [10:24] MSG: Client Quit [10:29] Join: cees joined #corewars [12:51] Join: sf_ghoul joined #corewars [12:53] MSG: Quit: Lost terminal [12:54] Join: sf[2] joined #corewars [12:55] MSG: Ping timeout: 255 seconds [13:09] * sf[2] presses her face against the screen and to see if bvowky's sleeping [13:09] * sf[2] kicks lazy bvowk! [13:14] MSG: Quit: Think your current client is sexy? Check out Bersirc 2.2! [ http://www.bersirc.org/ - Open Source IRC ] [13:19] Join: Core_old joined #corewars [14:32] MSG: Read error: Connection reset by peer [14:59] ow! [14:59] sf is cruel. [15:48] Well, atleast i am not lazy. I wrote a whole page of pseudocode for my evolver today during the programming-lessons (and repseudocoded it during networking lessons) [15:59] Join: fiveop joined #corewars [16:01] Join: Fluffy joined #corewars [16:01] :) [16:02] . [16:19] hello [16:20] hey miz [16:20] hmm [16:20] my paper got knocked off the hill [16:21] Then write a new one. [16:22] excellent plan. [16:22] I know. I've already trademarked it! [16:22] Then write a new one (tm): [16:22] actually.. I've got prior art on that.. [16:22] except I pulled it up one level of meta ;) [16:22] i've got a patent on it. [16:23] But I have WoMD and am ready to use them ;) [16:23] those nice fellows at HP cross shipped me a nice new 4108 chassis! [16:23] I take back some of the bad things I've said about them [16:23] for free? [16:24] no, to help with the whole lifetime warranty thing [16:24] my one 4108 chassis blew some slots [16:24] [3jane:~/work-tmp/ewon/save$] ls |wc -l [16:24] 18460 [16:24] crap. [16:25] Did you delete some warriors? [16:25] [booty:~/ewon/save$] ls |wc -l [16:25] 69605 [16:25] double crap! [16:25] no, I'm over-run with new warriors! [16:25] they must have started not-sucking [16:26] lol [16:26] then send the best to the hill [16:26] ... or all [16:27] that would make it the oldest hill then [16:27] I think metcalf would fly out and beat me to death with a rolled up copy of scientific american if I did that [16:27] Yes, but where is the problem? [16:28] I'd really love to see that ;) [16:28] I don't want met to waste his corewars time on transatlantic flights? [16:28] hmm ... he'd probably write more than your 80000 something warriors during that time [16:29] all of them definitely non-sucking :) [16:29] are you going to release the code of some of your evolved warriors? [16:29] yeah, I was thinking about releasing all of the nano warriors.. [16:30] you want a tarball of 760k? [16:30] lol. [16:30] sure. [16:30] How did you decide which warriors are in it, bvowk? [16:31] its all the nano warriors I have [16:31] 760000 warriors or 760 KiB? [16:32] 760,000 warriors [16:32] size? [16:33] bvowk still hasnt published one of those 79cd's =/ [16:34] I have no idea yet, I haven't made it.. [16:34] I was running a score vs koen and infnano on it.. [16:34] it just finished a few hours ago [16:37] [3jane:~/work-tmp/jnsl$] find clean_save -type f -name \*.rc |wc -l [16:37] 751793 [16:38] so its not quite 760k after I removed everything that scores less than 100 [16:40] when its done tarring I'll post a url [16:46] \o/ [16:58] looks like its about 30M [16:58] sorry.. [16:58] huge-nano-pit.tar.gz 100% 62MB 2.7MB/s 00:23 [16:58] woah. [16:59] www.math.ualberta.ca/~bvowk/files/huge-nano-pit.tar.gz [16:59] enjoy. [16:59] you know, i have a feeling thats has been pretty crappily archived [17:00] meh.. default gzip level [17:00] -rw------- 1 bvowk wheel 59226366 Mar 12 09:53 huge-nano-pit.tar.gz.bz2 [17:00] bzip2 didn't help much ;) [17:00] * bvowk ducks. [17:01] hmm ... [17:01] well, i would try compressing it better, but i dont want to see what NTFS will think of 760k files [17:01] its file-allocation table would go over a gig, methinks [17:03] they're sorted into directories by form.. [17:03] so its not quite so ugly [17:03] you use ext3? [17:04] bah. [17:04] ufs [17:04] I'm not linux user [17:04] so, on my desk, there is 8TB of disk! [17:04] oh yeah, you were BSDard [17:05] 16 x 500GB disks [17:05] I remember eyeing a 8TB setup once that filled a whole datacenter.. [17:05] now it fits in a smallish box [17:07] getting 300+K/s on that file. [17:08] its a Univ server, you expected something laggy? [17:08] nope :) just happy [17:08] oh, I was about to appologize for the 300K/s [17:09] we've been having problems with the commercial feeds [17:09] appology accepted [17:11] MSG: Quit: Leaving. [17:11] where'd fluffy go [17:11] without even a wave.. [17:11] well. [17:11] if i ever get to put that mediaserver up here, i am certain i will install something that will make people cry when hearing (no, not windows, that would make me cry) [17:12] Plan 9 from Bell Labs seems awfully nice on the "security through obscurity" -department [17:13] however Solaris might have some actually read-worthy documentation available [17:13] just install openbsd :) [17:13] its nice in the security through security dept [17:13] bah [17:16] that would make you sort-of-happy [17:16] http://www.techworld.com/security/news/index.cfm?newsid=8227 [17:17] if your goal in server administration is to not make me happy... [17:20] MSG: Remote host closed the connection [17:39] afk [18:12] Join: brx joined #corewars [18:13] * bvowk sells brx's kidneys on ebay [18:17] * brx bids a few schilling [18:17] hrm, I guess that means you know about the condition of your kidneys and are bidding accordingly eh? [18:23] so, done anything interesting lately brx? [18:30] screwed a mongolian. [18:33] okay I just remembered that this channel is logged. [18:35] anyway, nothing too interesting. if you are into illegally downloading regular tv show torrent releases of, say, battlestar galactica or heroes, me and michal have written a little ruby script to track episodes you have already seen and automatically download new ones as they come out. [18:37] essentially it automates the task of going to mininova or torrentspy or piratebay, searching for the series, grepping through the list of search results and downloading good torrents. [18:37] then there is the horrible Imagine Cup [18:38] a roguelike and of course the evolver [18:44] Join: jon_ joined #corewars [18:45] bvowk: did your evolver come up with the 'spl/mov/mov/mov/djn/' pattern itself, or is that a predefined pattern that's someway hardcoded? [18:49] MSG: Quit: Lost terminal [18:49] either by itself or by chains( which i think are not included in the simple evolver) [18:53] sheesh, people dont even bother to search the files nowadays [18:54] persons intellectual growth demands use of creativity and training ones brain, just couching and watching tv/tvrips doesnt do any good [18:55] which is exactly why the intellectual development is slowing down in the western nations [18:55] end of rant [18:57] O.o [18:57] its not that hard for a nano-evolver co come up with smmmd anyway, especially if there is crossover or expansion by weighted replacement to help spread when one appears [18:58] yes [18:58] i was curious to know if it was biased towards that format. [19:04] Will's evolver would be one to heavily bias towards such format, which is silly, considering it uses 10 times the time to get the structure compared to Yace [19:05] wow. [19:05] 10 times? [19:06] it tries to find the imporant loops and or pieces of code and modify and or preserve them [19:07] too bad it doesnt work in an evolving enviroment, where it is generally more useful to screw around with the important code [19:07] not preserve [19:27] jon: the evolver started with completely random code.. [19:28] so everything that comes out of it is the result of it scoring better than what came before it [19:29] (although, its not *QUITE* that simple, as some runs loaded warriors generated from previous runs on other machines..) [19:30] goddamnit [19:30] the jnsl evolver wasn't really biased towards anything, it did use bias warriors to keep the scores from going completely static.. [19:30] its supposed to be 2048/512 ADSL but it chokes near 100k/s [19:30] and to lightly direct the evolution towards warriors that were human competitive [19:31] mizcu: don't forget that bandwidth is limited by latency [19:31] unless i am downloading from, say, local school server [19:31] higher latency = slower tcp flows [19:31] so at 100ms, a 100Mbit connection is nowhere near that fast [19:32] yeah, yea [19:32] (well, there's that and that ISP's oversell the bandwidth they actually have about 500x over [19:32] i need a focking uncapping to this DSL [19:32] but usually when people complain its because of the latency [19:33] windows isn't really optimized for high throughput [19:34] (windows isn't optimized for anything other than revenue maximization) [19:34] http://people.freebsd.org/~kris/scaling/mysql.html [19:34] so, it's only taken like 6 years, but smpng is finally starting to look good [19:34] (some of those years were *PAIN* tho.. serious *PAIN*) [19:36] me puts some "WAKE THE FUCKING UP" -music to play [19:37] your fucking fell asleep? [19:37] my head want to fall asleep [19:37] not while i have new mythbusters coming in under an hour [19:38] why not nap? [19:38] napping is good. [19:39] because then i wake up 8 at morning and feel like having taken tranquilizers [19:40] what the.. [19:40] dog just came to my room, went to my bed, sniffed for 2 seconds, and left [19:41] hes supposed to be asleep [19:41] (telling useful information like that is usually a sign of not enough IRC) [19:44] Mizcu: discover free-running sleep [19:44] i already do that during school-days [19:44] Mizcu: and if you are downloading myth buster episodes, maybe pouncer is helpful :) [19:47] Join: Fluffy joined #corewars [19:47] hrm [19:47] :) [19:49] where'd you get off to earlier fluffy? [19:50] Killed XWindows and didn't bother to restart my IRC client [19:50] Fluffy: you need to use a better irc client and then run it in screen. [19:51] I use screen, but what use would it be on a computer, which I power down every evening? [19:51] power down? [19:51] I don't understand.. [19:51] you turn a computer off? [19:51] lol [19:51] do you move it? [19:52] I like it quiet in the evening and especially in the night, so I switch off all of my computers [19:52] Fluffy: even i use screened irc-client nowadays, so maybe it is time for you too [19:53] Fluffy: do you have another box that you do not power down? :) [19:53] nope [19:53] Fluffy: move one into the toilet. [19:53] into the toilet? [19:53] That computer wouldn't live very long. [19:53] wouldnt it get, you know, down the drain? [19:54] and then youll have a shitty computer [19:54] brx: You probably mean the bathroom [19:54] Fluffy: no, I mean the toilet. I wouldn't want to be bothered by the noise while taking an otherwise relaxing bath [19:54] * Fluffy sighs [19:55] heh. [19:55] But I prefer not to be electrocuted [19:55] I have heard of people drinking out of a bathtub, namely eggnog [19:55] do you have a dog? [19:56] namely people named eggnog [19:56] Not at the moment, but I know that I'll be forced to have one :) [19:56] bathtub has only one use [19:56] to brew in [19:56] mizcu: cooking batches of meth? [19:56] cute dead bodys into pieces? [19:56] *cut [19:57] of course, some of you might have read the lsd-massproduction file using bathtub [19:58] Hey, nano has become too quiet :( [19:59] sorry fluffy.. I'm not nano'ing anymore [19:59] I know [20:01] brx: Got any better suggestions on where to run screen? [20:01] Fluffy: get a free shell account somewhere maybe [20:11] Usually I'm denied shell accounts because of my stupid nick ;) [20:11] sdf.lonestar.org is pretty good [20:11] but i think you have to pay a $36 one off to get screen [20:12] sdf is fun [20:13] satellite defence fund? self defense force? [20:13] social-democrat floofs? [20:13] shut de fuck? [20:14] super dimension fortress [20:14] sexual deviants faction? [20:14] http://sdf.lonestar.org/index.cgi?faq?BASICS?01 [20:15] I never thought it possible to get free shell accounts [20:17] hey, you can even get free* sex over internet [20:17] * movie and dinner not included [20:18] Why would I want that? [20:18] 'caus its free [20:18] Everybody can kill himself for free, but usually only a very tiny fraction relaly uses that opportunity. [20:19] *really [20:19] but you cant get that from the internet [20:19] And that makes it better? [20:19] bah, you are unfun [20:20] No, I still have a life ;) [20:20] you can get couple bux out of it at ebay, you wont be missing it anyway [20:21] wow, it's getting late and I am terribly tired [20:21] I still don't get the idea of SDF [20:21] some dumb fucks? [20:22] Super Dimension Fortress [20:22] >:( [20:22] Fluffy: most people find the free account a little too restrictive, so they upgrade [20:22] ok, but what's the reason for its existance? [20:23] Only to provide shell accounts? [20:23] yep. [20:23] its a public access service. [20:28] Join: fluffy2 joined #corewars [20:28] hmm ... seems to work [20:28] * Fluffy kicks fluffy [20:28] MSG: Client Quit [20:38] hrm. [20:38] enjoy [20:39] 51 2708ba11-a026b4b6-f70e260a.rc 87.35 WTL% 25.3 11.5 63.2 1 [20:39] ouch. [20:39] thats some score noise.. [20:40] it scored 162 in the original test ;) [20:40] thats only a 50% inflation of its ability [20:40] i guess thats an impy warrior [20:43] guess again.. [20:43] ah, wtl [20:43] not wlt [20:43] Join: NewFluffy joined #corewars [20:43] Nick Change: Fluffy changed nick to OldFluffy [20:43] Nick Change: NewFluffy changed nick to Fluffy [20:43] Bye, OldFluffy! [20:44] MSG: Quit: oldFluffy.i < 1, # 42 [21:08] hrm [21:09] thats my line [21:10] ok, ok [21:10] I'll continue to smile then [21:10] :) [21:10] ... noseless [21:11] Maybe I should change that into a Canadian smile. [21:12] :) [21:12] That should be about right [21:21] bbl, sleep() [21:22] hmm [21:24] Join: sascha joined #corewars [21:25] Oh, silence here [21:25] no [21:27] MSG: Excess Flood [21:27] apparently he had alot to say [21:27] yes, he had [21:28] Join: sascha joined #corewars [21:31] Join: fiveop joined #corewars [21:32] Hi fiveop [21:48] Join: pkhuong joined #corewars [21:49] hi pkhuong [21:49] hi [21:59] O [21:59] wpp[s [21:59] missed my homekeys. [21:59] ? [21:59] homekey? [21:59] I'm reading the hints.pdf on www.firetrench.com/corewar/ [21:59] and it says at one point: [22:00] djn.f -2, is that comment right? [22:00] yes, because of the .f [22:01] the B-Field @ DJNSTREAM1 is decremented [22:01] i was under the impression that djn.f only decremented the B field but only jumped if both A & B werent 0 [22:01] and both fields on the location, that DJNSTREAM points to (after decrementing), are decremented by the djn.f [22:02] djn.b does only decrement the B-field [22:02] djn.a the A-field [22:03] watching the execution of such an instruction usually helps [22:03] hmm [22:03] i thought i did [22:03] * jon_ tries again. [22:03] try a "djn.f # 0, < stream" [22:03] with stream EQU -100 (or so) [22:04] the "< stream" decrements the same B-field over and over again [22:05] so it gets a new value with each execution [22:05] the djn.f decrements these new fields once [22:06] (cdb) step [22:06] 00002 DJN.B $ -1, < -100 [22:06] (cdb) -100 [22:06] 07902 DAT.F $ 0, $ -1 [22:06] (cdb) -1 [22:06] 07901 DAT.F $ 0, $ -1 [22:06] [22:06] shouldn't 7901 be dat -1, -1 then? [22:06] djn.F not djn.B [22:07] you're right, i didnt put any modifier on the djn in the code. [22:07] doh [22:08] thanks.. i knew it was something small i was missing. [22:09] obvious mistakes are often the worst :) [22:09] I know from experience [22:11] I've been looking for a wrong '-' a whole month [22:12] but only in 10k of simple JAVA code [22:13] doh [22:13] yup [22:13] i do perl for a living, so i can sympathise entirely. [22:14] I've fallen in love with Python and that won't probably change [22:19] jon_: Does everything work now as described? [22:19] yep, confusion gone [22:19] now i'm curious about spl.x [22:20] * jon_ rtfms and gets the answer. [22:20] spl.x = spl.b [22:20] actually spl.whatever = spl.b [22:21] excellent. [22:22] but not very intuitive [22:23] Fluffy: find some way to generalize spl to have an insn modifier, and submit it for ICWS 2014. [22:23] No, but I'll use that generalization for Redcode 2007 [22:23] aka Fluffy's Redcode ;) [22:24] fluffcode [22:25] no, that sounds like programming mixed with porn [22:27] pkhuong: I'm still interested in implementing spz (split, if zero) and spn (split, if not zero) [22:27] part [22:27] Part: sascha left #corewars [22:27] not quite [22:28] hehe [22:28] pkhuong: add predication for all the insns instead [22:28] (and remove sne/seq) [22:28] erh [22:28] Fluffy even [22:28] that would change cw a lot [22:29] *if* we could make everybody use it [22:29] not that much. It'd open up a *lot* of possibilities, though. [22:30] Do you have any special ideas what to write then? [22:30] oh hell. Or read the CTM (close to the metal, ati's gpu programming asm) programming guide [22:30] and try to port the feel to cw ;) [22:31] Fluffy: not really. Haven't really played cw in years [22:31] Every time I read about GPU I feel the urge to implement a MARS, but then I read how slow that would be :( [22:31] pkhuong: Then it is time to start playing :) [22:31] :( [22:32] i still think i could implement a cut down ICWS-88 on a microcontroller [22:32] it'd be slow as hell though. [22:32] depends on the microcontroller [22:33] jon_: on an fpga, otoh... [22:33] Why don't you emulate it on a faster CPU to make it run faster [22:33] fluffy: why don't you implement pspace in pycorewars [22:33] (slacker) [22:34] bvowk: Yes, why don't I implement pspace in PyCorewars [22:34] bvowk: when are you ever going to add pspace to your evolver anyway? [22:34] pk: its not for the evolver, its for benchmarking against K [22:35] but benchmarking P-spacers is a really bad thing [22:35] because you can't :( [22:36] Fluffy: huh? [22:36] pkhuong: do you know fpga? [22:36] pmars -P does only test all possible start positions [22:36] but not all possible states of a P-Spacer [22:36] jon_: nope. [22:36] pkhuong: nor i, doh. [22:37] fluffy: all possible start positions are really all the possible inputs into the p-spacer system tho [22:37] I got a lotta hardware reading on my work queue. [22:37] bvowk: Maybe my brain doesn't work, but I don't think so [22:37] bvowk: not exactly. [22:37] order is important. [22:38] and vanilla cw allows repeats. [22:38] fluffy: the only inputs into the pspacer for state are the results, and the results will be the same w/ -P [22:38] so you now have (coresize-minlength)^[number of rounds] [22:39] abut that [22:39] *aobut [22:39] grmpf [22:39] *about [22:40] bvowk: Think of a P-spacer, which accumulates wins/losses and chooses strategy upon that ratio. Then run -P once and the -P in random starting oder [22:40] both results should differ [22:41] maybe only a little bit [22:42] Fluffy: however, as any sampling/measuring expert will tell you, you don't have to test all the possibles to have a good benchmark [22:42] heck, you usually can't. [22:43] yes, but for '94nop you can [22:43] so I do it [22:45] and bvowk might do it for '94 with P-space ;) [22:50] Fluffy: a warrior's perf characteristic could depend on the # of rounds too [22:50] what will you do? [22:51] yes, I've already thought about that ... especially when optimizing for ', but since I can't/don't know its F-value, I don't see how to optimize for a given number of rounds [22:52] overflows [22:52] huh? [22:53] it's using moduolar arithmetic [22:53] once you hit the thousand battles range [22:53] it's very possible for a calculation to overflow [22:54] but no hill uses >1000 rounds [22:54] -P does [22:55] what does -P has to do with modular arith? [22:56] apart from adding +1 the the last position ;) [23:03] interesting [23:03]