[00:02] MSG: [00:09] Join: Mannerisk joined #corewars [05:54] Join: MewLaue joined #corewars [05:55] MSG: Ping timeout: 240 seconds [17:26] Join: lain|afk joined #corewars [17:26] MSG: Client Quit [18:37] Join: bjoern joined #corewars [18:38] Part: bjoern left #corewars [19:21] Join: Fizmo joined #corewars [19:35] Join: pjwaffle joined #corewars [19:35] hi [19:35] hi [19:36] I have a question... has anyone used the microgp evolver? [19:36] I never could figure it out [19:36] have no experience with it [19:36] have you had experience with evolvers in general? [19:37] i've modified Yace for my own use.. little bit of redrace and redmaker [19:37] ah yes I remember you! [19:37] how did you mod yace? [19:38] took the source, removed lot of extra, tweaked the benchmarking a bit and the warrior-generation [19:38] cool... redrace had problems for me when I compiled with gcc did you have any? [19:38] i used binary [19:38] ah [19:38] mine gets a seg fault a few minutes in [19:39] anyway there really isn't any evolvers that I really like they all have issues [19:40] I'd make my own but I only know a little Python and thats about it [19:40] where little= if,then, while, random numbers, etc [19:40] have you heard of cuda? [19:41] nvidia cuda [19:41] yes [19:41] what would happen if you put that in yace LOL [19:41] they're going to replace it with some unified shader standard blah blah [19:41] CW might be a little bit hard to implement on cuda, heck, currently everything is [19:41] you'd beat really big hills if you evolved one on a geforce 8800 [19:42] that's what I have [19:42] i dont know really how many shader units would take to simulate a single core [19:42] and how much memory each would take [19:42] cuda has a lot of unused potential [19:43] evolution is very very very easy to parallerize, but its hard to implement as such [19:43] oh, im contradicting myself [19:43] all these dumb scientists are too lazy to do anything but write papers about it [19:44] oh yes you can rule the world if you genetically program some machine code app that modifies itself... but we're too stupid/lazy to do anything about it [19:44] hehe [19:44] a bit exagerated of courswe [19:44] but the basic idea remains [19:45] whenever some amazing technology comes out people are so amazed all they can do is write papers [19:45] what do you think, mizcu? [19:46] there isnt that much common ground between a research paper and reality [19:46] CUDA is real tested stable code [19:46] not some paper [19:46] and BOINC can use it too [19:46] there are a zillion people who are provoked by my previous sentence, but theres a point in it [19:47] cuda is underused because it will be replaced [19:47] GASP! [19:47] by what? [19:47] my memory is a bit flaky, but directx 13982 will have a directshader or equal [19:47] the unified shader arcitechture i mentioned earlier [19:48] crazy [19:48] sounds like another linux/microsoft fight [19:48] although Nvidia has been hammering through their story that cuda will be 100% compatible [19:49] microsoft will come out with some "revolutionary technology" and Linux communities will start something "easy-to-use" but more stable [19:49] the quotes are sarcastic [19:49] don't get me wrong im using linux this second [19:49] but it's somewhat... challenging [19:50] but microsoft just takes a year to hammer out some buggy piece of crap [19:50] i tested linux when there were 5 big distros on the market, once the number of "big" distros went over 10, i switched into something little more obscure [19:50] like? [19:51] oh, i have windows on my tabletop and laptop, and have no intention on switching, but my server is running on openbsd [19:51] I am running windows 7 on all my gaming stuff [19:51] and I made a mistake switching it before... no games for a year :( [19:53] nvidia does have one thing that confuses me... how can a gaming graphics card (the gtx 295) be 1.7 teraflops and cost $500 and a nvidia tesla cost $1300 and have 1.2 [19:54] license fees for using tesla's name! [19:54] like Nvidia Quattro [19:54] that sounds like a mistake on nvidia's part they wanted to have a gaming graphics card but forgot they had another important piece of the market that they basically ruined [19:55] and they're kind of pulling back on the gtx 295 too [19:55] they don't sell at tigerdirect site either [19:55] I think they noticed [19:55] yeah lol [19:56] it costs about $500k to build a petaflop computing cluster out of gtx 295s [19:56] but then you have to write code for it [19:56] according to a few mathmatical calculations I did [19:56] how much does one eat power? [19:56] >:) [19:57] :P [19:57] i think they'll eat a megawatt per day [19:57] but someones gonna rent all that power out and make a buck for it too [19:57] rent the speed out* [19:58] how much does IBM pay for their petaflop system? [19:58] I thought so a lit [19:58] lot* [19:59] well bye :) nice chat we had there [19:59] MSG: Quit: Leaving [21:00] MSG: Quit: ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]