[00:15] AGH [00:15] wrong button and 500 notepads open [02:31] MSG: Quit: Lost terminal [03:34] somehow i think blowfish is not exactly a best mascot to an OS [03:34] except a Microsoft-product [07:52] yes! yes! no! no! [07:54] just had a nice one coming, and it got outplayed by those that i hate [10:58] Join: Retro joined #corewars [10:58] mornin [10:58] I was hoping this channel wouldn't be too dead :) [11:02] then again maybe it is... anyone about? [11:32] Join: Fluffy joined #corewars [11:32] :) [11:32] Hi Retro [11:32] I've just read the logs and found that you were waiting :) [11:33] So if you have any questions don't hesitate to ask. [11:49] hey fluff [11:49] You've forgotten the 'y' ;-) [11:50] How can I help? [11:50] I've read the logs too :) [11:50] I spent the past few days doing a little study on core war and related programs [11:50] Nice to hear. We always need new players [11:50] (read the newsgroup from the first post to now amongst other things :) [11:51] and I figured I'd drop by and see how things were in the community now [11:51] Hmm ... unfortunately you've chose then wrong time. [11:52] main reason I found core war was because of writing my own similar program [11:52] After about 16:00 UTC people are a little bit more talkative [11:52] ah no problem I'll be around for a while [11:52] You write your own MARS? [11:53] I wrote my own corewar variant independantly actualy, so it's not a "MARS" atm :) [11:53] it has some other instructions but it's quite similar [11:53] Ah, I see. Did you publish it? [11:53] haha no I just built it a week ago :) [11:54] I hope that you have/had fun coding it. [11:54] I'm still working on the second version [11:54] oh yeah a great deal [11:54] I used the first version to evolve a few algorithms for a seminar presentation [11:54] Did they work? [11:55] actualy the second one is more core wars like in nature, the first one was used to operate programs outside the "core" [11:55] yeah, I evolved tank battle programs that duke it out in realtime [11:55] (3d simulated tank battle, although simplistic ofcourse) [11:55] But not RoboCode, right? [11:56] haha no, I wrote the tank program myself too [11:56] Join: Roy joined #corewars [11:56] * Fluffy kicks Roy [11:56] Hi Roy [11:56] Hi FluffY [11:56] And Retro :) [11:57] hey Roy [11:57] Retro: And do you plan to write some "real" warriors (for corewar)? [11:57] you're dutch right? [11:57] You said "Hey Fluff", I like you already ;-) [11:57] Yeah [11:57] me too [11:57] haha [11:57] Waar kom je vandaan dan? :) [11:57] yeah fluffy I'll write a few eventualy [11:57] * Fluffy ups global warming and waits for The Netherlands to be flooded. [11:58] mostly I'm thinking of some rather interesting things tho :) [11:58] Fluffy, don't.... we'll all be fleeing to Germany then, I bet you don't want that [11:58] btw. one of the instructions I'm using in my own Core to great effect is SLP (Sleep) [11:58] managed to implement it in a fairly non exploitable way [11:58] Roy: Think of the fun that I'd have. I could actually kick you ;) [11:59] You won't survive that Fluff [11:59] it sleeps a thread for (A field) cycles of it's own proces [11:59] SLP sounds like fun, intresting things you can do with that [11:59] How do you know, Roy, I can run pretty fast! [11:59] if there's no active threads it wakes a sleeping thread [11:59] and once a sleeping thread wakes up it's told how long it slept (B field) [11:59] And if there are active thread it has a timer? [12:00] yup [12:00] the telling it how long it slept has some neat implications [12:00] Nice, but that could seriously change the game [12:00] since you can use it as a source for a second thread to start if the first dies [12:00] so yes it's a serious change :) [12:01] but then I started without knowing about corewars [12:01] It will probably fade the different strategies.. [12:02] For example you could start a paper and if it gets killed the timer wakes up and starts to clear [12:02] Things like that.. [12:02] It is sad that there are so few (active) players. Otherwise it would probably be fun to test all those variations of corewar. [12:02] yeah, it's still limited ofcourse as your thread won't be auto-waked unless there's 0 threads awake [12:03] oh and roy I'm from Breda [12:04] (late response, lol) [12:05] what I'm currently doing is developing this CoreWar like system of my own design and sugar coating it [12:06] The problem indeed is getting players to play with it [12:06] essentialy I'm writing a smart, full 3D core simulator [12:06] an IDE to go with it [12:06] 3d? [12:06] There are hardly any players which play the normal version, so a new different implementation is harder [12:07] yeah fluffy it's a 3d representation of the core shown as a way to debug your program [12:07] Do you have a screenshot? [12:07] it's just a really accessible and "neat" way to show the system in action [12:07] I'll show you one in a day or 2 once it's up and running with propper graphics [12:08] right now I'm still writing some core interfaces and squeezing a few more megacycles of performance out [12:08] MIPS if you will :P [12:09] So far I haven't found a better acronym than MIPS :) [12:09] yeah :) [12:09] for us tech people that's fine ;) [12:09] tech people? [12:09] You must refer to bvowky with that ;) [12:09] technology savy people ;) [12:10] oh I did read a few MIPS benchmarks in the chat logs Fluffy [12:10] particularly of all the MARS environments [12:10] :) [12:10] but I couldn't see what system they were on so they weren't that usefull to me for comparison [12:10] I was trying to bench my cores performance against the MARS'es out there [12:10] huh? [12:11] and? [12:11] well I still haven't found any MARS benchmarks that stated both the MIPS in redcode and the GHz of the processor running the test [12:11] PyCorewar probably beat all others [12:11] so I didn't get a chance to compare yet [12:12] If I remember correctly I've benchmarked on a PII-866. [12:12] *PIII-866 [12:13] my current core which looks quite a bit like redcode runs at 16 MIPS on a 2.2 GHz (4400+) and 32 when run on both cores, ofcourse I'm not done optimizing yet [12:14] * Retro double checks [12:14] I have about 40-80 MIPS on bvowk's 2.0 GHz. [12:14] 94nop: 40 MIPS, '88: 80 MIPS [12:16] must be quite well optimized code then :) [12:16] mine is still all neat and shiny (no ugly optimalizations yet) but 128 instructions = 1 emulated instruction is a little steep ;) [12:17] The biggest problem is that not all needed data fits into the L2 cache. [12:18] which is odd, I thought it would, altho on several different pages obviously... [12:18] for smaller cores atleast [12:19] for '94nop I need 8000 * (4+4+4) bytes + 80000 * 4 bytes for optimal performance [12:19] that's about 512 KiB, but then then you also need a little bit for the code [12:19] yeah gotcha, the 80000 is for the special thread optimalization you discussed before I guess? [12:20] yes, probably [12:20] yeah you're right that's cutting it close [12:20] hehe ... I can't count [12:20] it is 8000*12 + 80000*4*num_warriors :) [12:20] yeah [12:20] And even that is suboptimal. [12:21] Having everything in L1 would be even better :) [12:21] haha yes [12:22] brb... [12:25] MSG: Ping timeout: 255 seconds [12:27] back [12:27] so those are the future threads for all warriors in that 80000*4 array [12:27] though wouldn't it be better for the chache to put those threads into a cyclic list? [12:28] it'd be smaller then the 80000 array [12:28] that 80000-array is nothing but a list of pointers (to a location in the core) [12:28] more like 8000*12 [12:29] yeah I know :) [12:29] I could use a 8000-array (per warrior) [12:29] but that would make the MARS slower [12:29] and I know that too, but then you'd have to loop the array [12:29] which would like you said slow mars down [12:29] I'm always open to improvements (altough I haven't touched PyCorewar in a while) [12:30] however if the size is the issue maybe a single link cyclic list would be better [12:30] it's like [12:30] Yes, then you have that 8000-array-version [12:30] struct node { InstPointer somewhere; node* next; } [12:31] It is nothing more than a simple (cyclic) list [12:31] yeah fluffy, but with the array [12:31] you need to check if arraypos > arraysize and loop [12:31] or alternately used % [12:31] use* [12:31] or any other way to loop the array [12:31] with a cyclic list you don't [12:32] yes and if I'd use a real list I'd have even more code to manage the list [12:32] goddamnit, i spend an hour diving in wikipedia/messaging a buddy/watching vidplayer and youve written a new bible already [12:32] Hi Mizcu [12:32] hehe well it's intresting to theorize about it I guess [12:32] tho doesn't insertion into the array cause you issues? [12:33] Why would I want to insert sth? [12:33] like when a process splits and you need to start accounting for the new thread in the future... [12:33] All I need to do is to take away from the start and append to the end [12:33] ah yes [12:34] ic how you're using it [12:34] haha mind you I need to do a little mental jumping between redcode and my own system [12:34] (due to sleep and such it's not that easy for me) [12:34] No problem :) [12:34] (the sleep instruction that is) [12:35] I do need some strange instertions at times, tho they might be weeded out due to simplification [12:35] Redcode is already convoluted enough. No need to implement SLEEP and all that :) [12:36] I'm trying to somehow ease up on the ammount of modifiers and in general make things look a little "easier" [12:36] the point I was making with the cyclic list is that it doesn't need any memory re-aranging, all it needs is insertion after SPL [12:36] question: why 3 dimentional memory? [12:37] oh no mizcu [12:37] no 3d memory :) [12:37] just a 3d view of a chip with the memory on it [12:37] and the CPU floating over it reading instructions and all :P (not for realism purposes) [12:38] just fancy looking and very easy to read all the steps of the process for debugging (even if you are clueless about code at first) [12:39] very few probably remember this, but i once read that there is not sense in implementing multiple dimensions in a core, because computer memory itself is one-dimentional, and i keep that in mind [12:39] yeah [12:39] you're right mizcu [12:40] ga_war generated warriors and set them on a 2-dimentional array, like bacteria on a petridish [12:40] and nobody uses it nowadays, surprise? [12:41] but besides that, when visualizing memory as 3d blocks in a "cube" it wouldn't be possible to keep an overview [12:41] and more than three(plus one) dimentions needs an savant to understand [12:41] the thing I'm doing is showing a 2D flat representation of the looping 1D core memory (just like all the corewar programs tend to do) and then show that grid using fancy 3d graphics [12:42] yeah, have nothing against that [12:42] along with an in core tracker that highlights, zooms in on and slows down important events [12:43] (Will plans to use 3-dimentional scoring for hes next evolver, which is actually very good idea) [12:43] so you can see the cool parts of the battle taking place and watch the rest in fast forward, unless you choose otherwise ofcourse... [12:44] I was tempted to write a quick evolver for CW [12:45] and I obviously will for my own system, if just for testing and academic purposes [12:47] * Fluffy waves [12:47] bbl [12:47] MSG: Quit: fluffy.i < 1, # 42 [12:48] The main reason I started reading the whole corewar archives, logs etc. was to look for interesting ideas for new commands people might have suggested. [12:48] ofcourse quite a few of these make little sense but it's usefull to read the ideas all the same [13:05] * Retro is back [13:08] Join: fiveop joined #corewars [13:10] hey fiveop [13:44] hi [13:48] oh hey there :) [13:49] that was a late response, lol [13:49] whatcha up to? [14:00] buying a train ticket :P [14:22] hehe ic [15:02] hrm. [15:07] hey man [15:07] afternoon [15:15] its morning here [15:24] ic [15:24] 4:24 PM here [15:25] so you're yet another old timer :) [15:26] it's nice to see some people really stuck to this game for a long time [15:35] I'm not so old.. [15:35] ;) [15:46] :P [16:07] lol [16:59] hrm. [17:01] I think I have a new (better) K topper.. [17:02] 28 MIPS and counting :) [17:02] mips? [17:03] million instructions per second [17:03] are you running a calculator? ;) [17:03] I'm running my own sort of "redcode" [17:03] I see [17:03] * bvowk is all about the Gflops. [17:03] it's not compliant with redcode standards as I designed the language myself [17:04] but it's quite close [17:04] I'm tweaking the performance [17:04] why are you making a new redcode? [17:05] we're small enough already without fragmenting further [17:05] well first of all I am because I didn't know about redcode when I started it... [17:05] I suppose thats an acceptable response ;) [17:05] :) [17:05] I just came to this community because it caught my eye afterwards and I wanted to check it out [17:07] mostly for inspiration sake, I actualy started this whole project because I needed something to show off at a seminar, clearly many good ideas have been thought of before :) [17:07] apparently metcalf says hello [17:07] "* jm says hi to Retro" (length 5) by "John Metcalf" has challenged the Nano [17:07] hill [17:08] lol [17:08] :) [17:09] haha [17:10] url? [17:10] wow.. I've got an 8 point lead on nano [17:10] sal.math.ualberta.ca [17:12] nice :) [17:12] I assume he's reading the irc logs then [17:15] * Retro whips up a nano to return the greetings... [17:20] bbl [17:20] dinner time [17:21] looks like I'll end up somewhere around 35-40 MIPS (but that's including for example a sleep mode for threads that regular redcode doesn't have) [18:32] Join: Roy joined #corewars [18:40] Join: John joined #corewars [18:41] Hi :-) [18:41] Hi [18:41] Aww, no time, have to go, visitors [18:41] :-/ [18:41] * John waves to Roy [18:41] Just wanted to start the sieving and look around who was here [18:42] * Roy got a birthday to celebrate ;-) [18:42] * Roy waves! [18:42] MSG: [18:56] Join: gnik joined #corewars [18:56] Hi Gnik :-) [18:56] Hi bvowk, fiveop, John, Retro, log_guy ;) [18:57] You stole my name, I used test tube before :-P [18:58] hey guys [18:58] *back* [18:58] yes, i remembered someone used it before, maybe that's why it came to my mind ... [18:59] Hi retro [18:59] :-) [18:59] hi retro [19:00] I have more ties than you, 48.9 :-P [19:00] It's paper/imp with mov.i #1,1 imps [19:00] nice! [19:04] I don't think scanner/imps score very well on the current hill :-( [19:18] Join: Fluffy joined #corewars [19:18] :| [19:19] hey guys, quick question, what's the hill key for nano? [19:19] ;redcode-nano [19:19] figures, lol [19:19] hi fluffy! [19:20] Join: jon_ joined #corewars [19:20] Hi jon_ [19:20] hello. [19:20] Hi Fluffy, Jon_ [19:21] newb here [19:21] and the email is? . [19:21] koth@* [19:21] bvowky: How is mitzi2 going? [19:21] stupid paste [19:21] SAL: [19:21] koth.org: [19:21] tnx [19:23] lol [19:23] Error processing your e-mail: I don't understand this line:--_f4697132-fd3f-454b-a814-e10ad1c76a42_ [19:23] jon_: Don't hesitate to ask questions [19:24] I'm quite sure that line wasn't in my mail, lol, maybe it wasn't ascii :-/ [19:24] stupid hotmail :P [19:24] Fluffy: thanks:) [19:25] Retro: If I remember correctly the first line in your mail body has to be ";redcode-..." [19:25] i remember trying to get into corewar when i was like 16, but never really grokked it, took another look yesterday, and am grokking it.. so i figured i'd give it a proper go now (at 24) [19:25] hehe [19:26] jon_: I guess that you've found the Redcode tutorial then, haven't you? [19:26] http://vyznev.net/corewar/guide.html [19:26] yeah fluffy I had that in there [19:26] yep.. read it on the subway home from work yesterday. I'm reading the coreware book now (the 2 chapter book on imps and stones) [19:26] maybe my hotmail was sending it html encoded [19:27] John: Don't you use a hotmail account from time to time? [19:28] Yes [19:28] It sents plain text for me, I assume there are some options somewhere [19:29] jon_: Then you should send some warriors to the beginner's hill :) [19:32] argh ... I can no longer hear it. I from now on refuse to believe that earth's climate changes! [19:34] hehe [19:34] Fluffy: of course. [19:34] has anyone thought of implementing corewar in hardware? [19:35] You seem to have pretty much money then, jon_ ;) [19:35] Couldn't it be done on one of those programmable logic thingies? [19:35] i was thinking it could be done on a pic [19:35] but with a small core. [19:36] At least I can see no commercial sense in that. And FPGAs are still a little bit too expensive [19:36] a top-end pic could probably only do a core size of 400 [19:37] hmm ... which would only include nano-settings and nano can be done very fast on recent cpus [19:38] yeah [19:38] Retro: Is your hotmail-problem solved? [19:38] most of the pics i have about 2k of avilable memory [19:38] but actually i was thinking of a 'tamagotchi' style device, where you could assemble programs on board (via an LCD, primitive keyboard, etc). and run them on the device it self. [19:39] Are you talking about a Palm? [19:39] with some method of transfering warriors to one to another [19:39] It shouldn't be that hard to port an existing MARS to it [19:39] Fluffy: a homebrew corewar playiing palm :) [19:41] going to try it again fluffy [19:41] was off for a sec [19:41] Take your time [19:41] jon_: might be a little bit slow nonetheless [19:42] jon_: I wouldn't want to benchmark a '94nop warrior on it [19:42] I think there's already a couple of Mars's for the Palm [19:43] Got a link? [19:43] Sorry no [19:43] Damn! I have to use my own brian again [19:43] I think there was some discussion on r.g.cw once [19:44] Fluffy: can clock a pic18 series chip up to 40mhz, iirc. [19:45] hmm ... don't know what you are talking about, but I know how to use a fork ;) [19:45] I'm not that much into hardware :) [19:47] ah. [19:47] jon_: In any case it might be tricky to fit everything (code, core, process queue) into it. [19:50] Even a size optimized code for a MARS is quite huge because of the convoluted way some instructions behave [19:50] you're probably right. [19:51] hrm. [19:51] Hi bvowky. [19:51] Hi Bvowk [19:51] whats up? [19:52] We've decided that you should give use access to mitzi and mitzi2! [19:52] *us [19:52] so when is your cheque for your share coming? ;P [19:53] How is your play money called again? Canadian dollar, right? [19:53] No problem then ;) [19:54] while it is colourful like monopoly money, its on better foundations than the USD ;) [19:54] Do you call it actually "mitzi2" or did you choose a different name? [19:55] 2mitzi [19:55] like 3jane [19:55] :) [19:57] so doing anything of note fluffer? [19:57] how about you met? [19:58] * Fluffy kicks bvowk [19:59] Yes, I think that I've figured out a way to win significantly more against a q^4.5 [19:59] Just preparing stuff for nw03 [19:59] w00t! nice met! [19:59] John: Preparing as in "it is finished soon"? [19:59] and soon = "now"? [20:01] Have to go in 2 mins [20:01] * Fluffy waves [20:01] oh and fluffy [20:01] huh? [20:01] Join: sascha joined #corewars [20:01] I did manage to get it around 35 to 40 MIPS by now :) [20:01] Hi Sascha [20:01] Retro: Nice :) [20:02] it's a little heavier then redcode [20:02] since it's got to track sleeping processes as well [20:02] but a nice improvement all in all :) [20:02] I've gotta check out for tonite, but I'll be back! :D [20:03] * John waves [20:05] hey! chimpanzees have been seen wielding spears! [20:05] thats *EXCELLENT*... pretty soon they'll have nuclear weapons and mankind will have to fight for its right to continue existing ;) [20:06] You've forgotten about sth. skynet-like ;) [20:06] bah, once you've figured out its easier to kill with tools, the rest comes easy. [20:09] MSG: [20:09] bvowk: And what about the ties ;) [20:10] chimps don't wear ties ;) [20:11] lol [20:12] hey bvowk! [20:12] nice warrior you got there on #1 [20:12] Has anyone got a pic of bvowk wearing a tie? [20:13] You could always take the pics from his homepage and use GIMP/Photoshop [20:13] where's the nano hill? [20:14] jon_: http://sal.math.ualberta.ca/hill.php?key=nano [20:14] its in my server room [20:14] fair enough [20:14] I've worn a tie on 3 occasions. [20:14] jon_: and there's a little bit of info about the nano hill at http://corewar.co.uk/nano.htm [20:15] bvowk: your warriors are evolved? [20:15] jon_: A lot of warriors for nano are [20:15] especially bvowk's [20:15] yup, even *I'm* evolved. [20:16] interesting [20:17] you evolving jon? [20:19] i'm going grey, does that count? [20:19] heh [20:19] I was asking if you were evolving *REDCODE* [20:20] nope. [20:20] (we'll give you the benefit of the doubt on your reproductive endevours) [20:20] There's a quick-start evolving kit here for the nano hill - http://corewar.co.uk/yacenano.zip - http://corewar.co.uk/ankerl/yace_1.2-inst.exe [20:20] haven't written a warrior yet. started looking into redcode yesterday [20:20] (Windows only) [20:21]