[01:10] Join: brad joined #corewars [01:10] hey [01:10] re [01:30] bye [02:13] MSG: [12:20] Join: Core_old joined #corewars [13:15] MSG: Ping timeout: 255 seconds [14:59] Join: Fluffy joined #corewars [15:00] :) [15:00] aye [15:16] bbiaw [15:16] MSG: Quit: Leaving. [16:00] Join: Fluffy joined #corewars [16:00] :) [16:22] Time to generate a new nano warrior :) [17:56] good morning! [17:58] 'nin [17:58] ares: nope [17:58] bad morning? [17:59] no morning [17:59] you know, i dont live with CET, but with "uptime" [17:59] but that doesn't have a morning [18:00] all relative :-) [18:00] Mizcu: yesterday you asked "http://www.ociw.edu/~birk/COREWAR/94/HILL/fatallure.red <- errmm.. is this pre or post HSA" [18:00] what is HSA? [18:01] he scans alone [18:01] name of a warrior [18:01] ah! thx [18:01] on '94nop Koenigstuhl [18:02] but now it is more a type of scanner [18:11] hmm.. i have put together some warrior-ideas for the beginner-benchmark(s), but i still could use couple possibilities [18:12] what's in it so far? [18:12] the goal would be to have two benchmarks: one for total starters, and one that is somewhere around wilkies and wilbez [18:13] together with a short README, which tells how to benchmark! [18:13] starter, so far only Mice, Commando, Trident, Nightcrawler/No ties allowed and XTC [18:14] about the "medium", 2 papers, PI, PS, Stone, Bomber, SI, SI, 2 oldie scanners, oneshot, blurish [18:16] Timescape for paper, Paper-imp could be themystery or diehard [18:17] Flashpaper is an example of early Paper-stone, but a little stronger could do [18:17] those "medium" paper need a little qscan. otherwise all these warriors are too short to warrant a qscan in the beginner's warrior [18:18] Q^2? [18:18] yes, why not [18:18] For plain stone a Tornado II would go good, Torch for bomber [18:18] Aeka will be excellent stone/imp [18:18] but before you use one, you should test, wheter a q^2 doesn't make the warrior worse [18:19] second stoneimp is a bit of a question, Bluefunk? Newt? Stoney the second? Impfinity? [18:19] (Newt would be overpowering, i think) [18:19] choose from their position at Koenigstuhl [18:19] i.e. the lowest ranking [18:20] gotta see [18:20] For scanners, Rave is a favorite of mine, so id like to include it [18:20] Mirage for a blur-scanner [18:21] But what about a oneshot (millllion possibilities) and other oldie-scanners [18:24] met's scanner-article does have couple suggestions, though [19:05] hoi, ares [19:05] aye? [19:05] how much problem would be to create a small program or a script for the beginner's benchmarker? [19:05] i cant really participate on that discussion about benchmarks, due to a lack of knowledge and expirience [19:06] you mean as a web page? [19:06] as a program to include with the warrior-pack [19:06] ares allready has a built in benchmark function [19:06] i just have to add a directory to the zip file with your benchmark warriors [19:06] hmm [19:06] well, corewin has too.. [19:07] fluffy, do you have an unix-script for benchmarking? [19:07] i maybe will put a benchmark function to ares' home page [19:07] Mizcu: Define "unix-script", but yes I have several scripts for benchmarking [19:09] something i can include in the webpage of warriorpacks to be used on nix [19:09] since i can suggest corewin or ares for windows already [19:20] ooohkay, koening doesnt have tornado v2.. [19:23] id really like to include Digitalis in the bench though, but i think that might not be the best idea.. [19:23] little too strong.. [19:31] http://koti.mbnet.fi/mizcu/bench.txt [19:34] all of those warriors are at koening, most between 300-400 [19:41] and of course, an category "other" could be included, but only could [19:52] list reupdated [19:53] for beginners, i would present benchmarks with and without pspace separated [19:54] Do you really think, that beginners should work on pspacers? [19:54] no [19:54] as you can see, i set "maybe" on it [19:55] no, i did not [19:55] my head is playing games with me again [19:55] maybe a benchmark for pspace-starters would be of interest [19:55] You could at least have pretended to have used invisible ink ;-) [19:55] people like me might be happy about that [19:55] ares: I'd put that way down the list [19:55] just brainstorming [19:57] no comments on the other warrior-selection, or the selection of warrior-types? [19:58] I'm not sure, whether Mice is too easy [19:58] hmm.... i dont know enough about core war strategies, but i can imagine, that benchmarks for single types would be nice, so that i can compare my scanner with other scanners only [19:59] ares: scanners should fight papers to see their performance [19:59] you are the experts *g* [19:59] no, I'm Fluffy [19:59] Mice can be defeated by nearly any offensive scanner-type [20:00] I am not putting Winterwerewolf in the easybench or anything [20:00] Maybe you should use a modified Timescape [20:01] i.e. clean up its sources [20:01] to make it easier to understand [20:01] like the source of Paperone on Koenigstuhl [20:01] same goes for the rest [20:02] its there is not much to clear in timescape, unless you prefer spl @0, 1234 | mov }-1, >-1 - style [20:02] yes, but I've meant comments [20:03] hmm, yes [20:03] you are right about that the bench-warriors should have commenting on them [20:04] i will most certainly write something about each warrior, yes [20:04] you've said to include a benchmark script. then why not benchmark the warriors of the benchmark themselves to make comparing the performance more easy [20:04] Against what? You mean benchmark against the next step (eg wilfiz, if i am to recommend it)? [20:05] no, benchmark each warrior of a benchmark against the benchmark [20:05] one-against all, or MTS-style all-against-all? [20:05] as if would be a hill [20:06] so MTS-like [20:06] I've never used MTS [20:06] the beginner could then (after benchmarking his own warrior) see how good it is in comparison to the warriors of the benchmark [20:07] well, i have the script that metcalf used to create the minitourney-pages [20:08] if you want to see the comparative scores, then might it not be better to use scoretable similiar to wilkies-benchmark? [20:08] http://koti.mbnet.fi/mizcu/stone1.log [20:09] yes [20:09] but a run with -P might be a good idea [20:09] to create the table [20:09] I can churn the results out and make an html-table, no problem at all [20:10] :) [20:12] ill wait until i get an opinion out of someone more experienced before finishing and publishing the benchmark though [20:12] Because some of the warriors still need to be decided [20:12] k [20:28] * Fluffy waves [20:28] MSG: Quit: fluffy.i < 1, # 42 [22:07] Join: Anon1411 joined #corewars [22:07] MSG: Client Quit