00:16:11: -!- Fizmomaster has quit (Quit: ChatZilla 0.9.91.1 [Firefox 37.0.2/20150415140819]).
10:02:08: -!- impomatic_ has joined #corewars.
10:05:00: -!- Doxin has joined #corewars.
10:05:01: -!- Doxin has quit (Changing host).
10:05:01: -!- Doxin has joined #corewars.
10:05:48: -!- impomatic has quit (Ping timeout: 272 seconds).
10:05:49: -!- impomatic_ is now known as impomatic.
10:11:48: -!- inversed has joined #corewars.
10:12:08: <inversed> Hi impomatic
10:12:18: <inversed> I've just uploaded the challenge results
10:14:11: -!- Fizmomaster has joined #corewars.
10:14:36: <Fizmomaster> hi hi
10:15:27: * Fizmomaster says thank you to inversed for the enjoying stone challenge
10:18:58: <inversed> Hi Fizmo
10:21:33: <Fizmomaster> Hi inversed
10:21:45: <impomatic> Hi Inversed, Fizmo :-)
10:21:56: <impomatic> I'm just looking at the challenge results now. :-)
10:21:59: <Fizmomaster> Congratulation impomatic :-)
10:22:16: <impomatic> Thanks, and congratulations Fizmo :-) Joint winners
10:22:37: <impomatic> inversed: did you know http://inversed.ru/Benchmark_Papers.txt gives a 404 error (not sure if you're still working on the page)
10:23:23: <impomatic> Also your .red files attempt to display in a browser as HTML instead of plain text. Might be a mimetype issue.
10:23:55: <inversed> Fixed the benchmark 404
10:24:03: <impomatic> I'll have a close look at all the entries after lunch. I'll be in trouble if I don't go now :-D
10:24:35: <Fizmomaster> Interesting entries. Funny, Creeping Horror and Lapis Impus 3 accidently looks quit similar but scores very different
10:25:00: <inversed> Don't know how to fix the .red display issue though
10:25:19: <inversed> The stones are very different
10:26:01: <impomatic> inversed: you could rename them as .txt? Or I might be able to help later. Are you hosting on a windows server or a linux server. (does it use Web.config or .htaccess?)
10:26:14: <inversed> Fizmo entered a bit late, maybe with more time he could've created some more competition for John
10:26:25: <impomatic> Anyway, lunch or I'll get told off!
10:26:49: <inversed> Ok, I'll rename them to txt for now
10:29:28: <Fizmomaster> Another good side effect of the challenge was that I manage to get onto the tiny hill the first time since a couple of years ;-)
10:30:25: <Fizmomaster> although the warrior wasn't according to the challenge rules, haha
10:32:07: <Fizmomaster> I'll publish it later
10:41:37: <inversed> Just uploaded all entries in a single archive
10:49:00: <Fizmomaster> ok :-)
11:37:58: <impomatic> Time to look through the entries :-)
11:38:19: <impomatic> inversed: http://inversed.ru/CoreWar/T-Stone_2.txt gives a 404 error :-(
11:41:24: <Fizmomaster> http://inversed.ru/CoreWar/T-Stone_2.red
11:41:30: <Fizmomaster> this is still working
11:44:59: <inversed> Fixed
11:45:48: <inversed> Can't get used to case sensitivity in file names
11:46:28: <impomatic> Thanks :-)
11:50:45: <impomatic> Fizmo: is there a similar stone to T-Stone published? (.5c self-bombing into a reverse clear)
11:53:00: <Fizmomaster> As far as I remember it was first published in one of DMs p-warriors
11:53:23: <Fizmomaster> I used it in some of my p-warriors as well
11:53:52: <impomatic> If you change the b-field of the SPL it would create a perfect gate for the endgame.
11:54:00: <impomatic> SPL #0, <-197
11:54:02: <impomatic> DAT <198, >1
11:54:30: <impomatic> (should be DAT <-198,>1)
11:55:54: <Fizmomaster> ups, there seems a bug in my stone, as it should create a perfect gate
11:58:06: <impomatic> Maybe the position of your bomb. It might get decremented and change the value.
11:59:19: <impomatic> I didn't notice, sorry. When I first saw the code I thought it'd be perfect to create a gate, then singled stepped to see if you'd already done it.
11:59:44: <inversed> Should be spl #0, <sHop instead of sHop+2
12:14:54: <impomatic> Wraith bombs alternately with MOV STEP,1 and MOV STEP-1,1
12:15:25: <impomatic> That means half of the bombs expand to the following (like normal TNT bombs)
12:15:29: <impomatic> MOV STEP-1, 1
12:15:30: <impomatic> MOV STEP, 1
12:15:32: <impomatic> DAT ?
12:15:42: <impomatic> But the other half expand to the following
12:15:59: <impomatic> MOV STEP, 1
12:16:00: <impomatic> MOV STEP-1, 1
12:16:02: <impomatic> MOV STEP-1, 1
12:16:03: <impomatic> DAT ?
12:16:47: <impomatic> I would've tried for a .6c version but didn't have time :-(
12:17:20: <inversed> Synergy uses much simplier, but also very efficient bombs, JMP Step
12:17:57: <impomatic> I have a JMP step bomber :-)
12:18:06: <inversed> Like TNT bombs, they kill single-process warriors but do not kill papers
12:18:06: <impomatic> I'll send it to SAL now
12:20:51: <impomatic> 130.4 :-( It actually uses DJN step, #0
12:21:31: <inversed> Nearly made it
12:21:43: <inversed> Try JMP *step, <1
12:29:33: <impomatic> Tried just now. Similar score 130.3.
12:36:59: <inversed> Synergy's main feature is SPL stripe that gives protection against papers
12:38:45: <inversed> BTW I've updated Randy, you may want to give it a try
12:41:31: <impomatic> I downloaded Randy 0.3 a few days ago and will use it for whatever I optimize next :-)
12:42:32: <impomatic> So the jumping processes eventually end up in the SPL stripe?
12:44:10: <impomatic> I've been trying to write a .5c bomber with TNT bombs which self bombs to mutate into a clear. No luck so far :-(
12:48:46: <inversed> Jumping processes may end up in the SPL carpet, but it could take lots of cycles if there are many active processes
12:49:34: <inversed> Self-bombing idea should be feasible
12:50:41: <inversed> Hit with mov step, 1 just before the loop's jmp so it would move an SPL and activate a clear
12:52:22: <impomatic> I was thinking more along the lines of a .5c Carbonite style stone which converts to a DAT clear when it self bombs.
12:54:08: <impomatic> inversed: I'm going to attempt an ASCII version of the results graph for the next issue of Tinywarrior
13:02:59: <inversed> That's going to be difficult
13:03:50: <inversed> Dave Hillis's results are remarkable, a true evolved stones
13:06:35: * impomatic tries to figure out how it works.
13:08:45: <impomatic> I wish pMARS had an '86 mode.
13:12:23: <Fizmomaster> sent a test entry to tiny hill. Curious how it will score
13:13:18: <Fizmomaster> hmm, not as good as I expected against scanner and oneshot :-(
13:21:17: <inversed> I know absolutely nothing about 86
13:21:40: * impomatic wonders if this could be fixed using by self-bombing into a reverse clear like T-Stone http://corewar.co.uk/halloween2014/soulofthedamned.red
13:23:31: <inversed> How is 86 different from 88?
13:24:56: <impomatic> No SLT. All addressing modes allowed (I think). New processes from a SPL go to the front of the queue not the end (I think). ADD and SUB on instructions only affect the b-field (which makes lots of '86 warriors break under '88)
13:27:29: <inversed> I can't understand the logic behind dat #0, #0 / dat $0, $0 initialization in 88
13:28:36: <impomatic> Nor me :-(
13:30:04: <impomatic> The only way to space components nicely is to boot. Or if you only need a small number of DAT $0,$0 put something like MOV <4000, 0 in those locations and execute them (sometimes quicker than booting)
13:30:21: <impomatic> I don't like to boot. It makes the code look messy :-(
13:31:39: <inversed> I also don't like the X hill settings because of the proportions
13:31:57: <inversed> Max processes should be equal to coresize
13:32:37: <inversed> With the current settings, imps are very hard to kill because of relatively small maxprocesses
13:33:06: <Fizmomaster> I guess it was a compromise to the processor power theses days
13:33:34: <inversed> I don't think maxprocesses affect execution speed
13:34:44: <inversed> There is a process queue and you just fetch a new process each instruction. Their total number is irrelevant.
13:37:17: <Fizmomaster> I think there are still enough interesting settings possible with coresize 8000
13:38:36: <Fizmomaster> the max. processes somewhere between 8 and 200 changing the game siginificantly
13:39:46: <Fizmomaster> I also still like the idea of a "self-cleaning" core, where all modifications are resetted after N cycles
13:41:08: <impomatic> The Halloween tournament had maxprocesses 250
13:45:44: <Fizmomaster> I would like a setting where LP-paper and silk-paper are equal strong :-P
13:55:19: <impomatic> I wish the multiwarrior hill had a bit more activity.
13:56:06: <impomatic> I automatically send my 94nop entries o 94m just to see if anything sticks around for a while. (unless I already have too many warriors on 94m)
13:57:30: <inversed> My testing indicates that the "inflection point", the true medium process value, is around 64 processes
14:15:30: <Fizmomaster> an alternative "94nop" with -p 64 might be an interesting option then
14:21:47: <inversed> I plan to add an infinite hill with 64 processes to my rankings
14:22:52: <Fizmomaster> It shouldn't be a big seal to open a new SAL hill with this settings
14:23:04: <Fizmomaster> deal
14:23:12: <inversed> That'd be nice
14:23:33: -!- waknuk has quit (Quit: waknuk).
14:23:56: <Fizmomaster> I need ask bvowk to restore my admin account to SAL, then I could set it up
14:24:40: <impomatic> I've been trying to write a parallel self-vamping vamp / clear but everything I try scores worse then the vamp from Forgotten Lore.
14:25:19: <Fizmomaster> Just a question. What's in your opinion the best hill size?
14:25:24: <Fizmomaster> 20 or 25?
14:25:49: <Fizmomaster> SAL usually uses 25
14:26:04: <impomatic> I've also been trying to make a Sunset-style paper without any bombing (for an example) but I can't get that to work either.
14:26:12: <Fizmomaster> except Nano and the test hills
14:26:21: <inversed> Never thought about it, both seem fine to me
14:26:32: <Fizmomaster> KOTH usually uses 20
14:26:39: <impomatic> I'd stick with 25 to match the others on SAL.
14:27:38: <inversed> I had an Idea about how FL2b might work: 2/3c scanning vamp + SSD clear in parallel
14:28:36: <impomatic> FL2b = Forgotten Lore 2b :-) (there was hill entry called Forgotten Lore 2 just before FL2b)
14:29:48: <impomatic> That's a good idea. With the SSD clear running one process until the vamp has finished?
14:30:11: <inversed> Yep
14:30:51: <inversed> I have a clear-directing scanner with a separate clear in the works
14:31:23: <inversed> It is quite strong against PWS: even if the scan is destroyed, the clear will finish its job
14:33:17: <Fizmomaster> hmm, that reminds me I did something similar as an airbag version
14:34:08: <Fizmomaster> but failed to get something reasonable scoring
14:34:12: <Fizmomaster> http://users.obs.carnegiescience.edu/birk/COREWAR/94/HILL/airscan.red
14:34:36: <inversed> Fizmo: I'll send a fixed, perfect gate version of T-stone 2 to the Koenigstuhl
14:34:52: <Fizmomaster> thanks :-)
14:35:39: <inversed> Although it only gained 0.3 points against my benchmark
14:35:53: <impomatic> FL2b uses DJN.F #1, 1 bombs
14:37:47: <inversed> How do you know?
14:39:26: <impomatic> A probe that waits for a while then seaches a small section of core for DJN.F #1,1. If found, it suicides. If not found it runs a clear.
14:40:07: <impomatic> Also tested the same probe looking for something different and compared the difference in score.
14:42:24: <impomatic> Originally I was probing for 1,1, then tried some variations. Also did probes for various imp steps, bombs from published warriors etc.
14:44:13: <inversed> That's some hardcore reverse engineering
14:46:46: <inversed> I have so many CoreOps issues planned, not sure which one to start with: mirrored imps, evolved warriors, repliscans, LP scanners
14:48:01: <impomatic> My vote is for mirrored imps.
14:50:07: <impomatic> The first mirrored imp was published by Jonathan Roy in May 1992 (5 months before normal imp rings appeared). They didn't get the name mirrored imps until 1996 though when (I think) Paul Kline first used the name.
14:51:19: <impomatic> See superimp 1.0 http://cristal.inria.fr/~doligez/corewar/by-author/XRoy-Jonathan.htm
14:53:33: <inversed> And then they never caught on somewhy
15:47:11: <Fizmomaster> grrrr, almost...
17:28:06: -!- Fizmomaster has quit (Quit: ChatZilla 0.9.91.1 [Firefox 37.0.2/20150415140819]).
17:34:34: <inversed> .йгше
17:34:48: <inversed> i mean, /quit
17:34:52: -!- inversed has quit ().
21:37:18: * bvowk_ pokes impomatic.
21:37:51: * impomatic pokes bvowk_