2012-10-15T00:12:23 *** thestinger has quit IRC (Read error: Connection reset by peer) 2012-10-15T00:12:29 *** thestinger1 has joined #aichallenge 2012-10-15T00:12:37 *** thestinger1 is now known as thestinger 2012-10-15T00:32:48 *** thestinger has quit IRC (Quit: sleep) 2012-10-15T01:09:50 *** mceier has quit IRC (Quit: leaving) 2012-10-15T02:15:20 *** mceier has joined #aichallenge 2012-10-15T02:40:55 *** replore_ has joined #aichallenge 2012-10-15T02:57:52 *** amstan has quit IRC (Ping timeout: 245 seconds) 2012-10-15T03:27:27 *** epicmonkey has quit IRC (Remote host closed the connection) 2012-10-15T03:29:30 *** pairofdice has joined #aichallenge 2012-10-15T03:41:59 *** mleise has quit IRC (Quit: Leaving.) 2012-10-15T03:49:35 *** replore_ has quit IRC (Ping timeout: 252 seconds) 2012-10-15T04:14:30 *** mviel has joined #aichallenge 2012-10-15T04:21:00 *** mcstar has joined #aichallenge 2012-10-15T04:22:44 *** epicmonkey has joined #aichallenge 2012-10-15T04:33:49 *** epicmonkey has quit IRC (Read error: Connection reset by peer) 2012-10-15T04:36:16 *** epicmonkey has joined #aichallenge 2012-10-15T04:49:51 *** mlp has joined #aichallenge 2012-10-15T04:54:52 *** epicmonkey has quit IRC (Remote host closed the connection) 2012-10-15T04:55:03 *** epicmonkey has joined #aichallenge 2012-10-15T05:15:01 *** coeus has quit IRC (Ping timeout: 248 seconds) 2012-10-15T05:30:22 *** sigh has joined #aichallenge 2012-10-15T05:47:41 *** antimatroidl1 has joined #aichallenge 2012-10-15T05:47:42 *** antimatroidl has quit IRC (Read error: Connection reset by peer) 2012-10-15T07:08:54 *** antimatroidl has joined #aichallenge 2012-10-15T07:08:54 *** antimatroidl1 has quit IRC (Read error: Connection reset by peer) 2012-10-15T07:41:24 *** antimatroidl has quit IRC (Ping timeout: 260 seconds) 2012-10-15T07:49:12 *** alehorst1 has joined #aichallenge 2012-10-15T07:51:09 *** alehorst has quit IRC (Ping timeout: 246 seconds) 2012-10-15T08:07:37 *** alehorst1 has quit IRC (Remote host closed the connection) 2012-10-15T08:12:50 *** antimatroidl has joined #aichallenge 2012-10-15T08:37:41 *** replore has joined #aichallenge 2012-10-15T09:20:45 *** replore has quit IRC (Read error: Connection reset by peer) 2012-10-15T09:21:37 *** replore__ has joined #aichallenge 2012-10-15T09:27:57 *** replore__ has quit IRC (Ping timeout: 246 seconds) 2012-10-15T09:32:31 *** replore has joined #aichallenge 2012-10-15T10:21:35 *** sigh has quit IRC (Remote host closed the connection) 2012-10-15T10:59:02 *** mceier has quit IRC (Quit: leaving) 2012-10-15T11:11:00 *** Scooper has joined #aichallenge 2012-10-15T11:25:10 *** ecidforiap has joined #aichallenge 2012-10-15T11:29:07 *** pairofdice has quit IRC (*.net *.split) 2012-10-15T11:29:07 *** Kakwa has quit IRC (*.net *.split) 2012-10-15T11:30:15 *** keroksi has joined #aichallenge 2012-10-15T11:38:30 *** mleise has joined #aichallenge 2012-10-15T11:40:38 *** Kakwa has joined #aichallenge 2012-10-15T12:03:37 *** alehorst has joined #aichallenge 2012-10-15T12:09:28 *** mceier has joined #aichallenge 2012-10-15T12:11:47 *** keroksi has quit IRC (Quit: Page closed) 2012-10-15T12:23:51 mcstar: The algorithm I run when the stack of known values is empty just slows down the program. It uses 14% of the CPU time 2012-10-15T12:24:19 thats horrible 2012-10-15T12:24:24 youve got to reduce that! 2012-10-15T12:24:57 And turning every function into a template, so I can sutomize it for all of the 81 shapes and and shapes (row/column/block) made things worse, too 2012-10-15T12:25:21 It seems these modern CPUs don't care if something is a runtime value or they have to dereference 2012-10-15T12:25:59 they are hit much more severly by my creating hundreds of instances of the functions, because they don't fit into the cache any more :-/ 2012-10-15T12:27:54 could it be that after all exact-coverage is the one perfect algorithm ? 2012-10-15T12:36:01 *** replore has quit IRC (Remote host closed the connection) 2012-10-15T12:41:20 *** thestinger has joined #aichallenge 2012-10-15T12:54:01 *** mlp has quit IRC (Quit: Page closed) 2012-10-15T12:57:23 lol, you exhausted the code cache? 2012-10-15T12:57:45 the representation for the perfect hitting, is optimal thats for sure 2012-10-15T12:58:00 (well, the compressed boolean matrix formalism, that is) 2012-10-15T13:00:06 *** alehorst has quit IRC (Remote host closed the connection) 2012-10-15T13:05:59 mcstar: yes, it seems so :-( 2012-10-15T13:08:07 I think ther is no separate code/data cache 2012-10-15T13:10:45 *** alehorst has joined #aichallenge 2012-10-15T13:18:19 on my cpu there is 2012-10-15T13:23:33 mleise: L1-Cache: 64 + 64 KB (Data + Instructions), per core 2012-10-15T13:33:23 *** foRei has joined #aichallenge 2012-10-15T13:57:40 *** amstan has joined #aichallenge 2012-10-15T13:57:40 *** ChanServ sets mode: +o amstan 2012-10-15T13:58:11 janzert: hey, remember that d2011 guy who was having trouble posting? 2012-10-15T13:59:08 janzert: he's a spammer! 2012-10-15T14:04:08 *** keezy has joined #aichallenge 2012-10-15T14:04:39 *** Blkt has joined #aichallenge 2012-10-15T14:06:34 Any information when the next ai challenge starts? 2012-10-15T14:06:59 when all the spammers die 2012-10-15T14:08:43 *** epicmonkey has quit IRC (Ping timeout: 245 seconds) 2012-10-15T14:14:48 janzert: ("Upgrade LISP to VirtualC+LISP The programming language VirtualC+Lisp is new upgrade2012 of lisp language embedding math equations in LISP" fyi; i've deleted his posts and deactivated the account already) 2012-10-15T14:15:26 lisp reborn 2012-10-15T14:15:50 (((((((((()))))))))) 2012-10-15T14:15:53 i didn't realize lisp was so popular as to get spammed about it 2012-10-15T14:16:13 it is, in soviet russia 2012-10-15T14:16:20 where lisp writes you... 2012-10-15T14:28:37 *** mattswe has quit IRC (Quit: Connection closed for inactivity) 2012-10-15T14:38:50 *** jacob_strauss has joined #aichallenge 2012-10-15T14:48:13 *** epicmonkey has joined #aichallenge 2012-10-15T14:52:24 *** epicmonkey has quit IRC (Remote host closed the connection) 2012-10-15T14:53:54 *** Garf has joined #aichallenge 2012-10-15T14:55:25 mcstar: oh great, basicall just by removing the templates and unrolled loops I got from 3 seconds to 1 second, which is the speed of c program from kudoku 2012-10-15T14:56:20 why would templates matter? 2012-10-15T14:57:07 thestinger: I wrote a switch case that maps a cell [0 .. 81] to a templated function call 2012-10-15T14:57:20 thought it would speed things up, if I can work with known indices 2012-10-15T14:57:46 but it turned out that in every little function where I removed the template parameters it got faster 2012-10-15T14:57:57 even for a trivial 4 line function 2012-10-15T14:58:19 where i was bypassing a check when a template flag was set 2012-10-15T14:58:25 so you were generating a bunch of specialized code? 2012-10-15T14:58:37 thestinger: exactly, hundreds of functions 2012-10-15T14:59:24 I'll never try that again. But it cleared up a myth for me. 2012-10-15T15:01:18 *** Accoun has quit IRC () 2012-10-15T15:02:17 Zannick: that's quite weird all around 2012-10-15T15:03:08 mleise: if you aren't actually doing something special with the specializations, it's just a bunch of code bloat :P 2012-10-15T15:04:28 janzert: yeah 2012-10-15T15:05:33 thestinger: well, it avoided looking up values and arrays in lookup-tables 2012-10-15T15:05:38 mleise: template metaprogramming aha moment 2012-10-15T15:05:45 and in case of arrays i also unrolled them mostly 2012-10-15T15:06:14 mcstar: yes. it's not the killer performance feature I thought it was, more of a performance killer 2012-10-15T15:06:25 :) 2012-10-15T15:07:19 aynway, now that I am a few ms below Kudoku, I wonder if the exact-coverage algorithm is essentially the same as the 'trivial' algorithm that finds single candidates 2012-10-15T15:07:49 e.g. such a similar runtime performance can only mean they are both doing the same work 2012-10-15T15:16:45 *** Accoun has joined #aichallenge 2012-10-15T15:17:19 i have doubts to what i said yesterday 2012-10-15T15:18:25 anyway, my big sorrow is, that i cant get even to timings of my previous program with the new one, that uses this new representation, but instead of arrays, im using persistent data structures 2012-10-15T15:19:25 http://www.jsoftware.com/jwiki/Essays/Sudoku 2012-10-15T15:19:29 this is a J version 2012-10-15T15:19:40 i cant even make it read the puzzle from stdin, and run it 2012-10-15T15:21:41 some J examples: 2012-10-15T15:21:43 http://www.jsoftware.com/jwiki/Essays 2012-10-15T15:22:11 *** Accoun has quit IRC (Remote host closed the connection) 2012-10-15T15:24:24 J ? 2012-10-15T15:25:42 j 2012-10-15T15:26:05 it is good at manipulating arrays... 2012-10-15T15:26:17 and it is quite functional 2012-10-15T15:26:24 but ill never use it 2012-10-15T15:27:46 diff =: * 0&=@}:@(0&,) 2012-10-15T15:27:51 WAT ?! 2012-10-15T15:28:34 *** Accoun has joined #aichallenge 2012-10-15T15:29:15 you can give compund verbs names 2012-10-15T15:29:23 compoun* 2012-10-15T15:29:25 f 2012-10-15T15:29:30 compound 2012-10-15T15:31:13 *** keezy has quit IRC (Quit: Page closed) 2012-10-15T15:36:01 "We have verified by exhaustive computer search that a 16-clue puzzle does not exist." 2012-10-15T15:36:41 there are ~10^22 valid sudoku puzzles 2012-10-15T15:37:37 i think the bad performance with the persistent maps/hashmaps of mine is because im deleting elements from them 2012-10-15T15:37:43 not adding them, as i recurse 2012-10-15T15:40:08 meh, libstdc++ sucks 2012-10-15T15:40:15 *** jacob_strauss has quit IRC (Remote host closed the connection) 2012-10-15T16:08:31 *** replore_ has joined #aichallenge 2012-10-15T16:09:36 *** dici has joined #aichallenge 2012-10-15T16:12:34 *** FireFly has quit IRC (Excess Flood) 2012-10-15T16:13:49 *** sh4wn has quit IRC (Ping timeout: 246 seconds) 2012-10-15T16:14:01 *** FireFly has joined #aichallenge 2012-10-15T16:16:16 *** sh4wn has joined #aichallenge 2012-10-15T16:23:35 *** replore_ has quit IRC (Ping timeout: 252 seconds) 2012-10-15T16:35:19 *** Blkt has quit IRC (Ping timeout: 255 seconds) 2012-10-15T16:35:21 *** thestinger has quit IRC (Ping timeout: 246 seconds) 2012-10-15T16:45:58 *** sigh has joined #aichallenge 2012-10-15T17:04:04 *** ecidforiap has quit IRC (Quit: In girum imus nocte et consumimur igni.) 2012-10-15T17:21:32 *** foRei has quit IRC (Read error: Connection reset by peer) 2012-10-15T17:23:14 *** dici has quit IRC (Read error: Connection reset by peer) 2012-10-15T17:24:12 *** sigh has quit IRC (Remote host closed the connection) 2012-10-15T17:24:42 *** thestinger has joined #aichallenge 2012-10-15T17:35:56 *** Scooper has quit IRC (Quit: Leaving) 2012-10-15T17:52:42 *** coeus has joined #aichallenge 2012-10-15T17:54:25 *** mcstar has quit IRC (Quit: mcstar) 2012-10-15T18:02:42 *** antimatroidl has quit IRC (Quit: Leaving.) 2012-10-15T18:51:45 *** thestinger has quit IRC (Quit: WeeChat 0.3.9) 2012-10-15T20:59:40 *** Apophis has quit IRC (Ping timeout: 265 seconds) 2012-10-15T21:13:38 *** coeus has quit IRC (Ping timeout: 252 seconds) 2012-10-15T21:15:34 *** Apophis has joined #aichallenge 2012-10-15T21:24:14 *** coeus has joined #aichallenge 2012-10-15T21:35:48 *** Garf has quit IRC (Quit: Make a new plan, Stan!) 2012-10-15T22:15:02 *** antimatroidl has joined #aichallenge 2012-10-15T23:13:50 *** thestinger has joined #aichallenge