00:00:14 <pikhq> Mmm, yeah. Apple does very good hardware design.
00:00:26 <pikhq> ehird: There's large fans to the side of the heatsinks.
00:00:45 <ehird> Bit of a restrictive grill there
00:00:48 <ehird> I'd mount them directly on the heatsink
00:01:08 <pikhq> They probably did that for noise reasons.
00:01:22 <ehird> pikhq: Then they're dumb. That obstruction will make the fans whine
00:01:25 <pikhq> By having them to the side of the heatsink like that, they can have a large, low RPM fan.
00:01:28 <ehird> (and indeed, I've heard reports about whining under load)
00:01:36 <ehird> pikhq: Those heatsinks are big enough to fit a 120mm fan
00:01:47 <ehird> Just like the higher-end PC CPU heatsinks
00:02:12 <ehird> But seriously, it's two Nehalem Xeons. Those things are gonna need high RPM fans at load regardless of anything else.
00:02:54 <pikhq> Oh, right. They use *Intel* processors.
00:03:02 <pikhq> You can cook eggs on them if you're not careful.
00:03:04 <ehird> Since 2005, bud. :P
00:03:23 <ehird> But yeah, I'd scrap the grill and mount the fans directly on the heatsink.
00:03:37 <ehird> I wonder how they got the CPU/RAM on another board? I don't think that's in any consumer-purchasable motherboard setup.
00:03:40 <pikhq> Intel's had CPUs you can fry eggs on since about 2001. :p
00:03:55 <pikhq> ehird, that's not hard to do, actually.
00:04:04 <pikhq> First, they design the motherboard.
00:04:16 <ehird> pikhq: Apple use Tyan server-grade motherboards.
00:04:28 <ehird> but they don't make their own
00:04:43 -!- inurinternet has quit (Connection timed out).
00:04:44 <pikhq> It's a matter of having the FSB go over the edge connector is all.
00:05:00 <ehird> (Mac Pros also use server RAM w/ ECC. You need a server board for two Nehalem processors (which also requires you use Xeon, not i7, iirc), but they seem to be doing it willingly too.)
00:05:05 <ehird> pikhq: no performance hit?
00:05:18 <ehird> Oh, oh, really stupid thing about new Mac Pros;
00:05:24 <ehird> Here's the selection of graphics cards:
00:05:26 <pikhq> Why *would* there be a performance hit?
00:05:39 <ehird> [[NVIDIA GeForce GT 120 512MB
00:05:39 <ehird> 2x NVIDIA GeForce GT 120 512MB [Add $150.00]
00:05:40 <ehird> 3x NVIDIA GeForce GT 120 512MB [Add $300.00]
00:05:42 <ehird> 4x NVIDIA GeForce GT 120 512MB [Add $450.00]
00:05:44 <ehird> ATI Radeon HD 4870 512MB [Add $200.00]]]
00:05:59 <ehird> You can choose up to 4 low-end consumer cards, or one low-high-end gamer's card!
00:06:08 <ehird> In a professional's ultra-high-end boutique machine!
00:06:21 <pikhq> It's not like having the electrons go through a connector instead right on the board makes them magically go slower.
00:06:29 <ehird> pikhq: True, but, longer distance :P
00:06:52 <pikhq> And electrons travel a bit short of the speed of light.
00:07:03 <ehird> pikhq: http://static.arstechnica.com/macpro0409/14benchwellht.png Okay, WTF. Yeah, okay, so hyperthreading thrashes the cache. SO WHY IS IT SO GODDAMN FAST!
00:07:39 -!- Judofyr has quit (Remote closed the connection).
00:07:46 <ehird> http://static.arstechnica.com/macpro0409/16power.png ← Huh, I wonder what Apple did to keep the power consumption so low?
00:07:50 <ehird> 378 watts at load is great.
00:07:59 <ehird> (For a dual-high-end processor machine w/ 12GB of RAM)
00:08:01 <pikhq> ehird: Because rendering doesn't use the cache that much?
00:08:10 <ehird> pikhq: Er, whyever not?
00:08:24 <ehird> SourceForge Acquires Ohloh
00:08:25 <ehird> Today SourceForge has acquired Ohloh. We at Ohloh are pretty awed and excited at the opportunity (and challenges) ahead. I plan on blogging more deeply over the next few weeks but I wanted to give you some background on why this makes se... CONTINUE...
00:08:59 <pikhq> ehird: Because you don't exactly access the same memory all that close to the same time in a render?
00:09:02 <ehird> pikhq: Hahahaha. The low-end gfx cards the Mac Pros ship with,
00:09:15 <ehird> is just a rebranded GeForce 9500 GT. Which is the ultra-crappy $40 card I got for bsmntbombgirl's rig.
00:09:24 <ehird> But you can have FOUR of them!111111
00:09:45 <pikhq> Not that Apple charges $150 for each additional card in there.
00:09:56 <ehird> pikhq: Alternatively, you can have a gaming card. Never mind, say, a Nvidia Quadro or Tesla.
00:09:59 <ehird> No, this is totally a gaming machine.
00:10:13 <ehird> Not even a 4890 or an X2. It's just one 4870.
00:10:21 <ehird> Well, anyone knows that Apple is $$$.
00:10:36 <ehird> They just shove the costs for assembly and general design sense and OS onto the other components.
00:11:02 <pikhq> If I'm spending $450 on GPUs, I'm thinking a Tesla would be nice. :p
00:11:08 <ehird> 16GB (8x2GB) [Add $500.00]
00:11:08 <ehird> 32GB (8x4GB) [Add $6,100.00]
00:11:16 <ehird> Bit of a price jump there; wonder where they got 4GB sticks of DDR3?
00:11:21 <ehird> I didn't know they even existed.
00:11:36 <pikhq> (knowing my usage of GPUs, if I'm spending that much, then it's for GPGPU usage)
00:11:48 <pikhq> ehird: Note the price. That's why you don't see them much. ;)
00:12:21 <ehird> *gawp*. The Apple Cinema Displays — you know, really nice things —
00:12:23 <ehird> They now use OLEDs.
00:13:03 <ehird> They're just calilng LCD "LED".
00:13:09 <ehird> Stuuuuuuuuuuuuuuuuuuuuuuuuuuupid
00:13:49 <pikhq> I didn't know that liquid crystals had anything to do with light-emitting diodes.
00:14:07 <pikhq> Dammit, people: these terms mean something other than "LAWLZ IT SHOWS STUFF".
00:15:34 <ehird> [[New LA-based Mac cloner hopes to magically alter status quo
00:15:34 <ehird> Yet another company has decided to toss its hat into the Mac cloning ring. This latest venture hopes to avoid Psystar's fate by opening a retail store to supplement online sales. Can we get a WTF?]]
00:15:41 <ehird> pikhq: How apropos to what you were saying.
00:16:04 <ehird> [["It's exciting. We are trying to stay as close to Apple as we can with our products," Rashantha De Silva, Quo founder, told CNET. "We are trying to mimic things as much as we can. I'm hoping that Apple sees the value in what we are doing."]]
00:16:10 <ehird> For values of "value" equal to "ripping you off".
00:17:18 <myndzi> it's not ripping them off if they're overchanging anyway
00:17:36 <ehird> how does that follow at all
00:17:42 <myndzi> if apple products are overpriced simply because there's no competition
00:17:47 <myndzi> than having competition doesn't = "ripping them off"
00:17:53 <ehird> they're overpriced because they're overpriced
00:17:54 <myndzi> it's a conditional statement
00:17:57 <ehird> myndzi: what they're doing is illegal
00:18:06 <myndzi> what are they doing that is illegal?
00:18:08 <pikhq> Make a very nice EFI-based machine and do the whole wink-wink nudge-nudge thing.
00:18:10 <myndzi> i honestly haven't followed this
00:18:18 <ehird> myndzi: selling computers with os x on them
00:18:22 <ehird> a violation of the OS X EULA
00:19:21 <myndzi> have eulas even been tested to hold up?
00:19:31 <ehird> apple sued the pants off a company doing the same
00:19:41 <ehird> but eulas themselves aren't generally held to work uncontroversially
00:19:48 <ehird> this is fairly unquestionably illegal, though
00:20:00 <ehird> anyway, apple products aren't overpriced because of a lack of competition; they're overpriced because they're boutique
00:20:02 <myndzi> if you can't buy a computer with OS X on it
00:20:06 <myndzi> what are people who buy apple products doing?
00:20:27 <ehird> myndzi: apple are an apple-partnered retailer, and they sell Apple-labeled computers
00:20:39 <ehird> you can install OS X on apple-labeled computers, and you can sell computers with OS X on them
00:20:50 <ehird> they're selling computers that they installed OS X on that aren't apple-labeled
00:20:53 <ehird> therefore they are breaking the law
00:21:10 <myndzi> seems silly anyway but *shrug*
00:21:24 <ehird> myndzi: Apple are a hardware company that only survive because of their software.
00:21:27 <ehird> It's an odd situation.
00:21:43 <ehird> They wouldn't make enough money if they just sold OS X for PCs, but nobody really buys an Apple because of its hardware.
00:21:46 -!- Patashu has joined.
00:29:29 <ehird> "(for example, getting up close to the Mona Lisa reveals the crackled texture of the dried oil paints, not a bunch of little Mona Lisas)"
00:29:53 <ehird> It's Mona Lisas all the way down.
00:31:54 <ehird> http://i42.tinypic.com/302yjdg.png
00:31:59 <ehird> That rightmost compression is amazing.
00:32:02 <coppro> the question is whether that part of the EULA is enforceable
00:32:03 <ehird> Too bad it's a blackbox windows binary.
00:32:27 <ehird> coppro: well, apple don't sell "Mac OS X", they sell "Mac OS X for use on Apple Macintosh computers"
00:32:48 <coppro> but that may be illegal
00:32:50 <ehird> GregorR-L: what's Odikeh's rationale?
00:33:13 <coppro> EULAs are high on the list of contracts that may be viewed as unenforceable for the position they put the consumer it
00:33:30 <GregorR-L> ehird: Odikeh was a conlang, þe Odikeh alphabet was juſt þe alphabet for it.
00:34:16 <ehird> I put English in it, see.
00:35:11 <ehird> What would be fun is a Standard English Abbreviated Alphabet.
00:35:23 <ehird> e.g., "the x" could be x with a slash or something.
00:37:11 <ehird> pikhq: http://imgur.com/0dPo3.png
00:37:56 <GregorR-L> Yeah, Odikeh haþ more letters þan Engliſh, but haþ no upper- vs lower-caſe, ſo a few of þe upper-caſe letters are unuſed.
00:38:12 <ehird> I just typed it in the original cas.
00:38:19 <ehird> i.e., T is the only uppercase.
00:38:27 <GregorR-L> A few of þe lower-caſe letters are unuſed too, mebbe :P
00:38:39 <GregorR-L> It's been at leaſt a year and a half ſince I've looked at it.
00:38:39 <pikhq> GregorR-L: My custom script is an English shorthand. ;)
00:38:54 <GregorR-L> Þy cuſtom FACE is Engliſh ſhorþand
00:39:03 <GregorR-L> Whoops, definitely ſhouldn't be a þorn þere.
00:39:07 <pikhq> If I had a scanner, I'd give some examples.
00:39:16 <ehird> pikhq: Just use the GIMP or something
00:39:27 <pikhq> GregorR-L: Shortḧand?
00:39:41 <coppro> what's with these thorns and such?
00:39:42 <ehird> pikhq: h is not a vowel
00:39:47 <ehird> coppro: GregorR-L wrote a scirpt to do it
00:40:27 <pikhq> Somewhere here, I have a script to start up a nice Xnest for Gimpiness.
00:40:45 <ehird> pikhq: HAY TILING WMS ARE AWSUM
00:40:47 <GregorR-L> .... why do þou need an XNeſt for þe GIMP?
00:41:01 <ehird> GregorR-L: it uses 3 windows
00:41:04 <ehird> and he uses a tiling wm
00:41:13 <ehird> so it fucks up because his WM sucks and forces everything into an unnatural layout.
00:41:23 <ehird> So he needs an xnest to make it seem like one window.
00:41:53 <pikhq> And Xnest is broken.
00:41:59 <GregorR-L> Is þere any caſe of þe word "do" where "doſt" is incorrect?
00:41:59 <pikhq> Off to workspace two, then.
00:42:00 <ehird> xnest, as an side, is sucky; there should be a way to use it without starting a whole new server.
00:42:04 <ehird> It doesn't need that.
00:42:24 <GregorR-L> No, þere is. To doſt þat would be bad. (<-- example)
00:42:55 <ehird> Right, I'm bringing back a letter of my own.
00:43:43 <ehird> But eth doesn't feel very th to me.
00:44:15 <ehird> GregorR-L: Perhaps yoȝ.
00:44:27 <ehird> Ȝes, this could be nice. Albeit rare.
00:44:46 <GregorR-L> I only brought back letters þat had digraph replacements *ſhrugs*
00:45:01 <GregorR-L> OK, þe long-s þing is juſt ſillineſs :P
00:45:12 <ehird> GregorR-L: It's y and gh and stuff.
00:45:32 <ehird> The disadvantage to this is that Y AND GH ARE REALLY UNCOMMON.
00:45:40 <GregorR-L> But þe Ȝ->y was more of a letter evolution þan a linguiſtic change, waſn't it? Þat is, þe letter is þe ſame, it juſt looks different.
00:46:08 <kerlo> ...hey, you said that.
00:46:10 <ehird> GregorR-L: No; the letter is "yogh", which was spelt yoȝ.
00:46:15 <ehird> So it's not *exactly* modern y.
00:46:34 <GregorR-L> kerlo: Yeah, but not every inſtance of "do" can be changed for "doſt" :(
00:46:59 <ehird> dz with a flick, bitch.
00:47:15 <kerlo> Hey, you got the s at the end of a word right.
00:48:28 <kerlo> Though perhaps "ſillineſs" should be written with... that fancy German lowercase SS thing.
00:49:37 <ehird> Esh (majuscule: Ʃ, minuscule: ʃ; Unicode U+01A9, U+0283) is a character used in conjunction with the Latin alphabet, introduced by Isaac Pitman in his 1847 Phonotypic Alphabet to represent the voiceless postalveolar fricative (English sh), and is today used in the International Phonetic Alphabet as well as in the alphabets of some African languages.
00:49:39 <ehird> Its lowercase form ʃ is similar to an italic long s ſ or an integral sign ∫; its uppercase form Ʃ is based on the Greek letter sigma.
00:49:51 <GregorR-L> at's not a dead letter, at's a was-never-alive letter :P
00:49:59 <ehird> GregorR-L: btw, you can use ∫ as an s thing that looks italic.
00:50:21 <ehird> GregorR-L: Ʃut up, anyway; I'll use this ʃit if I like.
00:50:24 <kerlo> So, I think I may drop s2.normish.org. There's a good chance that ehird would encourage me to do so.
00:50:35 <ehird> kerlo: Yer a mind reader :P
00:51:04 <kerlo> Is there some backstory where?
00:51:12 <ehird> eh, maybe I'll just use the schwa. dəmon.
00:51:23 <ehird> Ou (Majuscule: Ȣ, Minuscule: ȣ) is a ligature of the Greek letters ο and υ which was frequently used in Byzantine manuscripts. This ligature is still seen today on icon artwork inside Greek Orthodox churches.
00:51:30 <GregorR-L> kerlo: And ehird being all wompo about it
00:51:44 <kerlo> GregorR-L: ask him; I don't think there's much to explain.
00:52:15 <ehird> GregorR-L: I just thought that the idea of starting a nomic from scratch was the silly. Also he used mysql and PHP and stuff and *vomit*
00:52:31 <GregorR-L> OH, I þought it was ſomeþing about þe domain name itſelf.
00:52:49 <ehird> Turned v (majuscule: Ʌ, minuscule: ʌ) is a letter of the Latin alphabet, based on a turned form of V.
00:52:56 <ehird> I declare it to mean fjkghdfiugherdfl.
00:53:30 <kerlo> "is a letter of the Latin alphabet"? Which Latin alphabet is that?
00:55:46 <kerlo> Hmm, I like the Unicode character LATIN LETTER TWO WITH STROKE.
00:56:05 <pikhq> Okay, I created a rough example via the GIMP.
00:56:09 <kerlo> Because 2 is totally a letter now.
01:01:49 <kerlo> Anyway, I think I'll write my vowels enPR-style.
01:03:25 <kerlo> Just as soon as I manage to install the Alt-Latin keyboard layout.
01:07:04 <kerlo> Th@ kwik brown fox jumps _ov@r th@ l_az_e däg.
01:07:21 <kerlo> Actually, circumflexes work pretty nicely.
01:07:43 <kerlo> Th@ kwik brown fox jumps ôv@r the@ lâzê däg.
01:08:55 <kerlo> Is this môr ôr less @nnôying th@n Greg@rR's schêm?
01:10:06 <kerlo> Yôô didn't êv@n ûz th@ sâm w@rd f@r "has".
01:10:31 <GregorR-L> At leaſt mine is all valid Engliſh (of ſome era)
01:10:40 <ehird> pikhq: PINGGGGGGGGG
01:10:55 <kerlo> Wh@t's invalid @bowt mîn?
01:12:18 <kerlo> If yôô wont, sâ Î'm just wrîting pr@nuncêâsh@nz insted @v spellingz.
01:12:36 <ehird> kerlo: Yours is so annoying that I refuse to read it.
01:12:37 <GregorR-L> Did ou really pronounce 'l' twice ere? ;)
01:13:00 <kerlo> And yes, Î am ûzing cons@n@nts s@mw@t inc@nsist@ntlê.
01:13:24 <GregorR-L> I actually find at quite readable.
01:14:06 <kerlo> Well, how wood Î distingwish b@twên "unâmd" and "unnâmd" if Î didn't pr@z@rv cons@n@nt lengths?
01:15:57 * GregorR-L is trying to decide what "unamed" means.
01:16:37 <kerlo> It mêns "not âmd".
01:17:41 <GregorR-L> e ing is, ou actually pronounce e n twice in un-named.
01:18:01 <GregorR-L> But þou don't pronounce þe l twice in ſpelling (ſpe*ling)
01:18:28 <kerlo> (Shoodn't that bê "pronouncest" or something?)
01:18:37 <kerlo> I guess I don't do it.
01:18:54 <kerlo> Î gess Î dôn't dôô it.
01:19:05 <GregorR-L> Let's not get all ſchwer-pedantic.
01:20:18 <kerlo> C@nect tôô Nôrmish and run @ whôôiz. :-P
01:20:47 <kerlo> quasnaart pts/1 May 17 00:37 (97-84-4-245:S.0)
01:20:48 <kerlo> tetsu pts/7 May 28 16:29 (m7a5e36d0:S.0)
01:20:48 <kerlo> tetsu pts/0 May 29 23:37 (m7a5e36d0:S.1)
01:20:48 <kerlo> tetsu pts/5 May 29 23:39 (m7a5e36d0:S.2)
01:20:48 <kerlo> ihope pts/8 May 30 12:40 (c-98-209-136-197.hsd1.mi.comcast.net)
01:20:50 <kerlo> tetsu pts/9 May 30 02:00 (m7a5e36d0:S.3)
01:20:57 * coppro is irritated by Enigma's last tutorial level
01:21:38 <kerlo> Î w@nd@r whî Î dôn't hav @ :S.0 aft@r mî hôstnâm.
01:26:05 <pikhq> ehird: http://imgur.com/B8SqH.png
01:26:30 <kerlo> Now, to start working on a different project to replace Normish! >:-)
01:26:44 <kerlo> Which I say solely to scare ehird.
01:27:10 <pikhq> Very, *very* rough example.
01:27:16 <pikhq> Bezier curves ahoy.
01:27:18 <kerlo> (Hey, "solely" is another double-consonant word.)
01:36:31 -!- psygnisfive has quit (Read error: 104 (Connection reset by peer)).
01:37:31 -!- psygnisfive has joined.
01:56:11 -!- psygnisfive has quit (Read error: 54 (Connection reset by peer)).
01:56:19 -!- psygnisfive has joined.
01:56:56 <bsmntbombgirl> http://static.arstechnica.com/macpro0409/3neh_internalshr.jpg
01:57:00 -!- pikhq has quit (Read error: 104 (Connection reset by peer)).
02:00:08 -!- FireFly has quit ("Later").
02:01:06 -!- pikhq has joined.
02:34:44 -!- psygnisfive has quit (Read error: 104 (Connection reset by peer)).
02:34:53 -!- psygnisfive has joined.
03:15:04 <bsmntbombgirl> the most expensive i could make a mac pro was $15k :(
03:39:42 -!- Corun has quit ("Leaving...").
03:41:24 <kerlo> ¿El agua hierve? ¿Qué hierve?
03:42:45 <pikhq> bsmntbombgirl: LAME.
03:43:01 <pikhq> I can get a system up beyond that before starting on hard drives.
03:44:02 <pikhq> (granted, with hard drives, the theoretical max is oo. There's RAID chassis that hook up via eSATA. Infinite storage FTW)
03:57:21 -!- inurinternet has joined.
04:03:57 -!- bsmntbombgirl has changed nick to bsmntbombdood.
04:05:31 -!- zzo38 has joined.
04:05:47 <zzo38> Hay! You told me I didn't need a e-mail address to register
04:05:59 <zzo38> But now it says I do
04:06:18 <zzo38> It also says I need to register and IDENTIFY to post message on ##C channel.
04:06:34 <zzo38> But IDENTIFY is not a valid command and I can
04:06:47 <zzo38> 't register without password/email but I don't want either of those things
04:07:24 -!- inurinternet has quit (Read error: 60 (Operation timed out)).
04:07:34 -!- zzo38 has left (?).
04:45:34 -!- KingOfKarlsruhe has joined.
05:45:18 -!- GreaseMonkey has joined.
06:23:25 <Patashu> !bfjoust 2_3weave (>(+)*20>(-)*20)*4(>[(-)*20[+]](-)*4>[(+)*20[-]](+)*4>(+)*4)*10
06:23:36 <EgoBot> Score for Patashu_2_3weave: 80.1
06:24:53 <GregorR-L> Alſo, you're not allowed to frown after getting þe top ſcore.
06:25:28 <Patashu> ya but I might be abusing... ~~:'(~~ HASH BASED RANDOMNESS ~~:'(~~
06:26:55 <pikhq> BTW, "you" would be correct in this case.
06:27:14 <GregorR-L> I þought "þou" was juſt þe polite form of "þou"
06:27:20 <GregorR-L> I thought "thou" was just the polite form of "you"
06:27:29 <GregorR-L> And so both are correct in any case.
06:27:39 <Patashu> anyway it ought to cause maximum discomfort to programs that expect a certain behaviour
06:27:41 <pikhq> "Thou" is the polite second person singular.
06:27:47 <Patashu> because it does not attack every third cell and has mod 3 behaviour
06:27:55 <pikhq> "You" is the impolite second person singular and the second person plural.
06:29:43 <Asztal> thou was the impolite one, I think
06:30:53 <GregorR-L> Note: "In Old Engliſh, generally, þou is þe language of a lord to a ſervant, of an equal to an equal, and expreſses alſo companionſhip, love, permiſsion, defiance, ſcorn, þreatening: whilſt ye is þe language of a ſervant to a lord, and of compliment, and furþer expreſses honor, ſubmiſsion, or entreaty." --Skeat.
06:31:17 <GregorR-L> It's hard to ſpeak a language I don't actually ſpeak :P
06:31:22 <pikhq> Asztal: Ah. Right.
06:33:04 <pikhq> COMMIES ARE PINK AND ÞEY ÞINK!
06:33:53 <GregorR-L> It's juſt not ... modern Engliſh :P
06:34:02 <pikhq> Þis is more Engliſh þan you would ever know.
06:34:16 <pikhq> GregorR-L: It's considered modern Engliſh. Early Modern, to be ſpecific.
06:34:23 <pikhq> s/considered/conſidered.
06:34:37 <GregorR-L> pikhq: In þe ſame way þat cro-magnon is human. Juſt not Anatomically Modern Homo Sapiens Sapiens (AMHSS)
06:34:51 <GregorR-L> pikhq: But y'know what people ſay as ſhort for AMHSS? Modern human.
06:35:22 <pikhq> I ſentence you to 100 years of Chaucer.
06:35:45 <GregorR-L> I ſentence þou to 100 years as a Homo Heidelbergenſis.
06:36:12 -!- psygnisfive has quit (Read error: 104 (Connection reset by peer)).
06:36:14 -!- psygnisf_ has joined.
06:41:03 <pikhq> "Þyself þou þouëst."
06:44:26 <GregorR-L> I þee þy þou, and raiſe þee þe þouſands!!
06:53:55 -!- oerjan has joined.
06:57:02 -!- amca has joined.
07:02:24 -!- myndzi has quit (Remote closed the connection).
07:09:28 -!- kar8nga has joined.
07:11:52 -!- myndzi has joined.
07:24:34 -!- oerjan has quit ("leaving").
07:27:43 <GregorR-L> Pataſhu: Pataſhu_2_3weave doþ in fact do far worſe wiþ different randomization :P
07:27:49 <GregorR-L> Patashu: Patashu_2_3weave does in fact do far worse with different randomization :P
07:35:51 <myndzi> seems to be pretty common
07:36:01 <myndzi> the random thing is gonna have to change if we are to have any chance of optimizing things based on the code itself
07:36:34 <GregorR-L> But if þou give me ten minutes to get þe initial cache, I þink it ſhould be faſt enough for normal runs.
07:36:59 <myndzi> i like that two of my programs are near the top after that change :>
07:37:07 <myndzi> maybe now i can optimize them 'for realz'
07:37:09 <Patashu> else 2/3weave would go dooown
07:37:22 <GregorR-L> What difference ſhould conſtitute a win?
07:37:27 <myndzi> well somebody tell me when to check
07:37:29 <GregorR-L> OK, even I'm getting tired of at :P
07:37:33 <myndzi> GregorR-L: ideally it should be scored on everything
07:37:48 <myndzi> if it were me, i'd give each program X points for a win and Y for a tie
07:37:58 <myndzi> so if they win all matches against another program, that's 21*X
07:38:00 <Patashu> are you checking reverse polarity as well?
07:38:18 -!- jix_ has quit (Read error: 60 (Operation timed out)).
07:38:27 <myndzi> maybe divide it by something to keep them from getting out of hand, i dunno how the results would be with just that simple scoring
07:38:46 <GregorR-L> Well, it would be from -20 to 20 for each run.
07:38:50 <myndzi> but if you have the info from 21 battles, it seems kind of arbitrary to "reduce" that information to just win or loss
07:39:03 <Patashu> you'd reduce it to a fractional win/loss I presume
07:39:17 <Patashu> 20 wins and 1 loss would add 19/21
07:39:35 <myndzi> i don't really know why? the most wins is the most desirable
07:39:46 <Patashu> instead of + - and 0 though
07:39:56 <Patashu> you'd need some set of 1 or 2 symbols indicating what amount was won/lost
07:39:57 <myndzi> oh, yeah. i don't know what you will do there
07:40:02 <GregorR-L> More accurately, that's a major change to the system ^^
07:40:08 <myndzi> http://sal.math.ualberta.ca/hill.php?key=94b
07:40:13 <myndzi> ^ an example of a pretty decent way to do it
07:40:21 <myndzi> what, running against all lengths ISN'T a major change? :)
07:40:51 <Patashu> ah, that doesn't display it as a grid
07:40:57 -!- jix has joined.
07:41:09 <myndzi> i thought you meant results wise
07:41:22 <myndzi> Patashu: click the name of one warrior
07:41:22 <GregorR-L> myndzi: There are two components, the interpreter and the reporter. The only interaction between them is execl.
07:41:39 <GregorR-L> I kind of don't want to change that, since I'd like the interpreter to be usable independently ...
07:42:00 <myndzi> GregorR-L: have the interpreter return a score
07:42:13 <myndzi> wins-for-1st-program wins-for-2nd-program ties
07:42:26 <GregorR-L> myndzi: return ohyeahthat'srightthisisaninteger;
07:43:03 <myndzi> technically you COULD do it like...
07:43:14 <myndzi> ties + losses * 32 + wins * 32^2
07:43:35 <myndzi> or for the lazy ties & losses << 5 & wins << 10
07:43:39 <myndzi> or something like that
07:43:53 <myndzi> or have the reporter run the interpreter 21 times
07:43:59 <myndzi> instead of running the multiple runs in the interpreter
07:44:09 <myndzi> then you can get individual results using your precious return ;)
07:44:59 <GregorR-L> Somebody give me a quick lesson on associative arrays in Python :P
07:45:23 <myndzi> Otyg - Holy Diver (03:51)
07:45:26 <GregorR-L> I'm not really that much of a Pythoner, but I've decided to be.
07:45:27 <myndzi> haha this sounds amusing
07:45:32 <myndzi> i like the dio version much better
07:45:50 <Patashu> http://docs.python.org/library/stdtypes.html#mapping-types-dict not good enough for you?
07:46:08 <myndzi> how do i get hg to update a cloned directory?
07:46:13 <myndzi> hg update doesn't seem to do it
07:47:14 <GregorR-L> hg up updates from your /working directory/ from the /local repository/, hg pull updates your /local repository/ from the /remote repository/.
07:47:47 <myndzi> i just realized when i rebooted i lost my query scrollback to EgoBot
07:47:52 <myndzi> and thus all the code i was playing with
07:48:45 <myndzi> i don't know if you heard it, but i guess the package in ubuntu was old, that's why i had problems
07:48:51 <myndzi> i installed the windows one and it worked great
07:49:03 <myndzi> took me quite some time to figure out how to display the most recent version of a deleted file though
07:51:58 <GregorR-L> That sort of stuff, not getting it installed.
07:52:19 <myndzi> yeah, i just was saying it 'cause i think you went off before i got it working
07:58:20 <GregorR-L> OK, I just set it up to use polarity too.
07:58:44 <myndzi> so far it hasn't updated for me, is it still (re?) running?
07:59:35 <GregorR-L> 400 runs as in 20 programs vs 20 programs.
07:59:46 <GregorR-L> (Which is a huge overestimate, actually, but eh :P )
07:59:59 -!- clog has quit (ended).
07:59:59 -!- clog has quit (ended).
08:00:00 -!- clog has joined.
08:00:00 -!- clog has joined.
08:00:00 <GregorR-L> Hey look, slowrush is back on top :P
08:00:18 <myndzi> i still get the old report though
08:01:00 <Patashu> you'd only need 210 matcups
08:01:32 <Patashu> would be the number of ways to order 20 elements
08:01:37 <Patashu> and is also a VERY big number
08:01:41 <Patashu> what you want is 1+2+3+...20
08:01:56 <myndzi> i forget, but there's a simple formula for it
08:02:13 <GregorR-L> !bfjoust pooper_scooper (>(-)*32>(+)*32)*5[>[-].+]
08:02:21 <GregorR-L> That will take a long while, because it's running a full uncached run.
08:02:23 <myndzi> oh now the report blanks
08:02:38 <myndzi> i figure this'll show us the real report
08:02:52 <myndzi> 20^2 + 20 doesn't sound right
08:02:57 <myndzi> the squared part, i mean
08:03:11 <myndzi> right, but i remember vaguely this formula and that doesn't sound correct
08:03:17 <Patashu> try and plug in the numbers
08:03:50 <myndzi> but i could have sworn there was a -1 and some shit
08:03:50 <Patashu> hmm so if entering the hill is going to be expensive from now on
08:03:57 <myndzi> maybe there's another formula for the same thing
08:04:07 <GregorR-L> Patashu: It's not that bad, it's just that this is an uncached run.
08:04:09 <myndzi> Patashu: not this bad, once it caches all the matchups
08:04:11 <GregorR-L> It's still on the order of a few seconds.
08:04:17 <Patashu> could you add another command that doesn't enter a program on the hill but instead makes it do a match against a program on the hill with your program and return the result
08:04:29 <Patashu> EVERY matchup just for the first time
08:04:40 <Patashu> then after that you only need 20
08:04:55 -!- whtspc has joined.
08:05:14 <myndzi> that's the formula i knew
08:05:18 <myndzi> which is of course n^2+n
08:05:22 <myndzi> but that's why it looked weird to me
08:05:39 <Patashu> why does it need to do a match of x against y then y against x :)
08:05:43 <EgoBot> Score for GregorR-L_pooper_scooper: 28.0
08:05:57 <GregorR-L> Patashu: It doesn't, but the framework is from and old system where it does, and it's threaded so it's hard to fix that >_>
08:06:18 <myndzi> 16 | + + + 0 + - + + + + 0 + + + + + + + + | 85.7 | 15 | myndzi_slowrush.bfjoust
08:06:36 <GregorR-L> 2 | - - 0 + + 0 - - - - - - - + + - + - - | 28.0 | -7 | GregorR-L_pooper_scooper.bfjoust
08:06:38 <myndzi> i still never managed to top 100 points
08:06:43 <Patashu> 2/3 weave is surprisingly good still
08:07:07 <Patashu> 3 | + + 0 + + + 0 - - + + + + + + 0 + 0 + | 71.3 | 11 | Patashu_2_3weave.bfjoust
08:07:10 <myndzi> and 3pass is still up there too, i am moderately happy about that
08:07:23 <myndzi> i don't want to work on optimizing them now but i'm glad i'll be able to try and tweak them better
08:07:34 <Patashu> I wonder if a 3/4weave would be worthwhile
08:08:23 <Patashu> (>(+)*20>(-)*20)*4(>[(-)*20[+]](-)*4>[(+)*20[-]](+)*4>(+)*4)*10
08:08:35 <myndzi> i'm surprised it did that well, i guess the speed gain is worth it
08:08:47 <Patashu> -(step decrement step decrement step) loop
08:08:57 <myndzi> but, perhaps you can get the same score simpler by simply skipping those lines at the beginning
08:09:00 <myndzi> except i guess vs defend
08:09:35 <myndzi> i'm assuming that you're going to skip the flag X times
08:09:41 <myndzi> where X i haven't quite determined
08:09:48 <Patashu> !bfjoust 2_3weave >>>>>>>>(>[(-)*20[+]](-)*4>[(+)*20[-]](+)*4>(+)*4)*10
08:09:49 <myndzi> and that if you put those skips at the start it will score similar
08:09:51 <Patashu> this is a decoyless version
08:09:59 <EgoBot> Score for Patashu_2_3weave: 51.5
08:10:05 <Patashu> !bfjoust 2_3weave (>(+)*20>(-)*20)*4(>[(-)*20[+]](-)*4>[(+)*20[-]](+)*4>(+)*4)*10
08:10:06 <myndzi> 1) skipping will overstep the tripwires in defend
08:10:18 <EgoBot> Score for Patashu_2_3weave: 71.3
08:10:27 <myndzi> 2) since you don't skip them all at the start, hrm... maybe 2 is wrong
08:10:34 <myndzi> should have looked at what you lost points to
08:10:35 <Patashu> it will foil any setup that relies on multiple tripwires
08:10:43 <Patashu> and half the time it's going to be able to win it will foil a 1-tripwire setup
08:10:45 <myndzi> possibly, unless they are spaced right
08:10:54 <myndzi> 1/3 of the time it'll foil a tripwire
08:11:17 <myndzi> indeed, it is interesting
08:11:24 <myndzi> also note how fast your submissions were
08:11:29 <Patashu> note to self: fiddle with everything else now that I know there's no HASH BASED RANDOMNESS.
08:11:45 <myndzi> i felt bad for fiddling after i learned that
08:11:53 <myndzi> i had thought i was doing something like adjusting the timing
08:12:08 <GregorR-L> !bfjoust weirdo >((+>)*10(+<)*10)*128
08:12:24 <myndzi> one side effect of doing the scoring this way
08:12:29 <EgoBot> Score for GregorR-L_weirdo: 0.0
08:12:31 <myndzi> is that you only have to win 51% to get points vs a program
08:12:44 <myndzi> which means you can effectively skip a large part of the tape for free
08:13:05 <Patashu> I thought you'd get more points for getting more wins vs losses against a specific program
08:13:13 <Patashu> so like 21/21 wins against a prog would help more than 11/21
08:13:22 <myndzi> Patashu: he hasn't done something like that yet
08:13:31 <myndzi> he just arbitrarily decides based on the results if it's a win or a loss or a tie
08:13:37 <myndzi> that's what i was talking about earlier
08:14:13 <Patashu> do the >> < >> < >> < thing
08:14:25 <myndzi> i've considered that multiple times
08:14:28 <myndzi> but it's just a great way to run off the tape
08:14:47 <Patashu> how about > decrement > decrement << > decrement > decrement <<...
08:14:58 -!- whtspc has quit ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]").
08:15:28 <myndzi> !bfjoust oddly (>)*18(>(-)*20[-[+]])*15
08:15:44 <EgoBot> Score for myndzi_oddly: 4.3
08:16:10 <myndzi> !bfjoust oddly (>(-)*19>(+)*19)*9(>(-)*20[-[+]])*15
08:16:10 <EgoBot> Score for GregorR-L_AHHHMOTHERLAND: 0.0
08:16:23 <EgoBot> Score for myndzi_oddly: 48.2
08:16:40 -!- whtspc has joined.
08:16:46 <myndzi> !bfjoust oddly (>(-)*19>(+)*19)*3(>)*12(>(-)*20[-[+]])*15
08:16:59 <EgoBot> Score for myndzi_oddly: 35.0
08:17:12 <myndzi> !bfjoust oddly (>(-)*19>(+)*19)*5(>(-)*20[-[+]])*15
08:17:29 <EgoBot> Score for myndzi_oddly: 54.7
08:17:34 <myndzi> !bfjoust oddly (>(-)*19>(+)*19)*6(>(-)*20[-[+]])*15
08:17:50 <EgoBot> Score for myndzi_oddly: 57.1
08:18:13 <myndzi> !bfjoust oddly (>(-)*19>(+)*19)*4(>(-)*20[-[+]])*30
08:18:35 <EgoBot> Score for myndzi_oddly: 60.4
08:18:45 <myndzi> !bfjoust oddly (>(-)*19>(+)*19)*3(>(-)*20[-[+]])*30
08:19:09 <EgoBot> Score for myndzi_oddly: 44.5
08:20:03 <Asztal> !bfjoust charge [>]>([-].[+].[-]..[-].[+])*100
08:20:34 <EgoBot> Score for Asztal_charge: 9.7
08:21:28 -!- whtspc has quit ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]").
08:22:11 <EgoBot> Use: !bfjoust <program name> <program>
08:23:04 <EgoBot> Score for KingOfKarlsruhe_test: 8.5
08:23:04 <EgoBot> Score for KingOfKarlsruhe_test: 8.5
08:23:06 <Asztal> !bfjoust charge [>](>([-].[+].[-]..[-].[+]))*21
08:23:18 <EgoBot> Score for Asztal_charge: 0.0
08:24:14 <EgoBot> Score for KingOfKarlsruhe_test: 12.5
08:24:38 <Asztal> !bfjoust charge [>](([+].[-].[+]..[-].[-])>)*21
08:25:04 <EgoBot> Score for Asztal_charge: 0.0
08:25:05 <Asztal> it's supposed to kill the flag while most programs are still setting up the first defence cell
08:26:41 <Asztal> there's no way to actually do that, is there?
08:27:29 <myndzi> whoops i did something silly
08:27:38 <myndzi> i thought i'd do a rear attack
08:27:49 <Asztal> !bfjoust charge (>[+])*50
08:27:53 <myndzi> but of course, tapelength 20 doesn't mean flag @ 20, tapelength 30
08:28:02 <EgoBot> Score for Asztal_charge: 19.0
08:28:06 <Asztal> that goes off the end every time, surely
08:28:34 <EgoBot> Score for KingOfKarlsruhe_test: 8.5
08:28:57 <Asztal> !bfjoust charge (>[+].>[-].)*50
08:29:11 <EgoBot> Score for Asztal_charge: 28.5
08:30:53 <GregorR-L> Why is it that every time I make a (much requested) change to the hill, everybody immediately tries to break it? X-P
08:31:34 <GregorR-L> Well, "take advantage" is more accurate.
08:31:41 <myndzi> that's the nature of the game
08:31:57 <myndzi> if i don't do it someone else will :>
08:31:58 <GregorR-L> You should be taking advantage of the other code, not the scoring mechanism :P
08:32:07 <Asztal> !bfjoust charge >->+>->+>+>->+>+(>[+].>[-].)*50
08:32:09 <myndzi> i just wanted to see if it would work
08:32:14 <myndzi> i was mixed up in my head
08:32:21 <GregorR-L> Well, at least the scoring mechanism is basically fair now.
08:32:22 <EgoBot> Score for Asztal_charge: 66.2
08:32:56 <myndzi> i appreciate your patience with our griping and your effort too
08:33:06 <EgoBot> Score for GregorR-L_reverse_charge: 0.0
08:36:22 <myndzi> i own the hill again :>
08:40:19 <EgoBot> Score for KingOfKarlsruhe_abc: 13.4
08:44:25 <Asztal> how do you get a 0 on the hill now? do you need an equal number of wins and losses?
08:46:40 <EgoBot> Score for KingOfKarlsruhe_abc2: 19.7
08:47:09 <myndzi> heh i made keke2 a little better and might have done more, but primarily it just did better against my own warriors :P
08:47:25 <GregorR-L> OK, I implemented the more-points-for-more-wins mechanism.
08:47:40 <myndzi> maybe i'm about to lose the lead :)
08:47:41 <GregorR-L> I think I effed up the range of scores somehow :P
08:47:42 <EgoBot> Score for KingOfKarlsruhe_abc3: 8.2
08:47:53 <GregorR-L> No, your score is now OVER FOUR THOUSAAAAAAAAAAAAND
08:48:47 <myndzi> gregor i just wanted you to know that i love you. in a totally not-gay way, of course.
08:50:22 <GregorR-L> I'm certainly not updating it in EgoBot until it works X-P
08:50:29 <GregorR-L> Yeah, all the scores are much lower now.
08:50:51 <GregorR-L> A score of 100 would mean that you won /every/ round, in /every/ configuration.
08:51:02 <myndzi> oh, so it's based on percentage?
08:51:24 <myndzi> am i still winnning? :D
08:51:55 <EgoBot> Score for KingOfKarlsruhe_loop: 12.3
08:52:00 <Patashu> !bfjoust juggernaut +(>(-)*128.(-.)*512>(+)*128.(++.)*512)*15
08:52:14 <EgoBot> Score for Patashu_juggernaut: 16.0
08:53:11 <EgoBot> Score for KingOfKarlsruhe_loop: 12.3
08:53:37 <Patashu> !bfjoust juggernaut +(>(-)*128.-.--.---)*29
08:53:49 <EgoBot> Score for Patashu_juggernaut: 21.4
08:54:02 <Patashu> !bfjoust juggernaut +>>>>>>>>(>(-)*128.-.--.---)*21
08:54:13 <EgoBot> Score for Patashu_juggernaut: 20.5
08:54:24 <Patashu> !bfjoust juggernaut +>--->--->>>>>>(>(-)*128.-.--.---)*21
08:54:34 <EgoBot> Score for Patashu_juggernaut: 21.3
08:54:38 <Patashu> !bfjoust waiter ((++-)*10000-(--+)*10000)*2(+)*10000(-)*10000(+-)*10000(-+)*10000
08:55:10 <myndzi> aw don't make me add all the blink defenses back in
08:55:12 <EgoBot> Score for Patashu_waiter: 33.3
08:55:21 <Patashu> !bfjoust lazy >(+)*5>(-)*5>(+)*5>(-)*5>(-)*5>(+)*5>(+)*5>(-)*5(>(+.)*129)*21[-]((-)*2048(+)*2048.)*2
08:55:31 <Patashu> don't think juggernaut's gonna work
08:55:33 <EgoBot> Score for Patashu_lazy: 20.4
08:55:37 <myndzi> guess it doesn't matter, ALL my warriors beat waiter
08:55:41 <Patashu> !bfjoust lazy >+>->++>-->--->+++>+>->((-.)*128>)*21
08:55:53 <EgoBot> Score for Patashu_lazy: 11.9
08:56:57 <myndzi> time to see the new results
08:56:58 <GregorR-L> Now I have to kill both, 'cuz that'll be effy.
08:57:50 <GregorR-L> OK, now you'll have to wait again.
08:58:20 <Patashu> is it possible to make a program that does worse than <
08:58:28 <Patashu> by causing a turn 0 syntax error?
08:58:50 <GregorR-L> Yeah, it detects some syntax errors.
08:59:00 <GregorR-L> And that gives FULL points to the opponent.
08:59:23 <Patashu> only if the syntax error is encountered?
08:59:26 <myndzi> points for invalid code wasn't so cool
08:59:54 <GregorR-L> You'd have to read my increasingly-horrifying code to get all the possibilities :P
09:00:05 <EgoBot> Score for Patashu_stall: 14.3
09:00:05 <EgoBot> Score for GregorR-L_reverse_charge: 0.0
09:00:20 <Patashu> !bfjoust stall (+-)*100000
09:00:27 <Asztal> oh wow, that's a very different hill now
09:00:37 <EgoBot> Score for Patashu_stall: 6.5
09:00:58 <GregorR-L> The points are now floating-point :P
09:01:32 <myndzi> interesting that defend7 does better than 9 now
09:02:07 <Patashu> !bfjoust stall ((+)*1001-)*100
09:02:41 <EgoBot> Score for Patashu_stall: 14.3
09:02:52 <Patashu> so what does a +, - or 0 indicate on the grid now?
09:03:06 <GregorR-L> Patashu: It's a 0 iff every configuration ties.
09:03:36 <Patashu> what if it's 10 +s, 10 -s and a tie
09:03:56 <GregorR-L> More accurately, the number of wins equals the number of losses.
09:03:57 <EgoBot> Score for KingOfKarlsruhe_loop: 5.6
09:04:16 <Patashu> so it's just indicating more wins than losses or vice versa
09:04:20 <Patashu> could there be a degree of gradation to it?
09:04:36 -!- KingOfKarlsruhe has quit (Remote closed the connection).
09:04:47 <GregorR-L> Patashu: YOU PEOPLE ARE NEVER HAPPY
09:04:52 <Asztal> !bfjoust charge >->+>->+>+>->+>+(>[+].>[-].)*11
09:05:00 <EgoBot> Score for Asztal_charge: 18.7
09:05:39 <Patashu> bouncing permutations of rushpolarity off egobot
09:06:15 <Patashu> 2/3 weave now has positive points but low score
09:06:19 <GregorR-L> I kinda don't want to see rushpolarity go :P
09:07:00 <myndzi> that doesn't quite count ;p
09:07:10 <myndzi> nobody's looking there! hehe
09:07:34 <GregorR-L> Is somebody running them via query?
09:07:46 <Patashu> want me to stop for a mom?
09:07:52 <GregorR-L> Ah, that explains why report.txt keeps disappearing :P
09:08:05 <Patashu> trying permutations of rushpolarity
09:08:12 <myndzi> yeah, here's another one of those complaints: why clear it before you're ready to write to it? :P
09:10:43 <GregorR-L> myndzi: ../report ../egojoust ../cache *.bfjoust > report.txt
09:11:37 <myndzi> mk, i guess that's just what happens eh
09:13:05 <Patashu> it's so nice to see improvements come gradually
09:13:09 <Patashu> rather than in bumps and jerks
09:15:54 <GregorR-L> OK, I'm done hacking at EgoBot for the moment.
09:19:02 <Patashu> okay, after searching the local space I've determined this is better
09:19:04 <Patashu> !bfjoust rushpolarity >((+)*20>(-)*19>(-)*19>(+)*19)*2>([(+)*19[-]](+)*21>[(-)*19[+]](-)*21>)*11
09:19:12 <EgoBot> Score for Patashu_rushpolarity: 53.5
09:20:12 <Patashu> next time on #esoteric, Patashu tweaks 2/3 weave
09:20:50 <Patashu> ON THE NEXT EPISODE OF DRAGON BALL Z...
09:21:40 <Asztal> I guess when mine loses, it *really* loses :(
09:21:49 <Patashu> yeah that's the weird thing now
09:21:54 -!- GreaseMonkey has quit ("YES -> thor-ainor.it <- THIS IS *DELICIOUS*!").
09:22:11 <Patashu> you can get lots of shallow losses or a few real bad ones
09:47:48 <myndzi> lol did we end up with practically the same program?
09:53:18 <myndzi> 16 69.88 12.60 myndzi_slowrush.bfjoust
10:28:18 <Deewiant> How does the scoring work now?
10:41:08 <GregorR-L> http://codu.org/eso/bfjoust/SCORES
10:44:43 -!- Judofyr has joined.
10:50:38 <GregorR-L> Heh, I just momentarily confused myself with the 0-100 thing.
10:50:54 <GregorR-L> Then I remembered that since you can't win against something which itself won against everything, you can never gain the >1 worth.
11:16:02 -!- kar8nga has quit (Read error: 54 (Connection reset by peer)).
11:31:33 <Patashu> rushpolarity lost a whole bunch?
11:31:39 <Patashu> was it beating slowrush before and now losing to it?
11:37:33 -!- amca has quit ("Farewell").
11:51:34 <Patashu> now it's getting 0 for everything o.O
11:51:43 <Patashu> !bfjoust 2_3weave (>(+)*23>(-)*24)*2>>>>(>[(-)*20[+]](-)*5>[(+)*21[-]](+)*5>(+)*5)
11:51:48 <EgoBot> Score for Patashu_2_3weave: 0.0
11:51:53 <Patashu> !bfjoust 2_3weave (>(+)*23>(-)*24)*2(>[(-)*20[+]](-)*5>[(+)*21[-]](+)*5>(+)*5)
11:51:59 <EgoBot> Score for Patashu_2_3weave: 0.0
11:52:02 <Patashu> !bfjoust 2_3weave (>(+)*23>(-)*24)*2>>>>(>[(-)*20[+]](-)*5>[(+)*21[-]](+)*5>(+)*5)
11:52:07 <EgoBot> Score for Patashu_2_3weave: 0.0
11:52:35 <Patashu> !bfjoust 2_3weave (>(+)*20>(-)*20)*4(>[(-)*20[+]](-)*10>[(+)*20[-]](+)*10>(+)*10)*10
11:52:42 <EgoBot> Score for Patashu_2_3weave: 14.9
11:53:54 <Patashu> brought to you by epicfailblog dot com
11:55:28 <Patashu> seems to be hovering at 20 now
11:56:55 <Patashu> !bfjoust 2_3weave (>(+)*23>(-)*24)*2>>>>(>[(-)*20[+]](-)*6>[(+)*21[-]](+)*6>(+)*6)*10
11:57:02 <EgoBot> Score for Patashu_2_3weave: 20.1
11:58:24 <GregorR-L> ... wasn't it doing better before? :P
11:58:39 <Patashu> before you changed it to return how many times it won
11:58:48 <Patashu> is get lots of shallow wins
11:58:59 <Patashu> as opposed to programs which get deep wins against a few programs they counter
11:59:23 <Patashu> so it looks good on the grid but has a shitty score
12:03:44 <Patashu> that I want to use two 'repeats' for attacking; one that leaves decoys after it moves but later one that doesn't
12:03:53 <Patashu> because if the tape's that long it's probably long passed its opponent
12:03:56 <Patashu> that'll make it attack faster
12:05:54 <Patashu> odd. it doesn't appear to help at all (for rushpolarity)
12:07:51 <Patashu> okay, rushpolarity is now at score 41.45 points 4.76, 2/3weave is now at score 19.16 points 2.67
12:08:42 <Patashu> stall is a worse version of waiter so I might kill it
12:09:12 <EgoBot> Score for Patashu_stall: 0.0
12:10:28 <Patashu> !bfjoust the_greatest_program_ever!!! +[>[-[+]]---]
12:10:43 <EgoBot> Score for Patashu_the_greatest_program_ever___: 15.2
12:17:13 <Patashu> defend6 is by far the oldest program (since last revision) on the hill
12:17:47 <GregorR-L> !bfjoust parisian_monastery (>->+)*5[>[-[+]]+]
12:17:56 <EgoBot> Score for GregorR-L_parisian_monastery: 28.2
12:17:59 -!- sebbu2 has joined.
12:18:05 <GregorR-L> It might be that I just plain suck at BF Joust :P
12:18:16 <GregorR-L> But nobody'll play FYB, so what else am I to do? X-P
12:18:21 <Patashu> 8th on the hill yeah you're terribad
12:19:26 <Patashu> do I need to enclose my posts in XML tags indicating the tone to be read in?
12:19:58 <GregorR-L> I guess logicex-2 has a four-year streak as king of the hill.
12:20:39 <GregorR-L> !bfjoust parisian_monastery (>->+)*5[>[(-)*5[+]]+]
12:20:46 <EgoBot> Score for GregorR-L_parisian_monastery: 30.8
12:20:56 <Patashu> think about whether you want lots of little decoys or fewer but bigger ones
12:21:13 <GregorR-L> Well, I'm thinking more about actually sinking the enemy flag :P
12:21:20 <GregorR-L> Which is perhaps not the real goal.
12:21:33 <Patashu> step 2. do it before they do it to you
12:22:26 <Patashu> there only seem to be so many variations on the basic program though
12:22:52 <GregorR-L> !bfjoust parisian_monastery (>->+)*5[(>)*2[(>)*2[+]]+(>)*2[+]]-]
12:22:58 <EgoBot> Score for GregorR-L_parisian_monastery: 0.0
12:23:28 <Patashu> can we get multithreading in bf joust?
12:23:57 <GregorR-L> !bfjoust parisian_monastery (>->+)*5[(>)*2[(-)*5[+]]+(>)*2[(-)*5[+]]-]
12:24:05 <EgoBot> Score for GregorR-L_parisian_monastery: 0.0
12:24:22 <Patashu> (>->+)*5[(>)*2[(>)*2[+]]+[(>)*2[+]]-]
12:26:55 <Patashu> what do you imagine the bombs in FYB looking like? this is an important question
12:27:26 <Patashu> ooh deeewiant's adding a new program
12:27:55 <Patashu> they're all slight tweaks on the
12:28:17 <Patashu> 2. use a loop like [(-)*n[+]]
12:28:26 <Patashu> 3. optionally set decoys while attacking
12:28:53 <Deewiant> Oh, is that what slowrush's success is based on? :-P
12:29:03 <Deewiant> Funny, I tried it pretty much by accident
12:29:32 <Patashu> slowrush looks more or less like rushpolarity now
12:30:29 <GregorR-L> Patashu: Most programs, since FYB has threads, employ multiple styles. My favorite style was logicex, or Logic Extermination, which looked for all logic in the program (loops, etc) and set bombs there, in the hopes that the logic would increase the chances of those spots being hit.
12:31:28 <GregorR-L> Simpler programs just looked for the enemy program counter and set a bomb in hopes that it would go there again, or focused more on defending ones one code than attack.
12:32:11 <Patashu> FYB cell arithmatic is mod 17?
12:32:38 -!- sebbu has quit (Read error: 110 (Connection timed out)).
12:32:40 <Patashu> only powers of 2 are holy cretin
12:32:58 <GregorR-L> The reason for 17 was solely that that's how many commands I had :P
12:33:05 <Patashu> . and , are both nops already
12:33:45 <GregorR-L> I had some rationale maybe, or maybe it was just to make 17 instead of 16 to bother you.
12:34:00 <Patashu> it was in case you decided to make . output values to your opponent later
12:34:12 <Patashu> if you turn one half of a loop into a bomb in FYB
12:34:17 <Patashu> where does the other half send its pointer?
12:34:22 <Patashu> same place or is it ineffectual?
12:34:30 <Deewiant> Alright, monorail is #2, I'm happy with that
12:34:35 <GregorR-L> If there's no match, it's just like a nop.
12:34:47 <GregorR-L> Which is why you have to place a bomb just /after/ the loop :P
12:35:04 <Patashu> so a pointer in a :; loop can escape its loop if you change one half of it
12:35:07 <Deewiant> It's also the third-smallest program on the hill (unexpanded) :-P
12:36:39 <GregorR-L> Deewiant: 46.33 is pretty darn good in the new scoring system.
12:37:12 <Deewiant> But I guess I have the best score/size ratio
12:37:30 <GregorR-L> It's not the size that matters, it's what you do with it.
12:42:04 -!- FireFly has joined.
12:46:55 <Patashu> op-code idea for FYB: swap program buffers but not code buffers :P
12:52:58 -!- dbc has joined.
13:05:28 -!- sebbu has joined.
13:07:55 -!- sebbu2 has quit (Read error: 60 (Operation timed out)).
13:24:56 <Patashu> 2/3 weave is now at 18.24, rushpolarity was at 36.76 now is 42.52
13:38:20 <Patashu> putting even one - or + on every cell you pass over vastly improves your score
13:38:30 <Patashu> since it forces any loop to be entered at the very least once
13:38:37 <Patashu> which takes them more cycles then it takes you
13:42:13 <Patashu> !bfjoust juggernaut >+>+>->->+>+>->-(>(-)*128.-.+)*29
13:42:19 <EgoBot> Score for Patashu_juggernaut: 18.6
13:43:25 -!- Corun has joined.
13:54:04 -!- selador has joined.
14:23:13 <AnMaster> [->-<] would be p[1] = p[1] - p[0]; right? For some reason this ends up as p[1] = 255 * p[0] + p[1] ... That can't be right...
14:25:10 <Patashu> is p[0] a negative number?
14:25:30 <AnMaster> Patashu, it is an unsigned char.
14:25:47 <AnMaster> the BF - is represented as += 255
14:26:20 <Patashu> if it's mod 256 arithmetic
14:26:25 <Patashu> aren't those two expressions equal?
14:26:39 <AnMaster> Patashu, that is what I'm trying to figure out
14:26:50 <Patashu> adding 255 is like taking away 1
14:27:32 <Patashu> are you using a bf optimizer?
14:28:04 <AnMaster> Patashu, but I'm not sure x - y = x + 255y (mod 256)
14:28:13 <AnMaster> which is what the code above seems to indicate
14:28:17 <Patashu> okay let's pretend that every time you hit 256
14:28:22 <Patashu> so if you add 256 that's like adding 0
14:28:32 <Patashu> adding 257 must be like adding 1
14:28:36 <Patashu> so...adding 255 is like subtracting 1
14:28:38 <AnMaster> the issue is that it turned into multiplication there
14:28:41 <Patashu> adding x*255 is like subtracting x
14:29:02 <Patashu> I have no reason to doubt it
14:29:17 <AnMaster> Patashu, I suck at this sort of math :/
14:31:20 <Patashu> did you have any other questions
14:31:47 <Patashu> p[1] = p[1] + 255*p[0] is equivalent to p[1] = p[1] - 1*p[0] in mod 256 arithmetic
14:31:57 <AnMaster> and yes it seems to be correct
14:32:11 <AnMaster> doesn't seem very logical to me :/
14:32:25 <Patashu> well let's take p[1] = p[1] - 1*p[0]
14:32:36 <Patashu> now let's imagine that for every 1 in p[0] we also add 256
14:32:42 <Patashu> that's a nop right? since it's mod 256 arithmetic anyway
14:32:57 <Patashu> but we turn it into p[1] = p[1] - 1*p[0] + 256*p[0]
14:33:18 <Patashu> and if you can go one way you can go the other way
14:33:39 <AnMaster> ok seems to make sense. Of a kind
14:33:45 <Patashu> you agree that adding or subtracting 256 any integer amount of times is going to be a nop?
14:33:51 <Patashu> so you can implicitly do it in any expression anywhere in your code
14:33:57 <Patashu> and all it does is change the way it looks
14:35:01 <AnMaster> hm. So y = x*256 is same as y = x in mod 256
14:35:59 <Patashu> but y = x*256 is the same as y = x*0
14:36:05 <Patashu> and y = x*257 is the same as y = x*1
14:36:13 <Patashu> and y = x*258 is the same as y = x*2...
14:36:36 <AnMaster> this makes complete sense now :)
14:37:05 <Patashu> because if you add an extra k 256s for every 1 you find in x you do nothing
14:38:52 <AnMaster> so what would [+>+<] end up as... p[1] += 256-p[0] right? (of course followed by setting p[0] to 0 afterwards)
14:40:15 <Patashu> you add 1 to p[1] (256-p[0]) times
14:40:37 * AnMaster wonders what his compiler turns that out as now
14:42:30 <AnMaster> hm that turns into p[1] = 255 + p[1] - p[0] though
14:42:50 <AnMaster> that one surely can't be right
14:43:30 <Patashu> if p0 = 255 it executes once
14:44:10 <AnMaster> p[1] = 255 + p[1] - p[0] should be p[1] = 256 + p[1] - p[0] ...
14:44:22 <Patashu> which is equiv. to what you said it ought to be
14:44:50 <AnMaster> Patashu, I wonder how C will handle that since p is unsigned char* ...
14:45:41 <AnMaster> wait. it will promote to p[1] and p[0] short since 256 isn't in the range of unsigned char
14:46:12 <AnMaster> and uh... then *truncate* it back to unsigned char when assigning?
14:46:44 <Patashu> and we know it's right because we've deduced it logically earlier, that any addition or subtraction of 256 is a nop
14:46:59 <Patashu> even if it doesn't immediately click
14:47:04 <pikhq> Since this is all mod 256.
14:47:56 <AnMaster> Patashu, well... that is a bit messy, since I don't have -= anywhere in the compiler. I work with all positive values due to implementation language issues (that is: the compiler is coded in erlang, which only has bignums, so all calculations are done mod 256 manually, working with negative integers make it a bit more complex)
14:49:00 * pikhq is reminded of interp.tcl in PEBBLE.
14:49:02 <AnMaster> Patashu, that could actually be done! since the polynomial code does handle negative coefficients
14:49:26 <pikhq> Manual modular arithmetic all over the place.
14:49:30 <AnMaster> Patashu, however it would be translated to += 255*p[0] in fact :)
14:49:45 <Patashu> I think ur polynomial is wrong....
14:50:10 <pikhq> proc Modvar {var} {variable memory;variable memmap;set memory($memmap($var)) [expr {$memory($memmap($var))%256}]}
14:50:52 <AnMaster> pikhq, I actually do it lazy... After all passes that can merge nodes I have a cleanup pass which goes through and mod 256s all relevant values
14:51:32 <AnMaster> which I admit is rather messy design
14:52:01 <AnMaster> it was the one easiest to implement when I realised "oops, this should have been "rem 256" all over the place.
14:52:13 <AnMaster> new code does the rem 256 though
14:56:27 <AnMaster> <Patashu> p[1] = p[1] + 255*p[0] is equivalent to p[1] = p[1] - 1*p[0] in mod 256 arithmetic
14:56:45 <AnMaster> wouldn't that mean += -p[0] ? is same as += 255*p[0] ?
14:56:46 <Patashu> because between the first and second you always add an integer number of 256s
14:57:09 <AnMaster> Patashu, so that is the answer to <Patashu> why 255 specifically?
15:08:45 -!- jix has quit ("leaving").
15:26:09 -!- Patashu has quit ("Patashu/SteampunkX - MSN = Patashu@hotmail.com , AIM = Patashu0 , YIM = Patashu2 , Googletalk = Patashu0@gmail.com .").
15:55:07 -!- tombom has joined.
16:30:27 -!- oerjan has joined.
16:33:34 -!- Corun has quit ("Leaving...").
16:36:11 <AnMaster> http://pastebin.ca/1442391 <-- funny mess up of mandelbrot.b
16:39:21 -!- kar8nga has joined.
16:47:03 <oerjan> The Font: Small button doesn't seem to be working :(
16:47:30 <oerjan> so i presume i'm seeing it even more messed up than it has to be
16:48:40 <oerjan> huh there was a badly formatted menu of fonts in the upper right corner, and _those_ work
16:49:44 <oerjan> Ooh, it's a pwetty buttewfwy!
16:59:03 <oerjan> <AnMaster> Patashu, but I'm not sure x - y = x + 255y (mod 256)
16:59:06 -!- kar8nga has quit (Remote closed the connection).
16:59:28 <oerjan> modulo arithmetic works across addition, subtraction and multiplication
17:00:45 <oerjan> division and powers are more subtle
17:00:53 <Deewiant> x - y = x - y + 256y = x + 255y (mod 256)
17:06:20 <oerjan> <AnMaster> wait. it will promote to p[1] and p[0] short since 256 isn't in the range of unsigned char
17:06:44 <oerjan> i vaguely thought C converted all chars before doing arithmetic anyway
17:07:31 <oerjan> or wait was this about function arguments...
17:08:28 <oerjan> 06:47:56 <AnMaster> Patashu, well... that is a bit messy, since I don't have -= anywhere in the compiler. I work with all positive values due to implementation language issues (that is: the compiler is coded in erlang, which only has bignums, so all calculations are done mod 256 manually, working with negative integers make it a bit more complex)
17:09:02 <oerjan> does erlang give negative remainders if you do remainder(negative, positive)?
17:09:22 <AnMaster> you mean negative rem positive?
17:10:04 <oerjan> well yes then keeping things positive may simplify things
17:10:26 <oerjan> AnMaster: haskell has both rem and mod, where mod always gives positive for positive divisor
17:10:50 <oerjan> and then you don't need to worry about negative things
17:11:09 <AnMaster> oerjan, iirc the semantics for rem is "whatever the compiler that compiled the VM used". So for Erlang compiled with GCC that means C99 semantics.
17:11:18 <oerjan> haskell is more precise there
17:11:28 <AnMaster> or rahter... for BEAM compiled with GCC.
17:12:12 <AnMaster> (the bit of Erlang written in C, rather than in Erlang)
17:12:43 -!- Corun has joined.
17:15:20 <lifthrasiir> AnMaster: i'm redesigning expression class. it should canonicalize "(Expr[3] + 4) - Expr[3]" into 4 on-the-fly. :p
17:16:18 <AnMaster> lifthrasiir, well I have polynomials working mostly. That is a lot simpler to implement. Just I'm fixing bugs in various other passes and such that were uncovered by adding polynomials...
17:16:56 <lifthrasiir> i wish there were multiset in python... :S
17:17:11 <lifthrasiir> actually, what i needed was "frozen" multiset.
17:17:20 <lifthrasiir> AnMaster: set with duplicate elements. for example, {1, 2, 2, 3, 3, 3}.
17:17:40 <AnMaster> what I do is basically "convert the 50 first loops we can convert to polynomials. Do a binary search from that. Once found which exact convert caused issues try to figure out why.
17:18:11 <AnMaster> lifthrasiir, oh? How comes? I use sets in erlang because I only want one of each. Otherwise I would use some other data structure
17:18:31 <AnMaster> lifthrasiir, not really. Since I dump original loop and matching polynomial to stdout
17:18:39 <AnMaster> and some other in the middle data
17:18:59 <AnMaster> sometimes it is another pass. Where I can then do a binary search on that.
17:19:27 <AnMaster> anyway iirc erlang has some duplicate_set thingy
17:19:33 <comex> http://www.faqts.com/knowledge_base/view.phtml/aid/4355
17:19:38 <AnMaster> lifthrasiir, couldn't you implement it for python?
17:20:11 <comex> there's a real module for it
17:20:20 <lifthrasiir> AnMaster: a mutable multiset is easy to implement in python, since dict already provides sufficient methods. but what if it has to be immutable?
17:20:35 <AnMaster> lifthrasiir, copy it every time?
17:20:45 <AnMaster> instead of making a change to it
17:20:58 <oerjan> lifthrasiir: a multiset is merely a dictionary with integer values...
17:21:11 <lifthrasiir> AnMaster: in python, "immutable" also conveys the meaning of "hashable".
17:21:23 <AnMaster> lifthrasiir, you can't hash other objects?
17:21:29 <lifthrasiir> oerjan: yes, but hashable dict can be hard to make.
17:21:37 <comex> store it as a tuple
17:22:47 <comex> and just have the add method maintain sorted order
17:23:01 <lifthrasiir> AnMaster: that's exactly my current method: def __hash__(self): return hash(tuple(self.items())). but i'm not sure it's dependent to implementation.
17:23:53 <AnMaster> lifthrasiir, why do you need it hashable btw? Going to store it in a dict?
17:24:02 <lifthrasiir> AnMaster: convert the dictionary into hashable list of items; but it depends on the order of items, and i'm not sure it's unique.
17:24:40 <comex> so don't use a dict, use a tuple for storage
17:24:40 <lifthrasiir> for example (a + a + a) == (a * 3), and i store them as same internal format: {a: 3}.
17:24:43 <AnMaster> I generally don't have this problem in Erlang. I can use lists as keys of dicts. I do that when simplifying polynomials
17:24:58 <comex> which is faster anyway, as you don't have to make a new tuple for every hash
17:25:42 <AnMaster> Term = Coefficient * Variables Variables = List of 2-tuples: {Offset,Exponent}
17:25:48 <lifthrasiir> comex: after some thought, i agree to you. it could be simpler using sorted list.
17:26:12 <AnMaster> to merge 2*x + 4*x I use a dict with the list of variables as the key
17:26:17 <lifthrasiir> (but i'm also not sure that tuple manipulation costs less than dict.)
17:26:41 <AnMaster> (since I just merged 2*x*x into 2*x^2 and similar before)
17:27:16 <comex> I don't mean manipulation
17:27:24 <comex> hash(tuple(self.items()))
17:27:42 <comex> is going to make a new list, and then a new tuple out of it
17:27:59 <comex> every time you want a hash
17:29:32 <lifthrasiir> comex: afaik in python __hash__ is called only once per object, but that could be a problem when copying objects.
17:31:06 <lifthrasiir> AnMaster: anyway i'm going to remove AdjustMemory ("add" in your code) in favor of SetMemory ("set" in your code). :p
17:31:21 <AnMaster> lifthrasiir, how will you handle ,++ or such
17:31:29 <AnMaster> or when you don't know the value before at all
17:32:13 <AnMaster> for some reason THAT one breaks...
17:34:10 <AnMaster> ok it seems to be related to the whole converted loop being:
17:34:14 <AnMaster> [{bfn,add,0,255,undefined,undefined},
17:34:14 <AnMaster> {bfn,set,2,0,undefined,undefined},
17:34:14 <AnMaster> {bfn,add,4,1,undefined,undefined}],
17:34:17 <lifthrasiir> comex: ah, i figured out why i didn't use sorted list. in python 3 cmp doesn't work on different types.
17:34:41 <AnMaster> where that is "{bfn,Opcode,Offset,Value,StuffUsedForLoops,MoreStuffUsedForLoops}"
17:35:29 <AnMaster> lifthrasiir, default value in struct definition for bfn (which stands for bf node)
17:35:39 <AnMaster> the second undefined contains the "body" of loops and ifs and such
17:35:50 <AnMaster> the third is various metadata for loops and polynomials
17:36:10 <AnMaster> contains stuff like what cells are modified by loops and so on
17:37:00 <AnMaster> lifthrasiir, and the "bfn" there is just due to using a record. Internally in erlang it ends up as a tagged tuple
17:38:10 <AnMaster> http://pastebin.ca/1442460 <-- I fail to see anything wrong there hm
17:38:17 * AnMaster goes to dump it just after that pass
17:38:20 <oerjan> hmph DMM leaves for vacation and the next day the site is broken in several ways...
17:38:48 <AnMaster> oerjan, do you think it is intentional? I suspect it is some kind of practical joke...
17:39:21 <AnMaster> well not "practical joke". wrong word
17:40:01 <AnMaster> I could live with the font mess up.... but NOT with the broken image link
17:40:21 <oerjan> also a missing part of the navigation bar
17:42:01 <oerjan> what _was_ 2317 again? it worked yesterday but i've forgotten exactly what it was
17:43:26 * oerjan turned on the transcript
17:43:36 <oerjan> i have definitely _not_ seen that before
17:44:11 <comex> 12:34 < lifthrasiir> comex: ah, i figured out why i didn't use sorted list. in python 3 cmp doesn't work on different types.
17:44:22 <oerjan> could i simply have forgotten to visit yesterday? no, because i _did_ read Comments on a Postcard.
17:44:34 <oerjan> AnMaster: something is definitely fishy
17:45:53 <lifthrasiir> anyway sorting [1, (2,), [3]] etc will fail in python 3. that's why i choose not to use sorted list.
17:46:11 <oerjan> AnMaster: there was a different comic there yesterday, i think, although i don't remember what it was. i suppose the update _could_ have failed horribly to overwrite it
17:46:30 <AnMaster> oerjan, wasn't it the one two back
17:46:41 <AnMaster> also yeah, this must be intentional...
17:46:49 <AnMaster> interesting way to irritate users
17:46:52 <oerjan> oh wait you are right, look at the date
17:47:01 <comex> oh, it was deprecated in favor of writing 6 boilerplate methods
17:47:22 <oerjan> oh, 2318 is missing the date
17:47:28 <oerjan> while 2317 has today's
17:51:48 <AnMaster> ok I minimised the broken diff
17:52:09 <AnMaster> oerjan, ok. This MUST be intentional...
17:52:33 <AnMaster> I just don't believe it would be this messed up otherwise
17:57:58 -!- selador has quit ("ChatZilla 0.9.84 [Firefox 3.0.10/2009042513]").
18:00:04 <oerjan> oh yeah the forum reminded me there was another glitch: "Intriguingly, the previous poll isn't showing when you answer the current one"
18:00:22 <oerjan> and that was there yesterday too
18:01:04 <oerjan> this appears to be approaching astronomical improbability :D
18:01:47 <oerjan> oh no, DMM actually noticed that before he left
18:04:37 -!- davinci has joined.
18:06:01 -!- davinci has left (?).
18:18:01 <AnMaster> oerjan, what did he say about that
18:18:20 <oerjan> that it was a simple typo
18:18:27 <oerjan> (that was the poll error)
18:40:56 -!- kar8nga has joined.
18:42:05 -!- ehird has left (?).
18:42:32 -!- ehird has joined.
18:49:09 <ehird> 00:26 pikhq: ehird: http://imgur.com/B8SqH.png
18:49:09 <ehird> 00:26 pikhq: Had to go cook.
18:50:09 <ehird> 03:05 zzo38: Hay! You told me I didn't need a e-mail address to register
18:50:09 <ehird> 03:05 zzo38: But now it says I do
18:50:11 <ehird> 03:06 zzo38: It also says I need to register and IDENTIFY to post message on ##C channel.
18:50:13 <ehird> 03:06 zzo38: But IDENTIFY is not a valid command and I can
18:50:15 <ehird> 03:06 zzo38: 't register without password/email but I don't want either of those things
18:50:17 <ehird> lol— tough. also IDENTIFY is a nickserv command like register.
18:50:19 <ehird> 03:07 coppro: /join #freenode ← he uses telnet to irc
18:54:51 <ehird> AnMaster: "DuPont Displays Surpasses Million Hour Milestone for Lifetime of New OLED Material" >100 years
18:55:15 <AnMaster> ehird, I wonder how they tested that...
18:55:21 <ehird> 06:36 GregorR-L: I'm now uſing all lengþs. ← imaginary patents.
18:55:46 <ehird> lifthrasiir: pikhq's english shorthand system.
18:56:00 <pikhq> The result of being bored in high school.
18:56:04 <lifthrasiir> at first glance it looked like japanese :|
18:56:09 <ehird> pikhq: what does it say?
18:56:22 <pikhq> "The quick brown fox jumped over the lazy dogs."
18:56:46 <ehird> anyhoo since GregorR has all but obsoleted my interp by adding all legngths I won't bother working on it :P
18:57:00 <ehird> pikhq: how does it work?
18:57:01 <pikhq> And managed not to include all of my symbols, courtesy of diactrics. :p
18:57:09 <Asztal> it also pushed my program to the bottom >:(
18:57:20 <ehird> AnMaster: adding polarity is a ~3 minute job
18:57:31 <Deewiant> ehird, AnMaster: And was also done.
18:58:03 <AnMaster> ehird, I guess it is because you took too long for him ;P
18:58:11 <Deewiant> I noticed the lack of vowels; inferred only from context?
18:58:24 <ehird> 18:57 pikhq: ehird: Abjad.
18:58:25 <ehird> 18:57 pikhq: n vwls.
18:59:19 <pikhq> Aside from that, it's just the Roman alphabet, simplified into 1 stroke per character.
18:59:59 <Deewiant> What do t and h look like in isolation
19:00:19 <pikhq> Deewiant: T looks like the first character without the ` over it.
19:00:55 <pikhq> h looks like... Hmm. Actually, it looks closest to 'þ' written backwards.
19:01:00 <ehird> AnMaster: So, looks like I'll be buying my display from DuPont. Maybe.
19:01:24 <ehird> I won't get this fancy long-lived one, just a regular OLED display.
19:01:27 <pikhq> No, only when with another letter. "gh", "th", "ph", etc.
19:01:55 <pikhq> It's amusing that it looks like kana, since I didn't learn kana until a bit *after* that.
19:02:02 <ehird> AnMaster: Either 163.92 or 136.6.
19:02:10 <ehird> Either way, ridonkulously high.
19:02:30 <ehird> 1280x1024 at 10" and 12", respectively.
19:02:41 <pikhq> (I write a ' to indicate a starting vowel, makes it a bit easier to read)
19:02:51 <ehird> With DPI set accordingly in the OS, of course; that high resolution's just for extra smoothness and for media.
19:02:54 <Asztal> `sztl would work on IRC, I suppose
19:03:03 <AnMaster> pikhq, He fails at faking nick change for all clients expect possibly xchat.
19:03:17 <ehird> AnMaster: since it's impossible to do it for all clients, who carse?
19:03:33 <ehird> who cares if he fails faking it
19:04:02 <AnMaster> I don't care a lot about it. Just for informative purposes.
19:04:15 <Asztal> if I could get a -!- there, I'd do that, but I don't really know what the other clients do
19:04:27 <ehird> ubuntu/gnome are really good for high-dpi displays
19:04:31 <ehird> everything just works if you set the dpi
19:05:16 <ehird> Huge advantage of high DPI displays: Subpixel rendering + no hinting — without any blurriness whatsoever.
19:05:25 <ehird> Although you get that at about 100dpi.
19:05:50 <Deewiant> Your "huge advantage" is my "can't tell the difference"
19:06:14 <pikhq> Deewiant: You're blind.
19:06:17 <ehird> Why don't you go and buy a braille screen, I'm sure you'll be just fine
19:06:24 -!- tombom has quit (Read error: 54 (Connection reset by peer)).
19:06:26 <ehird> not having fully-functioning eyes and all
19:06:48 <Deewiant> For this particular case I note that I actually haven't had a chance to compare
19:06:55 <Deewiant> But I suspect it'd be like that.
19:06:58 <ehird> Then your statement is very silly :P
19:07:07 <Deewiant> It applies to most of your monitor hype :-P
19:07:24 <ehird> you mean... i like good displays?
19:07:27 <ehird> i dont' recall hyping anything
19:07:50 <Deewiant> "to create interest in by flamboyant or dramatic methods"
19:08:38 <ehird> saying that 1280x1024 is way too small = dramatic?
19:09:07 <pikhq> 1280x1024 is kinda small unless you're talking TV set.
19:09:16 <ehird> pikhq: Deewiant said it was "just fine"
19:09:19 <pikhq> And even then... Go for full 1920x1080?
19:09:33 <ehird> yeah, it's not as if Full HD screens even cost a lot
19:09:36 <pikhq> ehird: Sure. In 1996.
19:09:50 <ehird> you can get a 23" LCD TV w/ real 1080p for ~£200
19:09:54 <Deewiant> ehird: Responding with "hahahahahaha" when I say it isn't is flamboyant :-P
19:09:59 <ehird> and most tvs cost about that much
19:10:11 <ehird> Deewiant: 512k is enough for anything
19:10:33 <pikhq> ehird: Eh, there's cheaper TV sets. Granted, those are at best EDTV these days...
19:11:00 <ehird> [[It would be better for the "Donky" of the U.S. Democratic Party to lick the carrot.]] —North Korea
19:11:01 <Deewiant> ehird: That style of response is flamboyant regardless of the statement.
19:11:15 <ehird> Deewiant: do we agree on the definition of flamboyant here?
19:11:16 <Deewiant> Aside: you guys have pretty damn cheap TVs.
19:11:39 <pikhq> Deewiant: TVs are cheap.
19:11:47 <ehird> Deewiant: not really the colloquial meaning
19:11:51 <ehird> pikhq: Tech is expensive in finland
19:12:00 <Deewiant> pikhq: I got a 19" TV for 199€, the next-cheapest one would have been around 299€.
19:12:15 <pikhq> Jebus. 19" for 199€?
19:12:30 <Deewiant> pikhq: Cheapest TV I could find.
19:12:39 <ehird> They're just badly-tuned monitors
19:12:51 <pikhq> I could just about get a 19" monitor, computer and *tuner card* for that.
19:13:30 <Deewiant> Tuner cards appear to be around 90-120 € here.
19:13:33 <pikhq> Oh. 199€ is according to Google $281.
19:13:43 <pikhq> Just about? Yes, I *definitely* could.
19:14:02 <Deewiant> That 23" with 1080p for £199 is ass-cheap.
19:14:06 <pikhq> I can pick up a DTV tuner for $20-$50....
19:14:34 <pikhq> If I skrimped, I could get a computer for another $150...
19:14:41 <Deewiant> pikhq: 19" monitors are upwards of 100 €.
19:14:45 <pikhq> Maybe even $100 if I didn't bother with CPU at all.
19:15:09 -!- jix has joined.
19:15:09 <Deewiant> And a computer on top of that will be more than the -20 to 10 € remaining. ;-)
19:15:15 <pikhq> ("CPU? Who needs that if it's just going to tune?")
19:16:04 <ehird> 19:14 Deewiant: That 23" with 1080p for £199 is ass-cheap.
19:16:24 <ehird> Deewiant: Yeah, we have pretty good pricse.
19:16:24 <pikhq> So, yes. It would be cheaper for me to build a barebones DVR setup than it would be for you to buy a TV set.
19:16:28 <pikhq> That's a bit crazy.
19:16:42 <ehird> But seriously, televisions are obsolete.
19:16:46 <Deewiant> I've become accustomed to it. :-P
19:17:01 <Deewiant> Nothing wrong with buying an obsolete device every now and then
19:17:02 <ehird> Deewiant: Here's the ass cheap TV: http://www.ebuyer.com/product/153223
19:17:15 <ehird> Deewiant: = 263.38 eur
19:17:33 <pikhq> ehird: DVR setup FTW. If you really *must* watch TV, at least you can skip the commercials.
19:17:52 <ehird> pikhq: Why even bother? Everything apart from the news is downloadable, and the news can be streamed for free online.
19:17:57 <ehird> (... assuming BBC, of course.)
19:18:02 <bsmntbombdood> for a monitor, is is 1920*1200, 22", or 19200*1200, 24" better?
19:18:14 <ehird> bsmntbombdood: the 19200. it has ten times the height
19:18:20 <Deewiant> ehird: About the same price here, actually. I'd've expected at least 290.
19:18:21 <ehird> bit of a skewed aspect ratio though ;)
19:18:36 <ehird> bsmntbombdood: But, depends what you want. 22" packs the same stuff into more detail, due to packing the pixels closer.
19:18:45 <pikhq> ehird: You're not very likely to find 10 Mbit/s MPEG2 streams from the net. :p
19:18:46 <ehird> bsmntbombdood: 24" is about 96dpi, which is what most images, etc on the interweb assume.
19:18:47 <Deewiant> ehird: But then, £30 on top does make it less ass-cheap. :-P
19:18:54 <ehird> And also, the default settings and such assume it, generally.
19:19:00 <ehird> So it'll work OOTB more smoothly.
19:19:06 <pikhq> ... Granted, if you've got cable, you're not even getting that.
19:19:18 <ehird> pikhq: Ultra-high quality is mandatory for breaking news, clearly.
19:19:32 <ehird> Deewiant: I have £30 a few meters away from me, want it? :P
19:19:36 <Deewiant> In this case, I just wanted something cheap with component inputs.
19:19:41 <ehird> (well, okay, so I don't have £30 lying everywhere.)
19:19:43 <ehird> (would be nice tho.)
19:20:02 <ehird> Deewiant: OK, that'll be £20 shipping, £11 handling charge
19:20:09 <pikhq> ehird: You're discussing the US. We think that making it possible for people to *download* shows they want to watch is EBIL!!!
19:20:39 <ehird> pikhq: I believe you can stream bbc news internationally
19:20:58 <pikhq> ehird: And what if you want to watch something other than BBC news?
19:21:08 <pikhq> So we'd rather use several hundred Mbit/s streaming it over the air. And over satellites. And over cables.
19:21:13 <ehird> "The BBC News channel is available in the UK only. Don't forget, to watch TV online as it's being broadcast, you still need a TV License."
19:21:24 <ehird> pikhq: If you want something else, it's probably not live, so download the series. In HD>
19:21:41 <pikhq> Thus my rant about the US.
19:21:54 <ehird> Jaywalking isn't legal either :P
19:22:00 <pikhq> And diverging into a rant about us not using multicast.
19:22:21 <ehird> TV/Radio have one architectural advantage:
19:22:28 <ehird> Broadcast scales way better than streaming.
19:22:37 <ehird> Because it doesn't have to do anything at all :P
19:22:43 <pikhq> Thus why I said multicast.
19:23:17 <pikhq> With multicast, the streamer just sends out a single stream. The routers make sure the stream gets to everyone. ;)
19:23:18 <ehird> [[For anyone who isn't familiar with the name "Prometeia," it is special device that chills an Intel or AMD processor down to an operating temperature of between -20°C to -30°C, with the end goal being better overclocking.]]
19:23:30 <ehird> Overclockers never stop surprising me with what they will buy.
19:23:49 <ehird> It's a fridge for your CPU.
19:24:08 <Deewiant> It's also nearly 10 years old :-P
19:24:13 <pikhq> I've seen some overclocker that said "Screw that, that's just stupid".
19:24:25 <pikhq> And bought a minifridge and made that into a computer case.
19:24:27 <ehird> Deewiant: Yeah, I bet it can only cool an i7 @ 4ghz to -5 under load.
19:24:39 <ehird> If there's one thing overclockers are, it's DIY.
19:25:01 <ehird> On the subject of crazy cooling systems...
19:25:26 <ehird> http://www.zalman.co.kr/DataFile/product/RESERATOR-1-V2_01_b(0).jpg Combined pump, reservoir and... space heater.
19:25:45 <ehird> bsmntbombdood: never said it wasn't
19:25:49 <ehird> i'm considering it for my new build
19:26:00 <ehird> one can only dissipate ~200W of heat continually
19:26:15 <ehird> probably drill a hole in the wall, and mount it on the rarely-used stairs, though
19:26:23 <ehird> don't want to be near two of them under load
19:26:35 <ehird> bsmntbombdood: i could just buy an industrial radiator that can dissipate 200w
19:26:42 <ehird> and hook it up to the reserator
19:26:46 <ehird> that'd be less fuss than DIYing it
19:27:06 <ehird> bsmntbombdood: the pump is a bit wimpy, though
19:27:12 <ehird> 5 watts, 30litre/hour
19:27:23 <ehird> otoh, anything more would be non-inaudible
19:27:36 <ehird> two of them is probably quieter than replacing the pump with something louder/more powerful
19:27:49 <ehird> but if I'm mounting this on the stairs it doesn't really matter :)
19:28:14 <ehird> FireFly: shush you :)
19:28:26 <ehird> bsmntbombdood: the reserator is ~8.5kg when filled
19:28:26 <bsmntbombdood> get a large sheet of copper, solder copper pipe on to it in a purdy design
19:28:39 <ehird> 6kg base rate + 2.5 litres of cooler
19:28:44 <ehird> bsmntbombdood: not on this wall, kthx
19:28:53 <ehird> the i7 and a gtx graphics card are gonna produce a lot of heat
19:29:02 <ehird> 25C-30C is hot enough for me as it is
19:29:07 <ehird> bsmntbombdood: I'm going to mount it on the other wall
19:29:21 <ehird> to the side of my monitor and further forward is a wall
19:29:25 <bsmntbombdood> it's not like you've got more heat than fan cooling
19:29:34 <ehird> beyond that wall, is air above rarely-used stairs
19:29:45 <ehird> step 1. drill three holes
19:29:54 <ehird> in, out tubes, and power lead
19:30:01 <ehird> step 2. get brackets, mount on stair wall
19:30:04 <ehird> step 3. connect it up
19:30:12 <ehird> bsmntbombdood: yes, but air heating just goes out the exhaust vent
19:30:14 <FireFly> And no-one in your house complains?
19:30:28 <ehird> it doesn't sit there 2ft high pumping out heat
19:30:43 <ehird> bsmntbombdood: the exhaust vent on my floor, facing the wall, you mean
19:30:56 -!- coppro has quit (Remote closed the connection).
19:31:04 <ehird> bsmntbombdood: anyway, the stairs are colder.
19:31:07 <ehird> also, this is cooler. ;)
19:31:18 <ehird> FireFly: weell, that may be a slight issue
19:31:26 <ehird> FireFly: but, the stairs are used like once every few months
19:31:33 <bsmntbombdood> anyway, if it's on the other side of a wall, you can use a cheaper, active radiator
19:31:34 <ehird> and it'd be too high to bash anyone
19:31:37 <ehird> FireFly: also, cool as in awesome.
19:31:48 <ehird> bsmntbombdood: the stairs are a great amplifier
19:31:51 <ehird> I can hear anything that goes on in there
19:32:09 <ehird> also, fans have significantly less cool factor
19:32:10 <FireFly> You should have your speakers there :D
19:32:34 <ehird> bsmntbombdood: yeah but if I go down that route my whole computer ends up in the basement (a basement which this flat does not have)
19:32:44 <ehird> and that's so much less fun than making it silent
19:32:51 <bsmntbombdood> i've already got a hole drilled in my floor down to the crawlspace :D
19:32:55 <ehird> bsmntbombdood: one of the less practical ideas I had while thinking about this,
19:33:02 <ehird> (note that I'm one storey up in a flat)
19:33:08 -!- tombom has joined.
19:33:09 <ehird> mount it on the outside wall
19:33:12 <ehird> put in protective box
19:33:14 <ehird> drill holes through wall
19:33:16 <FireFly> bsmntbombdood, that's my idea
19:33:23 <ehird> bsmntbombdood: i could overclock in the winter only
19:33:45 <jix> !bfjoust suicide <
19:33:47 <ehird> there's a word for that idea, though, actually two of them. impractical; expensive :-D
19:33:51 <EgoBot> Score for jix_suicide: 0.0
19:33:57 <bsmntbombdood> ehird: get a window-mounted air conditioner, and hook up a heat exchanger to the cold side
19:34:17 <jix> FireFly: nope
19:34:25 <ehird> bsmntbombdood: maybe
19:34:26 <jix> FireFly: leaving the arena makes one lose
19:34:27 <Asztal> !bfjoust true_suicide ]
19:34:30 <ehird> the reserator is pretty though
19:34:39 <EgoBot> Score for Asztal_true_suicide: 4.6
19:34:47 <ehird> bsmntbombdood: i'm almost totally set on a passive-radiator watercooling system, though
19:34:56 <ehird> although then the coil whine from the GTX 260 up becomes an issue
19:35:00 <FireFly> I thought one could still win if the opponent suicided afterwards (by clearing its own flag)
19:35:11 <tombom> !bfjoust suicide fake_suicide
19:35:15 <jix> FireFly: that's only if your program ends
19:35:17 <Asztal> heh, well, a syntax error loses to a <, AFAIK.
19:35:23 <ehird> bsmntbombdood: that was 'cuz I could passively cool a 9800gtx+
19:35:33 <ehird> bsmntbombdood: with watercooling, all those issues just - ehm - drip away
19:35:48 <ehird> step 1, get a decent radiator, step 2, put all the components you want in the loop
19:35:50 <ehird> bsmntbombdood: yeah
19:35:58 <ehird> bsmntbombdood: some of the coil things on the card make a whining noise
19:36:01 <ehird> apparently only for the first few months
19:36:05 <ehird> depending on who you ask
19:36:16 <EgoBot> Score for Deewiant_fake_suicide: 7.7
19:36:17 <EgoBot> Supported commands: addinterp bf_txtgen bfjoust daemon daemons delinterp fyb help info kill mush userinterps 1l 2l adjust asm axo bch bct befunge befunge98 bf bf16 bf32 bf8 bfbignum boolfuck c chiqrsx9p choo cintercal clcintercal cxx dimensifuck echo forth glass glypho google gregor hello kipple lambda lazyk linguine malbolge notecho num ook pbrain perl qbf rail rhotor rot13 sadol sceql sh show slashes test trigger udage01 underload unlambda whirl yod
19:36:17 <EgoBot> Score for tombom_suicide: 0.0
19:36:17 <EgoBot> Score for tombom_suicide: 0.0
19:36:54 <FireFly> Egobot, was that the one with lots of bf based commands?
19:37:00 <ehird> bsmntbombdood: that's mah plan
19:37:07 <ehird> bsmntbombdood: maybe even two, in SLI.
19:37:13 <EgoBot> That is not a user interpreter!
19:37:15 <ehird> depending on how excessive and moneybags I feel
19:37:37 <ehird> bsmntbombdood: GTX 260 and up
19:37:43 <ehird> all the GTXs suffer from it
19:37:52 <ehird> Deewiant: since you're the person who showed me that ssd article— do you think using an using intel ssd for an os drive, and then a stuttery-but-good-sequential-throughput SSD for a data might work well? i don't see why that wouldn't work
19:38:25 <Deewiant> If most of your data usage is sequential then sure, I guess
19:38:29 <ehird> bsmntbombdood: after all this, all I need is an anechoic chamber to enjoy it in
19:38:49 <bsmntbombdood> and why not put the computer on the other side of the wall?
19:39:25 <ehird> a bit less (few hundred) than $4,000 for the aircooled version ($4k was from endpcnoise.com preassembled; the rig i was going to build had some changes but would prolly come out cheaper due to no system integrator cost)
19:39:39 <ehird> so, remove costs for quiet case, heatsinks, fans etc and add watercooling costs...
19:39:47 <ehird> otoh, I found some ways to cut the costs down
19:39:52 <ehird> it's all very much in flux
19:39:56 <ehird> priority one, performance and silence
19:40:01 <ehird> priority two, money i can actually afford
19:40:10 <ehird> bsmntbombdood: well, I'm not going to
19:40:17 <ehird> i'm doing it iteratively
19:40:28 <ehird> make awesome computer, make affordable. if awesomeness is compromised, go back to step 1
19:40:33 <ehird> repeat until both criteria are fulfilled
19:40:51 <bsmntbombdood> gtx 295 - $500, i7 extreme, $1000, 4 x25-ms, $1200
19:41:09 <ehird> bsmntbombdood: watercooling costs a lot
19:41:24 <ehird> bsmntbombdood: 12gb of expensive ddr3 is more like $1000
19:41:38 <pikhq> Not if you do it the 8G stick way.
19:41:47 <ehird> pikhq: then it's $984758934579832478293784572893467834589347238947
19:41:51 <ehird> because you can't get 8g sticks
19:42:13 <ehird> pikhq: oh it wouldn't cost -that- much to get them invented and made ;)
19:42:51 <ehird> bsmntbombdood: $3100 for cpu, gfx card, ram and 160gb x25-m
19:42:54 <pikhq> That includes the cost of bribing the controllers of the market to allow them to be on the market.
19:43:02 <ehird> add water cooling costs, case etc and you get a bit over $4000
19:43:29 <ehird> conclusion: hardware is cheap. relatively.
19:44:07 <ehird> interwebs are being slow
19:44:19 <ehird> one thing that is totally stupid is any watercooling rig that isn't passive
19:44:23 <ehird> total waste of money
19:44:39 <ehird> bsmntbombdood: pfft, SATA?
19:44:59 <ehird> [[What's more, even Intel's fastest X25-E flash drive is limited to 250MB/s reads and 170MB/s writes]]
19:45:05 <ehird> FOCUS ON RANDOM PERFORMANCE FOR AN OS DRIE
19:45:07 <ehird> RAAAAAAAAAAAAAAAAAAAGE
19:45:16 <bsmntbombdood> so, it'll be 1000 times better to get a motherboard that can support that
19:45:32 <bsmntbombdood> because the processor's fastest and lowest latency connection to _anything_ is its ram
19:45:43 <ehird> i wonder what nice watercooled psus you can get
19:45:51 <ehird> probably not anything under 2345897234897238942389429347234 kilowatts
19:45:57 <ehird> because all watercoolers are excessive amirite
19:46:18 * pikhq wants a computer with battery backed RAM. Making a RAM disk feasible.
19:46:34 <ehird> bsmntbombdood: sata=3gb/s
19:46:43 <pikhq> bsmntbombdood: 6 gb/s soon.
19:46:48 <ehird> bsmntbombdood: well yeah
19:47:16 <bsmntbombdood> pikhq: it's very easy to do with a good operating system
19:47:40 <ehird> bsmntbombdood: can the dual 1366 server mobos take two i7s?
19:47:44 <ehird> i seem to recall that they only fit xeons or sth
19:47:54 <pikhq> bsmntbombdood: "Battery backed RAM".
19:48:03 <ehird> bsmntbombdood: you can have two xeons, why not two i7s
19:48:11 <pikhq> I didn't know an OS could magically make your RAM battery backed.
19:48:31 <bsmntbombdood> ups tells computer when to start backing up ramdisk
19:48:47 <ehird> fuck battery backed ramdisks
19:48:51 <ehird> ssd backed ramdisks, yes.
19:48:58 <pikhq> Sure, if you never shut your system down.
19:49:23 <ehird> bsmntbombdood: do you know?
19:49:31 <bsmntbombdood> pikhq: so it takes 2 minutes to shut down, big deal
19:49:54 <ehird> kinda lame but i don't need more than 4 cores
19:50:14 <ehird> doubt the performance advantage from 8 is really noticeable apart from in things specifically optimized for it
19:50:20 <ehird> if I wanna get mega parallel i'll talk to my gpu
19:50:52 <ehird> [[Busting the ANS-9010 open reveals eight 240-pin DDR2 DIMM slots]]
19:50:52 <pikhq> ehird: Like compile jobs, encode jobs, etc. :p
19:50:57 <ehird> bsmntbombdood: it isn't even fucking ddr3 :)
19:51:22 <ehird> pikhq: -j6 vs -j12 isn't going to make much difference unless your project is basically dependency-free
19:51:32 <ehird> pikhq: encode jobs can be done on the gpu
19:51:38 <ehird> and there are products doing that in development
19:51:53 <ehird> pikhq: can mencoder do that?
19:52:14 <pikhq> mencoder can do encodes in parallel. I don't think it can do encodes on the GPU, though.
19:52:29 <pikhq> *Decodes*, sure, but not encodes.
19:52:57 <ehird> Well, that's its problem
19:52:59 <ehird> bsmntbombdood: dunno
19:53:07 <ehird> i've seen figures using gpu encoders to mpeg4 and the like
19:53:17 <ehird> 5x faster than a multicore cpu or so
19:53:20 <pikhq> bsmntbombdood: All modern GPUs are TC.
19:53:27 <ehird> pikhq: he knows _that
19:53:30 <pikhq> And have builtin decoding hardware besides.
19:53:30 <ehird> but specialization = moar speed
19:54:00 <ehird> but yeah, gfx card of radeon 4850 or above caliber vs high-end gigahertzy 4-core cpu
19:54:06 <ehird> gfx cards kicks the cpu's ass at encoding video.
19:54:20 <ehird> bsmntbombdood: oh my god
19:54:25 <ehird> bsmntbombdood: their test system for that ddr drive
19:54:27 <ehird> bsmntbombdood: [[Processor Pentium 4 Extreme Edition 3.4GHz]]
19:54:30 <pikhq> If you've got an encoder for said graphics card, yes.
19:54:33 <ehird> TIME MACHINE TIEM MACHINE
19:54:50 <ehird> [[Memory size1GB (2 DIMMs)]]
19:55:00 <ehird> [[Graphics Radeon X700 Pro 256MB with CATALYST 5.7 drivers]]
19:55:04 <ehird> oh this shit is just piling up the power
19:55:12 <ehird> catalyst is up to 9.something :D
19:55:20 <pikhq> I've got a system better than that. And it's low-end 3 year old hardware.
19:55:21 <ehird> wow, xp service pack 2!
19:55:28 <ehird> yeaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaah
19:55:32 <ehird> The test systems' Windows desktop was set at 1280x1024 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.
19:56:21 <pikhq> I've got a lower refresh, but that's because my LCD's somewhat cheap.
19:56:26 <ehird> bsmntbombdood: http://techreport.com/r.x/acard-ans9010/worldbench-overall.gif
19:56:37 <ehird> bsmntbombdood: RAID-0ing two of those ramdisks?
19:56:38 <pikhq> But man. I outspecced their test system 3 years ago.
19:56:42 <ehird> only 8 points better than a x25-m
19:56:44 <pikhq> And I was buying *old hardware*.
19:57:00 <ehird> 160gb for cheaper than 16gb and the same performance hmmmmmmmmmmmmmmmmmmm.
19:57:01 <ehird> bsmntbombdood: i kno
19:57:04 <ehird> it's just so silly
19:57:07 <ehird> pikhq: they're prolly using a crt
19:59:02 <ehird> bsmntbombdood: oh lawd
19:59:11 * ehird closes article, too retarded
19:59:29 <ehird> bsmntbombdood: also, I'm not sure I should get a 295
19:59:39 <ehird> bsmntbombdood: it's two gpus
19:59:50 <ehird> bsmntbombdood: it is probably more effective to use two of the best single nvidia gpus
19:59:55 <ehird> "Now the GTX 295 is a bit of a weird combo. See, it has the memory volume and frequency of two GTX 260 cards yet the raw shader processor horsepower of two GeForce GTX 280 cards. "
19:59:59 <ehird> gets more memory power too
20:00:07 <ehird> 280 is discontinued though
20:00:18 <ehird> 2x275 is at least as good as a 295
20:00:22 <ehird> and more effective for cooling etc
20:00:40 <bsmntbombdood> the sad thing is, with a one line command, i can have a ramdisk half the size of theirs that is much cheaper and a zillion times faster
20:01:40 <pikhq> /dev/shm seems to exist without me trying.
20:01:53 <ehird> bsmntbombdood: ooh, there's a better single gpu than the 275
20:02:00 <ehird> 2x285 >>>>>>>>>>>>>>>>>>>>>>>>>>>>> 295
20:02:04 <ehird> probably comparable power usage too
20:02:31 <ehird> DISADVANTAGE OF WATER COOLING: First it drains your pocket, then it allows you to drain your pocket.
20:02:46 <ehird> Then your pocket shrivels and dies.
20:03:52 <ehird> bsmntbombdood: on the 295: "you can rest assured that when these two GPU's are fully utilized the card can peak towards 289 watts power consumption"
20:04:54 <ehird> bsmntbombdood: if I get 2x285, you have to buy me crysis, okay?
20:05:02 <ehird> otherwise that's just a waste ;)
20:05:51 <bsmntbombdood> pikhq: mount -t tmpfs -o size=8G tmpfs /mnt/faster-than-you
20:06:15 <ehird> bsmntbombdood: 8G is a bit big if you only have 12G of ram
20:06:43 <ehird> bsmntbombdood: well sure but then you're left with only 4
20:06:49 <ehird> i'd recommend a 4G ramdisk
20:06:53 <ehird> so you have 8G left
20:06:57 <Deewiant> Only 4G of RAM! Whatever shall you do with so little
20:07:00 <pikhq> bsmntbombdood: Fair enough.
20:07:47 <bsmntbombdood> but if you are backing it up it might be easier to use a ram device
20:08:47 <bsmntbombdood> what's the differences between the buffers and cached line anyway?
20:09:26 <ehird> bsmntbombdood: kernel reserves a fuckton of memory for usage later
20:09:36 <ehird> buffers/cache line subtracts that from the usage
20:09:44 <ehird> so you can know what's actually being used
20:10:50 <ehird> bsmntbombdood: wait whut?
20:12:32 <ehird> bsmntbombdood: yep, 285 SLI kicks the shit out of a 295
20:12:37 <ehird> http://www.anandtech.com/video/showdoc.aspx?i=3501&p=2 start of benchmarks
20:12:41 -!- inurinternet has joined.
20:12:45 <ehird> even on ATI-lovin' games
20:13:00 <ehird> same as a 280 sli on that though
20:13:22 <ehird> http://images.anandtech.com/graphs/gtx285launch_011509005146/18001.png
20:13:28 <ehird> 95 FPS at 2560x1600 4xaa.
20:13:34 <ehird> That's slightly ridiculous.
20:13:40 <ehird> *scrolls to crysis*
20:14:04 <ehird> http://images.anandtech.com/graphs/gtx285launch_011509005146/17996.png
20:14:09 <ehird> that's just ridiculous
20:14:20 <ehird> ... although how come the gtx 280 beats the 285?
20:14:31 <ehird> Deewiant: I'm talking about the sli configuraitons
20:14:43 <Deewiant> ehird: The difference is so small I'd chalk it up to error
20:15:00 <ehird> Deewiant: ps your card only gets 18fps haha looooser
20:15:09 <bsmntbombdood> i'm not seeing an i7 motherboard that supports more than 24 gb of ram
20:15:14 <Deewiant> Like there's a difference between 22 and 18 :-P
20:15:40 <ehird> bsmntbombdood: i'm not seeing a consumer use-case that requires more than 24 gb of ram
20:15:58 <Deewiant> I'm not seeing a consumer use-case that requires more than 4 GB of RAM
20:16:00 <GregorR-L> ehird: Then you haven't seen the latest line of games!
20:16:11 <ehird> Deewiant: it's ok to be a little excessive
20:16:20 <bsmntbombdood> pikhq: oh, and that stupid sata ramdisk does have a battery good for about 4 hrs
20:16:26 <GregorR-L> RAMSuckerShootShootShoot alone requires 16GB
20:16:40 <ehird> GregorR-L: NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
20:16:59 <ehird> GregorR-L: does it require a quad-SLI of gtx 295s too?
20:17:02 <ehird> maybe a render farm on the side
20:17:58 -!- KingOfKarlsruhe has joined.
20:18:05 <ehird> bsmntbombdood: why not
20:18:18 <pikhq> bsmntbombdood: Heh.
20:18:26 <ehird> bsmntbombdood: lame
20:18:32 <ehird> quad-SLI of gtx 285s, then
20:18:46 <Asztal> but the Asus Mars can be quad SLI'd.
20:18:58 <ehird> "The GeForce GTX 295 comes with a 289 watt TDP (peak wattage). Since your motherboard can only deliver 75 watts, you'll need to hook it up to a PSU"
20:19:03 <ehird> Reaaaaaaaaaaaaaaaaaaaaally?!!?!?!?237836248761238123
20:19:22 <ehird> Asztal: ah, it's two gtx 285s?
20:20:01 -!- Corun has changed nick to Corun|away.
20:20:45 <Deewiant> ehird: Most new GPUs these days need to be hooked up to a PSU.
20:20:52 <ehird> Deewiant: thus my sarcasm
20:21:02 <ehird> bsmntbombdood: what we need is thin clients + 5 billion lines of fiber-optics to communicate w/ the thick servers
20:21:12 <ehird> cooling vs silence: SOLVED.
20:22:25 <ehird> bsmntbombdood: yeah i could totally not hear a 4x asus mars sli + 2x nehalem xeons @ 4ghz + 2x 1200watt psu in another room
20:22:33 <ehird> everyone knows that walls block all noise entirely
20:22:46 -!- tombom_ has joined.
20:22:59 <GregorR-L> This picture rated X for violence.
20:23:11 <ehird> this picture rated x for xtrem
20:24:21 <ehird> this picture rated p for people having intercourse
20:26:30 <ehird> this is what 137dpi looks like:
20:26:37 <ehird> http://imggur.com/RAORj.png
20:26:40 <ehird> http://imgur.com/RAORj.png
20:27:12 <ehird> that's what 164 dpi looks like
20:27:16 <ehird> Resize picture-viewing window to be 10" :P
20:29:35 <ehird> if you have a 96dpi monitor,
20:29:48 <ehird> 750x600 is the right size
20:30:05 <ehird> GregorR-L: See my note about resizing.
20:31:14 <GregorR-L> That had nothing to do with any of that, I was just poking your eyes out so you can see what 0 dpi looks like.
20:31:26 <ehird> Very clever. Har har.
20:32:19 <ehird> bsmntbombdood: Wut?
20:32:25 <ehird> bsmntbombdood: and yeah
20:32:30 <ehird> but the fans needed to cool that shit = huge
20:34:13 <ehird> GRRRRRRRRRRRRREEEEEEEEEEEEEEEEN
20:35:20 <pikhq> Þis picture rated Þ for gratuitous uſe of þorn.
20:35:31 <ehird> gratuitous amounts of pornography
20:37:33 <ehird> bsmntbombdood: new cooling system
20:37:35 <GregorR-L> Dear me, I say, the pornography in this picture would have been acceptable 10 minutes ago, but now it's just gratuitous.
20:37:42 <ehird> your system goes in a thin box
20:37:44 <ehird> and is moved to finland
20:37:49 <ehird> cables are connected to it underwater
20:38:00 <ehird> or perhaps antarctica
20:38:04 <ehird> latency's lower from finland
20:38:29 <pikhq> GregorR-L: However, it is alſo gratis. Surely þat's important‽
20:38:56 <ehird> one day I'm going to start talking in real, bona-fide olde english
20:38:59 -!- inurinternet has quit (Connection timed out).
20:39:02 <ehird> and you're all going to be FUCKED
20:39:15 -!- inurinternet has joined.
20:39:40 -!- tombom has quit (Read error: 110 (Connection timed out)).
20:39:40 -!- tombom_ has changed nick to tombom.
20:39:53 <GregorR-L> It'll be difficult for you to say "esoteric programming language" in "bona-fide" Old English.
20:40:20 <ehird> GregorR-L: esoterice language of the magicke movinge stones
20:42:32 <ehird> Hwæt! wē Gār-Dena in geār-dagum, þeod-cyninga, þrym gefrunon, hu ða æþelingas ellen fremedon.
20:42:44 <AnMaster> lifthrasiir, Adding polynoms both made the generated code better and worse at the same time... Trying to fix back the cases where it is now worse
20:43:05 <AnMaster> mostly due to other optimisation passes knowing more tricks for the old nodes...
20:44:46 <ehird> bsmntbombdood: i am jealous of you. you and your noise-carelessness.
20:44:52 <ehird> if only I could be that desensitive :P
20:45:29 <ehird> wait, the reserator pushes 300l/hr not 30l/hr :-D
20:45:49 <AnMaster> <ehird> Hwaet! wē Gār-Dena in geār-dagum, theod-cyninga, thrym gefrunon, hu ða aethelingas ellen fremedon. <-- care to translate
20:46:14 <ehird> AnMaster: "Listen! We have heard of the glory of the Spear-Danes, of the kings of the people, in days of yore, [and] how those princes did deeds of glory."
20:46:27 <ehird> From Beowulf: http://en.wikipedia.org/wiki/Old_English_language#Beowulf
20:46:40 <AnMaster> I'm clearly going to have to ask for translation a lot from now on...
20:47:15 <ehird> bsmntbombdood: a good one? Uh, 0dB.
20:47:27 <ehird> Probably bad ones too.
20:47:32 <ehird> It's pumps that make noise.
20:47:53 <ehird> bsmntbombdood: I mean, it does nothing.
20:47:58 <ehird> It doesn't push anything, it just is.
20:48:06 <ehird> bsmntbombdood: Yes, that's what I'm going to do
20:48:09 -!- Corun|away has changed nick to Corun.
20:48:13 <ehird> By mounting it on the wall of the stairs
20:48:14 <AnMaster> <ehird> http://imgur.com/RAORj.png <--- the image metadata says it is 72 dpi...
20:48:22 <ehird> AnMaster: That's irrelevant; what dpi is your display?
20:48:31 <ehird> I'll tell you the appropriate resolution
20:48:36 <AnMaster> ehird, 86 iirc. 80-something anyway
20:48:50 <AnMaster> ehird, and I'm using gimps "not pixel for pixel" mode
20:49:04 <AnMaster> so if metadata was correct it would show as right size
20:49:09 <ehird> AnMaster: Set gimp to a-pixel-is-a-pixel, then resize it to 671x600.
20:49:21 <AnMaster> <ehird> AnMaster: Set gimp to a-pixel-is-a-pixel, then resize it to 671x600. <-- doing it that way is broken :P
20:49:34 <ehird> And it'll look right (modulo blurriness due to your monitor's dpi being lower than what the image was designed for)
20:49:44 <ehird> AnMaster: Blame ubuntu's screenshot too
20:50:18 <AnMaster> <ehird> AnMaster: Set gimp to a-pixel-is-a-pixel, then resize it to 671x600. <-- that doesn't keep the aspect ratio.
20:50:28 <ehird> AnMaster: Just do it, I know it'll work :P
20:50:36 <ehird> I worked it out and everything
20:50:48 <ehird> I think I got it the wrong way around
20:51:07 <ehird> AnMaster: 671.54 pixels TIMES something.
20:51:19 <ehird> AnMaster: For a more clarcut one, make it 600 second-dimension.
20:51:24 <ehird> And let it choose the first dimension.
20:51:32 <ehird> For perfect results, get out a ruler :P
20:51:44 <AnMaster> ehird, for perfect results use correct metadata :P
20:52:06 <ehird> Get it to as close as that as it'll be happy with
20:52:18 <ehird> 20:50 AnMaster: for 671 gimp suggests 537
20:52:36 <AnMaster> ehird, that is what I said -_-
20:52:40 <ehird> AnMaster: With linear interpolation
20:52:50 <ehird> Does best for scaling down text and stuff.
20:53:02 <ehird> Although at such a low dpi I'm not sure you could notice :P
20:53:51 <AnMaster> ehird, that is a very small screen!
20:54:03 <ehird> AnMaster: Measure its diagonal size
20:54:12 <ehird> And yeah, 10" is small. Netbook-size.
20:54:36 <ehird> AnMaster: But it'd be a lot better on the actual thing.
20:54:40 <ehird> It's for a laptop; you have it a lot closer.
20:54:44 <ehird> It's high-dpi, so it's sharper.
20:54:46 <AnMaster> ehird, are you still getting that speed monster too
20:54:52 <ehird> AnMaster: That's the main priority
20:55:03 <ehird> The netbook requires a sufficiently high DPI OLED display; so come back in a year or something
20:55:17 <ehird> AnMaster: The speed monster is what I'm watercooling w/ that gigantic passive radiator.
20:56:13 <AnMaster> ehird, the pump will still be noisy
20:56:27 <ehird> AnMaster: The 5 watt pump in the Reserator is inaudible.
20:56:37 <ehird> Even moreso if you add a drop of washing up liquid (seriously)
20:56:47 <ehird> AnMaster: Also, I'm mounting it on the wall of another room.
20:56:51 <ehird> (The stairs opposite this room.)
20:58:08 <ehird> bsmntbombdood: btw, the same basic idea as watercooling is used in heatpipe heatsinks
20:58:15 <ehird> zalman totally no noise is basically watercooling sans the water
20:58:20 <ehird> with the case as the radiator
20:59:30 <ehird> bsmntbombdood: howso
21:00:07 <ehird> also bsmntbombdood uh no
21:00:09 <ehird> that's a loop heat pipe
21:00:35 <ehird> bsmntbombdood: regular heat pipes are single-phase
21:02:17 <ehird> wikipedia said it therefore it is true.
21:02:39 <ehird> wait, that was talking about another thing
21:02:45 <ehird> even so, heatpipes != phase
21:05:54 <ehird> i wonder how many watercooled psus there are
21:06:58 <ehird> bsmntbombdood: well you just need a fanless psu
21:07:02 <ehird> and put a waterblock on it
21:07:19 <bsmntbombdood> http://www.koolance.com/water-cooling/product_info.php?product_id=665
21:07:27 <ehird> bsmntbombdood: that one's submerged
21:07:39 <ehird> i hate submersion :)
21:07:52 <ehird> bsmntbombdood: also, why bad idea
21:07:56 <ehird> it's just like cooling a hot cpu
21:08:26 <ehird> there are psus designed to run fanlessly
21:08:37 <ehird> putting a waterblock on them would run fine
21:10:18 <ehird> Silverstone's max fanless psu is 450w
21:10:30 <ehird> lame as a lame thing
21:11:10 * ehird looks around some more
21:11:22 <ehird> bsmntbombdood: dunno, just don't like it
21:11:59 <ehird> bsmntbombdood: also, that psu has way too high wwattage
21:12:12 <ehird> doesn't even look to be 80% efficient
21:13:39 * ehird calculates power usage for his rig-o-matic
21:15:18 <ehird> 344watt @ 85% cpu tdp, 90% load, 0 capacitor aging
21:15:33 <ehird> w/ 100% tdp, 100% load 402w
21:15:53 <ehird> that but w/ 20% capacitor aging = 482w
21:16:20 <ehird> if we say gtx 295 instead of 285, and same settings as last, 566w
21:16:29 <ehird> so i only need 500w
21:17:08 <ehird> bsmntbombdood: so 1700w is amazingly wasteful&excessive.
21:17:14 <ehird> for, well... anything.
21:23:31 <ehird> bsmntbombdood: 'pparently with a zalman reserator you still need one low-rpm case fan
21:23:33 <ehird> i don't see why though
21:23:41 <ehird> as long as you put absolutely everything that makes any heat into the loop
21:24:43 -!- inurinternet has quit (Read error: 113 (No route to host)).
21:26:39 <ehird> bsmntbombdood: any idea why?
21:36:40 <bsmntbombdood> you can't put everything that makes heat in the loop
21:36:50 <ehird> bsmntbombdood: why not
21:36:52 <bsmntbombdood> or you end up like this: <please hold while i google>
21:36:55 <ehird> northbridge, ram, gpu, ...
21:36:59 <ehird> perfectly possible.
21:38:57 <ehird> bsmntbombdood: your google is slow
21:38:59 <ehird> you should grease it
21:40:04 <ehird> bsmntbombdood: what was it?
21:40:26 <bsmntbombdood> a setup where he put a waterblock on every single heat producing component
21:41:06 <bsmntbombdood> http://forums.overclockers.com.au/showthread.php?t=744914
21:41:36 <ehird> bsmntbombdood: well i don't mean everything that produces heat
21:41:45 <ehird> everything that produces heat that would actually need a heatsink or airflow of any kind
21:41:58 <ehird> i really don't see why you'd need a case fan
21:42:45 <ehird> bsmntbombdood: northbridge, gpu, gpu ram, cpu, ram, psu
21:42:49 <ehird> and you've covered most everything
21:43:08 <bsmntbombdood> ehird: southbridge too, and the power circuitry for the cpu
21:43:22 <ehird> bsmntbombdood: do those make enough heat to need any sort of cooling?
21:43:26 <ehird> i'd imagine the heat would just waft off
21:43:40 <ehird> bsmntbombdood: leave that on, then
21:43:44 -!- psygnisf_ has quit ("Leaving...").
21:43:53 <ehird> bsmntbombdood: put a waterblock on, then
21:44:04 <ehird> bsmntbombdood: btw.
21:44:07 <ehird> yes they need airflow
21:44:13 <ehird> but your case almost certainly has some holes in it
21:44:23 <ehird> i highly doubt it needs more than a trickle of air moving about aimlessly
21:44:40 <ehird> yes. they're quiet. but if you're gonna get a fan, you might as well aircool
21:45:07 <ehird> 1 psu fan, 1 cpu fan, 1 gfx card fan, 1 intake, 1 exhaust = 5 fans; if you get really good 120mm fans, barely more audible than 2 fans
21:45:07 <bsmntbombdood> http://www.overclockers.com.au/images/monolith/DSC02981-600.jpg
21:45:14 <bsmntbombdood> "The top of the pump after running for a few hours got really damn hot, so i decided to cool it."
21:45:25 <ehird> the attraction of watercooling, to me, is running without any fans.
21:45:42 <ehird> bsmntbombdood: the pump's in a cold room
21:45:56 <ehird> bsmntbombdood: and as I said:
21:45:58 <bsmntbombdood> http://www.overclockers.com.au/images/monolith/DSC03035-600.jpg
21:45:59 <ehird> 21:44 ehird: but your case almost certainly has some holes in it
21:45:59 <ehird> 21:44 ehird: i highly doubt it needs more than a trickle of air moving about aimlessly
21:46:09 <bsmntbombdood> that's what happens when you try to watercool everything
21:46:21 <ehird> i'm pretty sure you could just poke out the exhaust fan grill on the back of your case
21:46:25 <ehird> and let the air flow by itself
21:46:32 <ehird> for such minor heatsources as what you said
21:46:56 <ehird> bsmntbombdood: no?
21:47:13 <ehird> bsmntbombdood: i'm putting my data hd in a fanless enclosur
21:47:21 <ehird> also, you can buy HD waterblocks
21:48:00 <AnMaster> <ehird> AnMaster: Also, I'm mounting it on the wall of another room. <-- heh
21:48:04 <ehird> bsmntbombdood: i'm reading it
21:48:22 <ehird> bsmntbombdood: right well, you can buy 'em
21:48:24 <AnMaster> ehird, I want 100% passive cooling.
21:48:31 <ehird> AnMaster: zalman totally no noise
21:48:35 <ehird> AnMaster: but the pump is inaudible
21:48:38 <ehird> way more so than any fan
21:48:47 <ehird> AnMaster: a reserator system can have no fan
21:48:49 <ehird> I'm almost certian
21:49:00 <ehird> AnMaster: and yes, zalman tnn = case is radiator, heatpipes
21:49:10 <ehird> AnMaster: giant radiator + reservoir + pump
21:49:23 <ehird> AnMaster: it's made by Zalman
21:49:40 <AnMaster> ah so that explains why all google results seemed to be about a specific product
21:49:42 <ehird> AnMaster: disadvantage is that it can't dissipate too much heat (so put two of them together), and it heats the room
21:49:56 <AnMaster> ehird, all cooling systems heat the room
21:50:10 <ehird> bsmntbombdood: YOU CAN REPEAT AnMaster
21:50:19 <ehird> and "moreso" didn't mean that
21:50:32 <ehird> but fact is, an air-based system won't be a gigantic space heater like the reserator
21:50:36 <AnMaster> ehird, what about only heatpipes...
21:50:48 <ehird> that's what totally no noise does. it cannot cool anywhere near as much as WC
21:50:59 <AnMaster> ehird, of course it can't cool as much!
21:51:02 <ehird> AnMaster: go forwards a bit from where my computer will be and you reach a wall. drill some holes through that wall. mount the two reserators w/ brackets on the wall. feed pipes through
21:51:18 <ehird> bsmntbombdood: no, it won't
21:51:20 <AnMaster> ehird, WC can cool below ambient temperature iirc
21:51:21 <ehird> (assertion match fuck yeah)
21:51:25 <ehird> AnMaster: it cannot
21:51:34 <ehird> and that isn't about how much heat the radiator lets out
21:51:36 <ehird> that's cpu temperatures
21:51:38 <AnMaster> Oh wait I'm mixing it up with compressor
21:52:04 <ehird> the hardest part of mounting the reserators
21:52:07 <ehird> is that they're 8kg when filled
21:52:17 <bsmntbombdood> do you think a radiator just magically creates heat somehow?
21:52:26 <ehird> you're not reading what I'm saying
21:52:29 <AnMaster> ehird, you need special support structures then
21:52:32 <GregorR-L> <bsmntbombdood> ANY COOLING SYSTEM HEATS THE ROOM // you could refrigerate the room and redirect the heat outside.
21:52:36 <ehird> AnMaster: yes, brackets.
21:52:39 <bsmntbombdood> then how will it heat the room more than air cooling
21:52:48 <AnMaster> GregorR-L, sure. But that isn't relevant!
21:52:53 <ehird> bsmntbombdood: it won't. the point is the concentration in one space, and the lack of moving air
21:53:24 <AnMaster> GregorR, I never said that the look of my face was relevant to this discussion...
21:53:42 -!- inurinternet has joined.
21:54:28 <ehird> (two reserators mounted above my head in an unused staircase)->coolness
21:54:30 <ehird> >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
21:54:36 <ehird> (two reserators sitting in my room)->coolness
21:56:03 <ehird> bsmntbombdood: so nyah.
21:56:41 <AnMaster> oerjan, http://www.irregularwebcomic.net/archive.html is messed up too
21:57:18 <AnMaster> ehird, got a good pic of that reserator?
21:57:28 <ehird> AnMaster: yep, sec
21:57:40 <ehird> AnMaster: http://www.zalman.co.kr/DataFile/product/RESERATOR-1-V2_01_b(0).jpg
21:57:48 <ehird> it's the big metal tower with the fins, in case you're confused. :p
21:58:10 <ehird> (that * 2) + wall-mounting + drilled holes in walls = silent, heatless water cooling from my position
21:59:46 -!- inurinternet has quit.
22:00:04 -!- inurinternet has joined.
22:00:12 <ehird> 'part from coil whine
22:02:52 <ehird> http://www.feser-one.com/site/popup_image2.php?pID=322
22:02:53 <ehird> 18.75"x5.85"x4.10" (LxWxH)
22:02:57 <ehird> dissipates 2000 watts
22:03:18 <oerjan> AnMaster: you mean that empty space at May 16, pushing everything out?
22:03:28 <ehird> bsmntbombdood: err
22:03:31 <ehird> it dissipates 2000 watts
22:03:43 <ehird> bsmntbombdood: as in, your system can go to 2000 watts of heat
22:03:50 <ehird> and it'll dissipate it into the air successfully
22:03:54 <ehird> a reserator manages 200w
22:04:17 <ehird> the 8kg is 6kg reserator + 2.5 litres of water
22:04:30 <ehird> bsmntbombdood: ... do you understand what i'm saying?
22:04:56 <AnMaster> ehird, how do you top up the water in it...
22:05:07 <ehird> AnMaster: you don't really have to top it up
22:05:12 <ehird> but to replace... drain it
22:05:34 <oerjan> AnMaster: hm, 2318 shows up in the correct place in the themed archive
22:05:35 <ehird> bsmntbombdood: 2000 watts of heat = 2000 watts of heat
22:05:36 <AnMaster> <ehird> run it for a while <-- ?
22:05:40 <ehird> a radiator can't magically remove heat
22:05:52 <ehird> AnMaster: if you turn on your computer while the tubes are empty it could overheat
22:05:59 <ehird> before the water gets to it
22:06:00 <oerjan> under Steve and Terry, as expected
22:06:02 <AnMaster> oerjan, I'm actually suspecting this isn't intentional any more...
22:06:06 <ehird> so you run just the watercooling for a while
22:07:50 <ehird> bsmntbombdood: but for a regular room I imagine dissipating 2000 watts would make it burn you if you touch it
22:08:01 <ehird> or at least sting a lot
22:08:06 <ehird> can't be more precise than that :)
22:08:30 <ehird> bsmntbombdood: it's the radiator heating the air around it
22:08:36 <ehird> no change to cpu temps.
22:08:38 <oerjan> 2000 wasps, on the other hand...
22:08:46 <ehird> it's just where the hot water goes...
22:08:56 <bsmntbombdood> if the radiator is hot enough to burn, the water going through it is, and so the cpu is too
22:09:13 <ehird> bsmntbombdood: well
22:09:20 <ehird> some cpus give up to like 900 watts of heat
22:09:26 <ehird> so i may be overestimating
22:09:35 <ehird> maxing out a 2,000 watt radiator would be hard
22:09:40 <ehird> it's just how much it can theoretically disperse
22:09:45 <ehird> bsmntbombdood: according to a quick google
22:09:55 <ehird> but anyway, it's just a theoretical maximum
22:10:38 <ehird> think the person saying it was a retard :)
22:10:46 <bsmntbombdood> you can move that much heat through something the size of a cpu heatspreader
22:11:25 <ehird> who cares about how much heat it can dissipate as long as it's enough
22:13:56 <ehird> bsmntbombdood: but ofc, if you're dissipating 2000 watts you'd better fucking have a fan on the radiator
22:15:14 -!- tombom has quit ("Peace and Protection 4.22.2").
22:22:32 <ehird> "Project wants strong AI within 8 years in giant distributed system and your computer could be a part of it."
22:22:38 <ehird> Gee, I sure hope it's friendly :P
22:23:20 <oerjan> oh i'm sure my computer _would_ be a part of it. whether i signed up or not. :D
22:23:37 <ehird> Will this project make public some of its source code?
22:23:38 <ehird> Will this project describe how knowledge is represented in the brain?
22:23:54 <ehird> "Let's completely revolutionize humanity. But first, PROFIT!!!!!"
22:24:16 <ehird> Why not make public the mechanisms of knowledge representation in the brain?
22:24:16 <ehird> Security is our main concern. As long as this concern is not addressed, it is in everyone's best interest that certain knowledge is restricted.
22:25:11 -!- kar8nga has quit (Remote closed the connection).
22:25:40 <ehird> [[We intend to build a system that has the brain capacity of all people living in the world. This way we can ensure that no one will be left behind.]]
22:25:43 <ehird> NO BRAIN LEFT BEHIND
22:30:42 <myndzi> they want to make a psycho MPD ai?
22:30:43 <ehird> bsmntbombdood: i wonder if it holds that since a reserator can handle a bit over 200w of heat, two of them can handle a bit over 400
22:30:54 <ehird> there are 6 core old xeons
22:30:59 <ehird> (old = core 2 days, iirc)
22:31:23 <ehird> bsmntbombdood: hmm then 400 watts should be enough to cool i7 + gtx 295
22:32:10 <ehird> bsmntbombdood: erm, for values of taking the heatsink off
22:32:23 <ehird> bsmntbombdood: yeah, so I gotta detach them
22:32:28 <ehird> AnMaster: stupid noname company
22:32:33 <ehird> AnMaster: see reddit
22:32:37 <ehird> bsmntbombdood: big deal
22:32:48 <ehird> bsmntbombdood: other option: 2x285 or just 1x285
22:33:20 <ehird> bsmntbombdood: also far more expensive
22:33:26 <ehird> 285~=$400 295=$500
22:33:43 <ehird> bsmntbombdood: should be fine
22:33:52 <ehird> as i said, reserator can handle a bit above 200, let's say 210
22:34:05 <ehird> so it could probably handle it
22:34:12 <ehird> (410 being an underestimate)
22:34:21 <ehird> bsmntbombdood: or I could just get that massive radiator)
22:34:32 <ehird> bsmntbombdood: or I could get 3xreserators
22:34:42 <ehird> bsmntbombdood: nothx
22:35:08 <ehird> bsmntbombdood: also, companies tend to overestimate
22:35:12 <AnMaster> +os("\n\nYou are standing at a junction. The path north continues along the sheer\ncliff and leads to a high plateau. Directly south you can see the tangled\ntrees an
22:35:12 <AnMaster> d vines of a small forest. In the distance, to the west, you can see\nthe small village that you now call home.\n");
22:35:18 <ehird> it'll actually be like 390, I imagine
22:35:31 <AnMaster> the spurious line break due to using less
22:35:44 <ehird> bsmntbombdood: lemme look at reserator's manual
22:36:54 <ehird> bsmntbombdood: cuz it's tiny
22:37:11 <ehird> people like having space.
22:37:23 <ehird> most people aren't enthusiasts, or are buying a non-main pc
22:38:08 <ehird> oh wtx is bigger atx
22:38:12 <ehird> AnMaster: eATX nowadays
22:38:32 <AnMaster> I like the name "workstation ATX" though
22:38:41 <ehird> bsmntbombdood: so buy an eatx mobo
22:38:45 <AnMaster> ehird, always liked the word "workstation"
22:39:11 <ehird> bsmntbombdood: this manual is shit, it doesn't tell you how much heat it can dissipate
22:39:26 <AnMaster> bsmntbombdood, "Sun Workstation". From the 80s. Sound so nice
22:39:30 <ehird> bsmntbombdood: but i really don't want to diy :P
22:39:39 <ehird> (having said that it is now inevitable, such is my life)
22:39:40 <KingOfKarlsruhe> yeah, now you can execute bf programs in my counterstrike source server
22:39:49 <ehird> KingOfKarlsruhe: Uh, wonderful.
22:39:51 <ehird> I so want to do that.
22:39:57 <ehird> "hey fagut u suk" "wel ,[.,] to u"
22:40:09 -!- inurinternet has quit (Read error: 60 (Operation timed out)).
22:40:49 <oerjan> all our counters are on strike
22:41:14 <ehird> bsmntbombdood: more fuss & the reserator looks pretty & mounting a separate rad/pump would sort of be a pain and also feels more like just putting the whole computer there
22:41:24 <AnMaster> oerjan, that explains why this is a while loop instead of a for one!
22:41:51 <ehird> "The Reserator V2 must be used with the provided G200 coolant. The use of other coolants with the Reserator V2 can cause damage to the Reserator V2 unit, as well as the CPU and VGA jackets."
22:42:01 <ehird> tons of people don't use the stock coolant :)
22:42:16 <bsmntbombdood> 1 megapixel display, 1 megabyte of ram, and 1 megabit of network
22:42:34 <AnMaster> bsmntbombdood, and a whopping 25 MHz CPU!
22:42:41 <ehird> bsmntbombdood: exactly, I don't want to do that
22:43:01 <ehird> putting a reserator makes me think "well let's just put the computer there" less than a separate pump/rad
22:43:51 <AnMaster> bsmntbombdood, up for some compiling again?
22:44:17 <ehird> AnMaster: buy your own i7 :)
22:44:31 <ehird> AnMaster: what‽ you would support EVIL MARKETING‽‽‽‽‽‽‽‽‽
22:44:39 <ehird> I am shocked, shocked
22:44:43 <AnMaster> http://rage.kuonet.org/~anmaster/LostKng.b.c.gz
22:44:49 <AnMaster> ehird, I wish I had that money...
22:44:49 <ehird> that's no benchmark
22:44:56 <ehird> even a naïve interp can start it fast
22:45:16 <ehird> AnMaster: use clang
22:45:21 <AnMaster> bsmntbombdood, yes my compiler is much better now
22:45:30 -!- oerjan has quit ("Good night").
22:45:39 <AnMaster> bsmntbombdood, -march=k8-sse3 -O2 ?
22:45:41 <ehird> bsmntbombdood: -march=i7 just to make him suffer :)
22:46:18 <AnMaster> ehird, and I pretty much hit the limit to what I can do to mandelbrot. A few more things. But that is it
22:46:37 <bsmntbombdood> AnMaster: it's only using a gigabyte of ram right now
22:46:44 <AnMaster> ehird, well what I can see "AHA here we can optimise"
22:47:03 -!- inurinternet has joined.
22:47:19 <AnMaster> bsmntbombdood, I do polynomials now too
22:48:40 <ehird> bsmntbombdood: i think the best solution if it's >420w heat output is 3 reserators
22:48:51 <ehird> could handle >600W of heat
22:49:18 <ehird> bsmntbombdood: i don't want to build my own
22:49:24 <ehird> AnMaster: miscalculated
22:49:38 <ehird> bsmntbombdood: a dollar per watt
22:49:43 <ehird> pretty good value ;)
22:51:03 <AnMaster> bsmntbombdood, currently I have lots to do on the new polynomial code... Since lots of other passes doesn't fully understand it...
22:51:15 <AnMaster> but this is enough to make sure that it runs
22:52:28 <ehird> bsmntbombdood: i'm sure that a reserator can dissipate more than 200W
22:52:43 <ehird> iirc it can do most things under 300W, someone said
22:52:50 <ehird> so if we say 250, 2 of them can dissipate 500W
22:52:55 <ehird> which is fine for a 420W heat system
22:54:05 <ehird> bsmntbombdood: ha, apparently some of the best standalone radiators only do like 180-580w
22:54:25 <ehird> bsmntbombdood: no fans. at all.
22:54:40 <ehird> the only non-passive components allowed is an HD and a pump
22:54:48 <ehird> bsmntbombdood: perfectly possible
22:54:59 <ehird> as long as I have a hole in the case for minimal airflow for the few minimal heat components left
22:55:08 <ehird> → just take out the exhaust fan grill
22:56:31 <AnMaster> bsmntbombdood, you could WC cool the HD maybe
22:57:21 <bsmntbombdood> hdds are more than sufficiently cooled with a slow fan
22:57:36 <AnMaster> bsmntbombdood, that isn't passive
22:58:01 <AnMaster> bsmntbombdood, yes. I want heatpipes + ssd
23:09:31 <AnMaster> bsmntbombdood, memory controller too
23:09:49 <AnMaster> but it needs to connect to memory
23:11:44 <AnMaster> then there is power, and various other things
23:15:04 -!- olsner has quit ("Leaving").
23:15:33 -!- nooga has joined.
23:16:07 <nooga> AnMaster: jag hittar en himmelsk drog
23:16:33 <AnMaster> bsmntbombdood, it is broken Swedish
23:16:41 <GregorR-L> IF ENGLISH WAS GOOD ENOUGH FOR JESUS, IT'S GOOD ENOUGH FOR #ESOTERIC
23:17:17 <nooga> AnMaster: smygande
23:17:20 -!- darthnuri has joined.
23:17:31 <Slereah_> Plain English is the best language ever
23:17:42 <GregorR-L> ehird: In case you didn't notice, I made egojoust use all lengths and polarities.
23:17:46 <nooga> except for that English is unplain
23:18:33 <AnMaster> or have he stopped using the silly script?
23:18:46 -!- coppro has joined.
23:18:52 -!- inurinternet has quit (Connection timed out).
23:19:08 <GregorR-L> AnMaster: It started bothering me at least as much as everyone else :P
23:19:15 <nooga> my English is totally broken, but idc
23:22:08 <FireFly> [00:16:07] <nooga> AnMaster: jag hittar en himmelsk drog
23:22:15 <FireFly> Wouldn't that be correct swedish?
23:22:30 <AnMaster> FireFly, well, just strange I guess
23:22:40 <AnMaster> "hittade" would make more sense
23:22:41 <GregorR-L> Bork bork bork spaghetti bork bork bork.
23:22:57 <nooga> AnMaster: it's from a song by Kent
23:23:35 <FireFly> Couldn't remember the english term :P
23:23:52 <nooga> http://www.youtube.com/watch?v=V9d3JqXr7fg that shit
23:23:56 <AnMaster> FireFly, What is the Swedish term for it now again
23:24:00 <nooga> my brother is addicted to them
23:25:30 -!- FireFly has quit ("Later").
23:26:53 <nooga> AnMaster: http://www.youtube.com/watch?v=irnr0HPt5f0 this is quite nice
23:36:50 <ehird> 23:17 GregorR-L: ehird: In case you didn't notice, I made egojoust use all lengths and polarities. ← yes, so I stopped working on my interp
23:37:11 <ehird> 22:58 AnMaster: bsmntbombdood, yes. I want heatpipes + ssd ← tnn, stfu
23:37:54 <ehird> GregorR-L: uh cuz yours does what mine was gonna?
23:38:05 <ehird> AnMaster: zalman totally no noise
23:38:09 <ehird> heatpipes + case is radiator
23:38:11 <GregorR-L> Yours was going to have proper expansionlessness :P
23:38:17 <ehird> it can't run a non-low-powered system
23:38:24 <ehird> so stop obsessing over it
23:38:49 <ehird> watercooling is passive modulo pump
23:38:51 <AnMaster> ehird, consider a recording studio
23:39:12 <ehird> AnMaster: recording studio people use quiet air cooled and watercooled pcs
23:39:16 <ehird> AnMaster: do you know how silent a pump is?
23:39:33 <ehird> inaudible to the human ear from just a tiny distance away, that's how queit
23:39:40 <ehird> quieter than any fan
23:40:29 <GregorR-L> The pumps built into the human body are nearly silent while inhabiting that very human body.
23:40:42 <GregorR-L> (Like how useless my commentary is?)
23:40:55 <ehird> AnMaster: space? what?
23:41:08 <ehird> AnMaster: what about it
23:41:10 <ehird> so use a pump instead
23:41:20 <ehird> so use a pump instead
23:41:25 <ehird> but seriously, what?
23:41:37 <ehird> AnMaster: space is 4 kelvin
23:41:40 <AnMaster> they use heatpipes and pel-wahtever-elements a lot of he the time
23:41:42 <ehird> good job needing cooling in that
23:41:53 <ehird> you don't mean directly in space? gee, then your argument's rubbish
23:41:55 <AnMaster> ehird, um... the sunny side gets very hot
23:42:14 <AnMaster> ehird, so "you don't need cooling" is very wrong
23:42:26 -!- Judofyr has quit (Read error: 104 (Connection reset by peer)).
23:42:39 <ehird> bsmntbombdood: gtx 285 = 204W
23:42:41 <AnMaster> ehird, vacuum is a good insulator too.
23:42:51 <ehird> bsmntbombdood: = 344W total w/ i7
23:43:01 <AnMaster> well not perfect vacuum of course
23:43:02 <ehird> bsmntbombdood: 2 reserators would handle that and yawn as a sideproject
23:43:48 <ehird> i admit, though, that it would be nice to be able to handle a gtx 295.
23:44:27 <ehird> ati cards tend to be a little less hot iirc
23:44:32 <AnMaster> ehird, I thought they called them "geforce 9000" and such iirc
23:44:41 <ehird> AnMaster: up to the 9800gtx+
23:44:50 <ehird> but their high end gaming cards are named gtx 2??
23:45:00 <ehird> gtx 260, gtx 275, gtx 285, gtx 295
23:45:02 <ehird> is the current range, iirc
23:45:04 <AnMaster> ehird, great... I remember when it was simple Geforce 3 Geforce 4a
23:45:17 <ehird> wait, there's also a gts 250
23:45:23 <ehird> the s presumably stands for shitty
23:45:28 <ehird> but they list it with the gtxes
23:46:05 <ehird> bsmntbombdood: i like how this started with "stfu the reserator pwns" and is now "dude the reserator sucks diy it"
23:49:56 <ehird> 19:25 ehird: http://www.zalman.co.kr/DataFile/product/RESERATOR-1-V2_01_b(0).jpg Combined pump, reservoir and... space heater.
23:49:58 <ehird> 19:25 bsmntbombdood: ehird: stfu, that thing's awesome
23:50:08 <ehird> 22:49 bsmntbombdood: build your own
23:51:03 <ehird> bsmntbombdood: i have a portable radiator/space heater behind me. think i should rip out the radiator and do it ghetto-style? :D
23:53:06 <ehird> bsmntbombdood: THAT WAS A JOKE :(
23:53:14 <ehird> bsmntbombdood: it probably can't even radiate much
23:53:47 <ehird> bsmntbombdood: ...right?
23:54:08 <ehird> i mean, it only has 3 heat settings.
23:58:21 <ehird> bsmntbombdood: i thought you said horse
23:58:34 <ehird> bsmntbombdood: also, um, that's kind of expensive
23:58:41 <ehird> i don't want to be refilling it every day :D
23:59:02 <ehird> it also, bsmntbombdood, begs the question of what to hose it on to.
23:59:34 <ehird> bsmntbombdood: yer joking... right?
23:59:52 <ehird> bsmntbombdood: i mean in general
23:59:55 <ehird> with the whole idea of using a hose.