Forum logs for 14 Aug 2017

Monday, 16 March, Year 12 d.Tr. | Author:
BingoBoingo: ^ Oh noes, he's gone again! [00:30]
deedbot: http://phuctor.nosuchlabs.com/gpgkey/06DF269F44B2F8C4AAEDA67C6A687B89D89B4535D07913078504B89CABF0FE6A << Recent Phuctorings. - Phuctored: 1348...7817 divides RSA Moduli belonging to '178.188.248.230 (ssh-rsa key from 178.188.248.230 (13-14 June 2016 extraction) for Phuctor import. Ask asciilifeform or framedragger on Freenode, or email fd at mkj dot lt) <ssh...lt> ' (Unknown AT) [00:47]
deedbot: http://phuctor.nosuchlabs.com/gpgkey/06DF269F44B2F8C4AAEDA67C6A687B89D89B4535D07913078504B89CABF0FE6A << Recent Phuctorings. - Phuctored: 1624...1809 divides RSA Moduli belonging to '178.188.248.230 (ssh-rsa key from 178.188.248.230 (13-14 June 2016 extraction) for Phuctor import. Ask asciilifeform or framedragger on Freenode, or email fd at mkj dot lt) <ssh...lt> ' (Unknown AT) [00:47]
deedbot: http://phuctor.nosuchlabs.com/gpgkey/EBB54B7022BA30DFD198B846506BFA051F02DCF89003A0C9917EF6AD8E2CD9DD << Recent Phuctorings. - Phuctored: 1516...4019 divides RSA Moduli belonging to '92.243.14.54 (ssh-rsa key from 92.243.14.54 (13-14 June 2016 extraction) for Phuctor import. Ask asciilifeform or framedragger on Freenode, or email fd at mkj dot lt) <ssh...lt> ' (xvm-14-54.ghst.net. FR) [01:42]
deedbot: http://phuctor.nosuchlabs.com/gpgkey/EBB54B7022BA30DFD198B846506BFA051F02DCF89003A0C9917EF6AD8E2CD9DD << Recent Phuctorings. - Phuctored: 1790...2957 divides RSA Moduli belonging to '92.243.14.54 (ssh-rsa key from 92.243.14.54 (13-14 June 2016 extraction) for Phuctor import. Ask asciilifeform or framedragger on Freenode, or email fd at mkj dot lt) <ssh...lt> ' (xvm-14-54.ghst.net. FR) [01:42]
mod6: mornin' [10:45]
asciilifeform: heiya mod6 [10:46]
shinohai: gm mod6 [10:46]
trinque: mircea_popescu: makes sense. can't stop the bits from leaking can however simply let the owner of an account know every time something happens, so he can yell at me before I do the next batch of withdrawals if he didn't do it. [10:56]
asciilifeform: in other lulz, https://www.youtube.com/watch?v=IzZb6kfctEU << 'my father would spin like a dreidel in his grave if he saw' [11:12]
asciilifeform: 'the police abandoned us!' etc [11:14]
trinque: cute backyard tiki torches. I wonder if they got the citronella ones, useful for keeping skeeters away from your /pol/ flash mob. [11:15]
asciilifeform: in other hilarities almost worthy of mircea_popescu's robotzi, https://www.youtube.com/watch?v=VKkcTpCur7g [11:18]
asciilifeform: ^ ar propaganda cartoon re 'malvinas', with 'fixed' sub [11:18]
asciilifeform: in other olds , https://www.iacr.org/archive/ches2009/57470141/57470141.pdf << traditional rsa prime generation is quite 'loud'. this is not a seekrit. subj demonstrates algo for actually recovering the prime. [12:13]
asciilifeform: ( tldr : superiority of the FUCKGOATS-enabled approach, of get-new-N-bits-from-rng-then-primalitytest-until-done, vs the kochian get-N-bits-then-increment-until-passes-millerrabin ) [12:14]
asciilifeform: the other thing, you don't need ANY trial-divisions in the prelude to miller-rabin, IF you have a constant-time gcd [12:15]
asciilifeform: ( then, elementarily, you gcd against '8ball', primorial of $largeint ) [12:16]
asciilifeform: as usual asciilifeform has deeply nfi why NONE of the published rsatrons, to date, do this. [12:18]
shinohai: http://archive.is/Uffst <<< lulzy ... "After a software update was sent to your lock, it failed to reconnect to our web service making a remote fix impossible" [12:45]
shinohai: Theonly 2 fixes proffered require AT LEAST one week to implement [12:46]
mircea_popescu: http://trilema.com/forum-logs-for-14-aug-2017#2322812 << don't you just love it how the implication is that your item they broke was "somehow" at fault ? it's not "we pushed a buggy piece of shit into your item and thereby broke it". not ever. [12:53]
a111: Logged on 2017-08-14 16:45 shinohai: http://archive.is/Uffst <<< lulzy ... "After a software update was sent to your lock, it failed to reconnect to our web service making a remote fix impossible" [12:53]
mircea_popescu: it's your racist victim that stood in the way of black code matters and progress and spice and everything nice. [12:53]
mircea_popescu: http://btcbase.org/log/2017-08-14#1697556 << the funny thing is that there's not enough of them by now. same exact thing happened to the german communists in the 30s. "folks this is the end", ie, "we used to think we're right because we're many, and now we're not many, and that's all we had." [12:55]
a111: Logged on 2017-08-14 15:12 asciilifeform: in other lulz, https://www.youtube.com/watch?v=IzZb6kfctEU << 'my father would spin like a dreidel in his grave if he saw' [12:55]
mircea_popescu: http://btcbase.org/log/2017-08-14#1697559 << the MOST hilarious thing in there is that the fucktard thinking himself an airplane pilot asks "me copia base". you understandf this ? subhuman orc language DOES NOT HAVE WORD. thinks it's ok, just as good as anything. [12:57]
a111: Logged on 2017-08-14 15:18 asciilifeform: in other hilarities almost worthy of mircea_popescu's robotzi, https://www.youtube.com/watch?v=VKkcTpCur7g [12:57]
mircea_popescu: and otherwise, in their own fucking idiota i mean "idioma", it goes like so : "con fuerza y corage dicimos adios". this. this is 100% argentina, brave, brave, brave sir robert turned about and galantly he chickened out... brave brave brave recursos humanos... [13:00]
mircea_popescu: http://btcbase.org/log/2017-08-14#1697562 << aha! [13:05]
a111: Logged on 2017-08-14 16:14 asciilifeform: ( tldr : superiority of the FUCKGOATS-enabled approach, of get-new-N-bits-from-rng-then-primalitytest-until-done, vs the kochian get-N-bits-then-increment-until-passes-millerrabin ) [13:05]
mircea_popescu: it is also a very typical difference, symbolic enough to go on our fucking flag. "the empire makes a test and then goes through parts until it finds one that goes through the republic makes a part and tests it until it is certain to be correct." [13:06]
asciilifeform: the important bit : if fails, make a ~new~ one, rather than n+1 [13:07]
mircea_popescu: yup. [13:07]
mircea_popescu: it's funny how all the things are the same thing and everything wraps into ideological identity. empire needs... a lot of really dumb ones, as a COLLECTIVE. we... make every one stand on its own INDIVIDUALLY. [13:08]
asciilifeform: 'rng bits are expensive' spawned quite a few idiocies , by itself [13:08]
mircea_popescu: http://btcbase.org/log/2017-08-14#1697564 << speaking of this, how large is phuctor 8ball ? ie largest prime ? [13:08]
a111: Logged on 2017-08-14 16:16 asciilifeform: ( then, elementarily, you gcd against '8ball', primorial of $largeint ) [13:08]
asciilifeform: 1.2G at last count [13:08]
asciilifeform: it's largely useless tho [13:09]
mircea_popescu: asciilifeform i believe the spawner is the same mother-of-idiocy, always pregnant, always knees spread, that spawned everythiong they do, from "voting" to "gpg" [13:09]
mircea_popescu: asciilifeform it would seem an 8ball usable for 4kb rsa key verification would be exceedinglty large. [13:09]
asciilifeform: you wouldn't use it alone , lol!! [13:09]
mircea_popescu: asciilifeform i don't mean the disk size of the 8ball, i mean the size of the largest prime in it [13:09]
asciilifeform: it's a prelude for avoiding expensive miller-rabin when the latter is doomed to fail [13:10]
asciilifeform: ah hm will have to see re largest prime [13:10]
mircea_popescu: a yes. a mb or so's worth is good to have. afaik all rsa impls have some small primes. heck, peterl's gossiptron had a list. [13:10]
asciilifeform: you wouldn't want a mb of anything in ffa tho [13:11]
mircea_popescu: asciilifeform amusingly, wouldn't it be len(factorial)/log(len(factorial)) ? :D [13:11]
asciilifeform: approx!yes [13:11]
mircea_popescu: aha. [13:11]
asciilifeform: idea is, for pre-millerrabin litmus, take gcd(candidate, Qw) where Qw is largest primorial that fits in the ffawidth [13:15]
mircea_popescu: ah, the primorial being 4kb [13:15]
mircea_popescu: this is not bad. [13:15]
* asciilifeform still devising a constant time gcd [13:15]
asciilifeform: the basic, naive method for magicking a conventional algo into a constanttime algo, is to [13:16]
asciilifeform: 1) replace termination condition with a mux that starts discarding new results of iteration in favour of old, at iteration T and after [13:17]
asciilifeform: 2) replace variable iteration params with fixed, conditional terminations with same [13:18]
asciilifeform: this is still difficult for gcd because gotta prove max number of shots needed [13:19]
asciilifeform: for given w. [13:19]
mircea_popescu: this is actually going to be teh magic number of the republic. so at this juncture i would like to ask everyone to compute "the largest primorial (ie, product of all successive primes) that fits in 515 bits", sign it and put it into deedbot. [13:20]
mircea_popescu: diana_coman, hanbot, trinque, bingoboingo, mod6, danielpbarron, mike_c, asciilifeform, davout, ben_vulpes, phf, lobbes, mike_c, jurov, peterl, pete_dushenski ^ [13:20]
asciilifeform: wai 515 [13:20]
mircea_popescu: the idea is to use DIFFERENT methods to compute it. [13:20]
mircea_popescu: asciilifeform http://btcbase.org/log/2017-08-10#1696601 [13:20]
a111: Logged on 2017-08-10 14:54 mircea_popescu: 257, 258, 515. [13:20]
mircea_popescu: tmsr rsa standard key is 515 bits, made out of a 257 and a 258 bit long prime. [13:21]
asciilifeform: waat [13:21]
asciilifeform: 1024b?! [13:21]
asciilifeform: breakable even today [13:21]
mircea_popescu: damn. bytes not bits [13:21]
asciilifeform: lool [13:21]
mircea_popescu: jesus [13:21]
BingoBoingo: !~ticker --market all [13:31]
jhvh1: BingoBoingo: Bitstamp BTCUSD last: 4305.0, vol: 12077.63805615 | Bitfinex BTCUSD last: 4297.4, vol: 34009.57289113 | BTCChina BTCUSD last: 4294.660483, vol: 18039.95370000 | Kraken BTCUSD last: 4331.509, vol: 8242.7757566 | Volume-weighted last average: 4301.87039373 [13:31]
BingoBoingo: ^ Markets love ffa [13:31]
hanbot: <mircea_popescu> this is actually going to be teh magic number of the republic. so at this juncture i would like to ask everyone to compute "the largest primorial (ie, product of all successive primes) that fits in 515 bits", sign it and put it into deedbot. << is this right (up to 2900)? http://wotpaste.cascadianhacker.com/pastes/9lxGb/?raw=true [13:33]
mircea_popescu: hanbot what's that, first 419 primes ? [13:34]
* mircea_popescu is not going to comment re right or wrong for teh obvious reason! [13:34]
hanbot: yeah [13:35]
asciilifeform: that's 4104 bits... [13:36]
asciilifeform: > 4096 [13:36]
mircea_popescu: fencepost eh :D [13:36]
hanbot: damn [13:36]
mircea_popescu: asciilifeform tmsr key = 515 bytes = 4120 bits. [13:36]
mircea_popescu: ie > 4096. [13:37]
asciilifeform: that ain't divisible by 64 [13:37]
asciilifeform: and won't work in ffa. [13:37]
asciilifeform: unless you push it up to next multiple. [13:37]
mircea_popescu: so the correct work spec here is then, 4160 bits ? [13:37]
asciilifeform: ( at which point you oughta use THAT as the cap, not it ) [13:37]
asciilifeform: btw does it makes sense why i put in that req ? [13:37]
mircea_popescu: so ffa only allocates by... 8 byte chunks ? [13:38]
mircea_popescu: notrly. [13:38]
asciilifeform: it has to do SAME THING on all iron [13:38]
asciilifeform: which requres a 64b quantum [13:38]
mircea_popescu: so ? [13:38]
asciilifeform: so it gotta work on 16, 32, and 64 bitness machine. [13:38]
mircea_popescu: and an update when the 128 computer hits the shelves ? [13:38]
asciilifeform: 64 divides 128 [13:38]
mircea_popescu: ... [13:38]
mircea_popescu: 8 also divides 128 [13:38]
asciilifeform: understand, it is not possible to use partial machinewords in the simplest possible (i.e. the one in ffa) arithmetic method. [13:39]
mircea_popescu: answer the q then! when a 128 bit computer is sold, ffa word will ahve to increase to 128 bits ? [13:39]
asciilifeform: no!! [13:39]
asciilifeform: nor 192bit cpu [13:39]
mircea_popescu: then how about you know, it uses 8 bit chunks now. [13:39]
asciilifeform: why?! [13:40]
asciilifeform: you want 8x slowdown? for what? [13:40]
mircea_popescu: mmm [13:40]
mircea_popescu: this has been the worst explanation of a rationale in recorded history. care to do it over ? [13:40]
asciilifeform: i have nfi where to even begin [13:40]
mircea_popescu: quote your spec. [13:41]
mircea_popescu: that's where you always begin. [13:41]
asciilifeform: 'ffa represents a W-bit integer as a contiguous array of N machine words of bitness B, W = N*B.' [13:42]
mircea_popescu: go on. [13:43]
asciilifeform: 'W is constrained, such that any permissible value of W must be representable in a whole number of machine words on 8, 16, 32, 64-bit ALU.' [13:44]
mircea_popescu: "N must be 64 because at some point i nthe past a 64 bit machine was released and we care N will not have to be 128 in the future because even though an 128 bit machine will probably be released in the future, we don't understand the future and consequently do not care" [13:44]
asciilifeform: mno. on 128bit machine, we still can use 64bit arith, [13:45]
asciilifeform: m [13:45]
mircea_popescu: and then why can't we use 32 bit aritm on 64 bit machine ? [13:45]
asciilifeform: just like ffa worx just fine on opteron today dialed down to 32 [13:45]
asciilifeform: because you lose >2x performance for no gain. [13:45]
asciilifeform: it is idiocy and i won't countenance it. [13:45]
mircea_popescu: so then we CANT use 64 bit aritm on 128 bit machine. [13:45]
asciilifeform: can. [13:45]
mircea_popescu: you are contradicting yourself. [13:46]
asciilifeform: goal is ADEQUATE performance on iron of TODAY [13:46]
mircea_popescu: i wasn't aware. [13:46]
asciilifeform: the hypothetical future 128ism is not , in that light, needed for anything. [13:46]
mircea_popescu: goal is definitive solution to math problem. [13:46]
asciilifeform: it can go to hell. [13:46]
asciilifeform: mircea_popescu: 'solution to math problem' existed in 1978. ffa goal is simplicity+correctness of implementation + adequateperformance. [13:47]
mircea_popescu: turns out my mock-rewrite was more on point than originally thought. [13:47]
asciilifeform: a trio that has not prev been achieved at any point. [13:47]
mircea_popescu: you'll have to re-read this log tomorro. [13:47]
asciilifeform: rewrite of what [13:47]
mircea_popescu: http://btcbase.org/log/2017-08-14#1697664 [13:48]
a111: Logged on 2017-08-14 17:44 mircea_popescu: "N must be 64 because at some point i nthe past a 64 bit machine was released and we care N will not have to be 128 in the future because even though an 128 bit machine will probably be released in the future, we don't understand the future and consequently do not care" [13:48]
asciilifeform: if you read the code, mircea_popescu , [13:48]
asciilifeform: would be very clear that the 'must sit in whole number of machine words' is not any kind of constraint imposed by asciilifeform , but by the machine [13:48]
asciilifeform: it is not possible to have anything that looks like ffa, without suffering this constraint. [13:49]
mircea_popescu: ah, that part is not in dispute. [13:49]
asciilifeform: and 'W % 64 == 0' is not any more or less nonsensical than 'W % 8 == 0' which you are not escaping from to anywhere [13:49]
asciilifeform: not to the moon, not to mars [13:49]
asciilifeform: ( perhaps to pdp8... ) [13:49]
mircea_popescu: but this important point has important consequences, because now we can't have my eccentric rsa keys. must be 4096, because the only alternatives ffa permits are 2048 which is too short and 8912 which is too long. [13:50]
asciilifeform: not quite [13:50]
asciilifeform: let's work an example [13:50]
asciilifeform: say for some peculiar purpose, an ffa run needs 192-bitness. [13:51]
mircea_popescu: and if your key isn't 2^n, further computers will wreak havoc on deployed code by requiring non-dividing Ns. [13:51]
asciilifeform: this becomes 3 64-bit words on opteron 6 32-bit words on pentium 12 16-bit words on 8086 and 24 8-bit 'words' on 6502. [13:51]
asciilifeform: each of which computes exactly same thing, correctly. [13:52]
mircea_popescu: and then when a 128 bit machine comes along ? [13:52]
asciilifeform: mircea_popescu: you can use any key bitness you like ! but gotta top it out with 0s to sit it into a ffa word [13:52]
asciilifeform: this is not end of the world [13:52]
asciilifeform: just like your opteron is happy to add 1 + 1, even though 1 is a '1-bit' rather than 64-bit int [13:52]
asciilifeform: think about it [13:52]
mircea_popescu: that part is not in dispute either. [13:53]
asciilifeform: when on 128-bit iron , which exists today, you simply gotta pad out your payload so it sits in a W multiple of 128 ( supposing you insist on squeezing every penny of horse out of the 128ness ) [13:53]
asciilifeform: you are already forced to pad to multiple of 8 on every iron in existence today, this is no different. and nobody will reprieve you from it. [13:54]
mircea_popescu: asciilifeform yes, and then you're going to compute 192 keys on 256 bits, and wonder why the fuck are you not using 256. [13:54]
mircea_popescu: which is how "it's either 4096 bits long or get lost" ends up in there. [13:54]
asciilifeform: so you find nearest multiple that is longer than your payload. [13:54]
asciilifeform: and use that. [13:54]
mircea_popescu: because 4096 will accomodate all computers up to 4096 bit ones. [13:54]
asciilifeform: btw the bitserial cpu thread prolly belongs linked here [13:54]
asciilifeform: but i lost the link. [13:54]
asciilifeform: ( there WAS attempt to make 'comp of no fixed bitness' ) [13:54]
mircea_popescu: http://btcbase.org/log/2016-02-22#1413259 [13:55]
a111: Logged on 2016-02-22 23:21 asciilifeform: bit-serial cpu is meaningfully and interestingly unlike anything that came before (or, afaik, after.) [13:55]
asciilifeform: aaaha here . [13:55]
asciilifeform: ftr i considered imposing a 'ffa W is power of 2 or fuckyou' [13:56]
* mircea_popescu is really pissy that "gotta use trad width keys" ended up imposed on him! [13:56]
asciilifeform: it would simplify many routines, in particular karatsuba [13:56]
asciilifeform: but i specifically did not like that it gives 'traditional keys' [13:56]
asciilifeform: or at least encourages them [13:56]
mircea_popescu: yes but i see no way out. [13:56]
asciilifeform: and so instead went with 'mult of 64' [13:56]
asciilifeform: this is open to debate, it is an aesthetic choice [13:56]
asciilifeform: however P progs MUST execute to same result on ALL known iron, is the idea [13:56]
asciilifeform: can't just say 'sorry , can't do that here' [13:57]
mircea_popescu: no, it is not, in that it would simplify routines. that's what makes it an aesthetic non-choice. [13:57]
asciilifeform: if instead of 'mult of 64' we had 'powers of 2', we could dispense with the odd split in karatsuba [13:57]
mircea_popescu: anyway, i shall go eat frozen pastry and fume. [13:57]
asciilifeform: and by extension, with the temp buffers in same [13:58]
asciilifeform: because could instead pass the splits directly [13:58]
asciilifeform: this is also a 10% or so speedup. [13:58]
asciilifeform: but then you can only 8,16,....,4096,... etc. W. [13:58]
* asciilifeform was going for max generality, at minimal expense [13:59]
asciilifeform: i'ma repeat that http://btcbase.org/log/2017-08-14#1697720 is a mistake -- you can still use any key width you like. just gotta 0extend up to the permitted multiple. [14:00]
a111: Logged on 2017-08-14 17:56 mircea_popescu is really pissy that "gotta use trad width keys" ended up imposed on him! [14:00]
asciilifeform: ( by a max of B-1 bits , where B is your machine word ) [14:00]
asciilifeform: http://btcbase.org/log/2017-08-14#1697693 << and importantly, current ffa works with ( see factorial demo ) any multiple of 64, that fits in your machine memory. [14:02]
a111: Logged on 2017-08-14 17:50 mircea_popescu: but this important point has important consequences, because now we can't have my eccentric rsa keys. must be 4096, because the only alternatives ffa permits are 2048 which is too short and 8912 which is too long. [14:02]
asciilifeform: e.g. 6666-bit keys work fine on ffa! [14:25]
asciilifeform: simply take a 6720-bit W. [14:26]
asciilifeform: recall also that, since we have karatsuba, cost goes up with W logarithmically, rather than quadratically [14:27]
asciilifeform: incidentally, there is no reason why the ~public~ exponent , on ffatronic rsa, should not also be a large prime [14:29]
asciilifeform: ( you no longer win anything by using small ones ) [14:29]
asciilifeform: 'large' means 'occupies most of W' here. [14:30]
asciilifeform: ALL ffa ops take time that is not dependent on the hamming weight. that's what 'constant time' means. [14:30]
mircea_popescu: asciilifeform no dude, consider the catechistic angle. "soo... why is your key 515 byts ?" "i dunno, his lordship mp said so" "why ?" "nobody knoiws, really. he just says things." "so how do you calculate it ?" "first, you set ffa to 520 bytes..." "why did he say 515 then ?" "uh... that's a good question." [14:32]
mircea_popescu: there's no fucking way out of it. [14:32]
asciilifeform: well theoretically fpga [14:32]
mircea_popescu: re exponent, it's traditionally 0101 because cheaper calcs. [14:32]
asciilifeform: but on your current iron you're stuck with cups of a certain size. [14:32]
asciilifeform: mircea_popescu: but not cheaper for us [14:32]
mircea_popescu: why, no penalty to use checkerpattern instead ? [14:32]
mircea_popescu: aha. prolly should have a "dense" prime then. [14:33]
asciilifeform: no reason to use any particular pattern, understand, ffa [14:33]
mircea_popescu: (ie, 50% bits set) [14:33]
asciilifeform: moar hamming weight, the merrier for the enemy to suffer crunching'em [14:33]
asciilifeform: with his variable-time ops [14:33]
mircea_popescu: anyway. i am sad, but defeated. the dream has to die, it's what it is. so regretfully... 4096 keys. [14:33]
mircea_popescu: !!up PeterL [14:33]
deedbot: PeterL voiced for 30 minutes. [14:33]
* asciilifeform points out that even very modest iron, ffa's quite acceptably over 8192b and higher. [14:34]
PeterL: I am still digesting logs, did you guys agree on a bit size for the primorial you want? [14:34]
asciilifeform: PeterL: these ain't hard to calculate [14:34]
mircea_popescu: PeterL we found a hole in the spec / [14:34]
PeterL: http://wotpaste.cascadianhacker.com/pastes/lObj3/?raw=true << should be 4160 bits [14:35]
mircea_popescu: ty [14:35]
PeterL: oh, wait, nevermind, that is wrong [14:37]
asciilifeform: mircea_popescu plox to expand on http://btcbase.org/log/2017-08-14#1697729 [14:39]
a111: Logged on 2017-08-14 17:57 mircea_popescu: no, it is not, in that it would simplify routines. that's what makes it an aesthetic non-choice. [14:39]
asciilifeform: 'non-choice' here means 'obviously Right Thing' or the opposite ? [14:39]
PeterL: http://wotpaste.cascadianhacker.com/pastes/x9YvV/?raw=true << this is a better one! [14:39]
PeterL: (I forgot to increment a variable, so the first number is just 2*(a big exponent of 3) ) [14:40]
asciilifeform: PeterL: that's a 4150b [14:40]
PeterL: yes, seems so. [14:41]
PeterL: could be the next prime would put it over the limit, right? [14:41]
asciilifeform: it is already over the limit [14:41]
PeterL: (I was using 4160 bits as the limit) [14:42]
PeterL: what limit do you want? [14:42]
asciilifeform: 4096, i thought, it was [14:42]
mircea_popescu: asciilifeform 1111111111111111111111111111111111111111111111111111111111000101 ? or rather something more like 1100111110111010111111000100100010011101001010001000100100011001 ? [14:43]
PeterL: http://wotpaste.cascadianhacker.com/pastes/fLrxv/?raw=true << should fit in 4096 bits [14:43]
mircea_popescu: asciilifeform it means there's no space for choosing, as there's an obvious right thing. [14:43]
asciilifeform: mircea_popescu: there's something to be said for 111111.....1 (max hamming weight), and there's something else to be said for max-entropy [14:45]
asciilifeform: i cannot currently say which i prefer [14:45]
asciilifeform: ( or describe a logic for showing that one is always preferable to the other ) [14:45]
mircea_popescu: me either. [14:45]
asciilifeform: as for the other thing, right now we have a 'classical' karatsuba that permits odd splits [14:46]
mircea_popescu: or at least show some logic whatsoever. "it's sometimes better", anything. [14:46]
asciilifeform: this means you can have, e.g., 192b, 384b, etc ffa [14:46]
mircea_popescu: asciilifeform there's nothing wrong with HAVING it. and people can use it. BUT for rsa we should use the even one. [14:46]
mircea_popescu: it's what it is. [14:46]
mircea_popescu: (at least we get this much from the sadness : if anyone asks "why 4096" user can link here._ [14:46]
asciilifeform: forcing 2^x = W, x in Z , also simplifies comba [14:47]
asciilifeform: and a few other things. [14:47]
mircea_popescu: yes. [14:47]
asciilifeform: the 192b cpu of the dark future will have to smoke sadly... [14:47]
mircea_popescu: nobody is ever making non 2^n anyways! [14:47]
asciilifeform: well i thought of making a 3^n. but there's no reason not to say 'be a whole power of the machine BASE ' ! [14:48]
PeterL: fight specificity of didling, use odd machine word size! [14:49]
mircea_popescu: asciilifeform aha. [14:49]
asciilifeform: PeterL: sorta why i wrote the most general , unconstrained ffa . [14:49]
mircea_popescu: PeterL that lemma refers to domain borders. there's no border re hardware. [14:49]
asciilifeform: meanwhile, usg.prb folx claiming they will soon 'broadcast blocks from satellite', 'use prb without internet!111' -- leveraging the orbit monopoly to diddle 'these blox, mined by us, legal, other, terrorist blox -- orphans nao' [14:53]
mircea_popescu: aww,\ [14:53]
mircea_popescu: i don't get it, if you have satellite you already have internet ? [14:53]
asciilifeform: 1way [14:54]
mircea_popescu: this is how i get it in cars etc. [14:54]
mircea_popescu: uh. really ? 1 way from owned source ? and how do you check for splits ? [14:54]
asciilifeform: 1way much cheaper [14:54]
asciilifeform: and noshit.jpg you check for nuffin, you eat the mother church's blox as the come, presumably [14:54]
asciilifeform: paypalization [14:54]
asciilifeform: '1 way from owned source' noshit. [14:55]
mircea_popescu: lol idiots. [14:55]
asciilifeform: they've been looking for how to monopolize relay, for years. [14:55]
asciilifeform: ( much cheaper than mining ) [14:55]
mircea_popescu: to quote, http://btcbase.org/log/2017-08-14#1697581 [14:55]
a111: Logged on 2017-08-14 17:08 mircea_popescu: it's funny how all the things are the same thing and everything wraps into ideological identity. empire needs... a lot of really dumb ones, as a COLLECTIVE. we... make every one stand on its own INDIVIDUALLY. [14:55]
mircea_popescu: asciilifeform who the fuck would depend on a satellite. for one thing they keep falling. [14:55]
asciilifeform: now this i cannot say [14:56]
mircea_popescu: it WAS cheaper before they got involved lol [14:56]
mircea_popescu: now it's more expensive than mining. cuz gotta airforce one it [14:56]
asciilifeform: presumably plan is part of 2-pronged approach where they interfere with ordinary propagation. [14:56]
asciilifeform: and claim 'sat is best' [14:56]
asciilifeform: yeah but they ~already own~ the sat [14:56]
mircea_popescu: ill interfere with a satellite lol. [14:56]
asciilifeform: http://btcbase.org/log/2015-10-15#1299189 << thread [14:57]
a111: Logged on 2015-10-15 01:03 mircea_popescu: if you want a Tatra truck, ask for it. [14:57]
mircea_popescu: tru fact : you can take down satellites TODAY. as an individual effort. [14:57]
asciilifeform: depends what means individual [14:57]
mircea_popescu: one guy. [14:57]
asciilifeform: it's ~same work as launching yer own [14:58]
asciilifeform: (elementarily) [14:58]
mircea_popescu: not really. all you need is a small rocket [14:58]
mircea_popescu: easy to home on them lol. for obvious reasons. [14:58]
asciilifeform: well yes [14:58]
asciilifeform: small rocket that goes to same orbit. [14:58]
asciilifeform: ergo same work. [14:58]
mircea_popescu: and it doesn't have to be fast or anything. nor does it need payload. [14:58]
asciilifeform: well needs a kg or so [14:58]
asciilifeform: ( you don't have blast, recall, must rely on frags ) [14:58]
mircea_popescu: asciilifeform no. different work to PUT satellite in orbit, and to send it an old bit of spanish grapeshot [14:59]
mircea_popescu: hardly. 70-80 grams will do it. [14:59]
asciilifeform: also depends, do you want to machinegun 1 sat, or... whole orbit [14:59]
asciilifeform: mircea_popescu: depends on distance, neh [14:59]
mircea_popescu: distance from what [14:59]
asciilifeform: contact detonator ? proximity ? [14:59]
mircea_popescu: but yes, glonass slightly better than usg sats because of orbit style. [14:59]
mircea_popescu: you get some warning [14:59]
mircea_popescu: asciilifeform no, just mass. [15:00]
mircea_popescu: i literally mean, dig up an old bit of grapeshot from age of sail, send it up., [15:00]
asciilifeform: if you're at musket range, even shot shell would work [15:00]
asciilifeform: problem is to get there. [15:00]
mircea_popescu: homing rocket. [15:00]
asciilifeform: this in practice is harder than simple sat launch ( the latter is not trying to fly to a specific point in space ) [15:01]
mircea_popescu: 2 stage, a 1-200kg booster thing to get it in gross range and a 2-3kg thing to power the last 100 grame spiked ball to the actual satellite [15:01]
mircea_popescu: the cost per ton of this (materials, excluding labour and knowledge) is under 100 bux. [15:02]
mircea_popescu: ie, less than car. [15:02]
asciilifeform: but at any rate this is not open problem, it was solved in ~same way in all 3 major inca empires -- fighter goes to max altitude, launches slightly modified radar-seeker rocket, vertically. [15:02]
mircea_popescu: aha. [15:02]
mircea_popescu: borrow phf 's ballooons lol [15:02]
asciilifeform: less, incidentally, conspicuous, than monster flare from the ground [15:02]
mircea_popescu: i thought the point was conspicuity. [15:03]
mircea_popescu: demonstration, yes ? [15:03]
mircea_popescu: ows occupies ws, tmsr takes out your sats. [15:03]
asciilifeform: mircea_popescu: 'cossacks write letter to the sultan' in order? [15:04]
mircea_popescu: something like that. [15:04]
mircea_popescu: anyway. lulz in being, so far. [15:04]
asciilifeform: ( http://www.loper-os.org/wp-content/sultan.jpg oblig ) [15:04]
mircea_popescu: anyway, to the general point : satellite comms are a very limited thing, and will definitely go the same way the old roman dinner (ie, fuckfest) went once the medievals discovered syphilis. [15:09]
mircea_popescu: there's a cost to exploration, typified by pandora. [15:09]
asciilifeform: sorta why the balloon thing keeps coming back -- search for a midpoint b/w 'tower' and 'satellite' cost [15:09]
mircea_popescu: aha [15:10]
asciilifeform: http://archive.is/264y7 << relatedly [15:13]
mircea_popescu: leller. [15:13]
asciilifeform: gotta point out that nailing 1 enemy sat is not enough [15:17]
asciilifeform: ( they have constellation ) [15:17]
asciilifeform: gotta show ability for sustained extermination. [15:17]
mircea_popescu: it's enough to make a point. [15:17]
asciilifeform: d00dz seem to be physically incapable of grasping a point by induction ( even the skyscrapers , apparently, didn't help... ) [15:18]
mircea_popescu: there is that, yes. [15:20]
mircea_popescu: then again, sept 11 marked the end of vhs-america. the remainder usg-sadness is not even at the level of hruschev's "reformed cccp" [15:21]
asciilifeform: buncha monkeys, sitting on remnants of massive imperial machinery made by men. and looking for any, any way to leverage'em into a few moar yrs of pushing back the inevitable [15:22]
mircea_popescu: and of course as women are the social scar tissue, this is going to end up with yet another "oh, it's THEIR fault!!!", like every historical case to date. [15:26]
asciilifeform: relatedly, a shortwave ( or even ultralong wave! ) transmitter of genuine ( i.e. best-known-pov ) blox, would be a fine thing [15:27]
asciilifeform: ( see old thread ) [15:28]
asciilifeform: *pow [15:28]
asciilifeform: but if it isn't obvious why : [15:28]
asciilifeform: it would make the kind of miner monkey work currently fashionable, ~impossible [15:28]
asciilifeform: ( withholding in particular ) [15:29]
asciilifeform: getting new blox to the transmitter asap would, theoretically, be a schelling point for miners ( all of whom currently 'eat the spoon of shit for free' from withholding chains ) [15:30]
mircea_popescu: not really much of a problem yet, that spoon. [15:35]
* asciilifeform not equipped to say at present how much of a problem [15:36]
asciilifeform: it ~would~ be a problem for any would-be defector from cabal [15:36]
mircea_popescu: fees not yet enough. give it a coupla halvings. [15:37]
asciilifeform: ( and is the principal reason why there even is cabal ) [15:37]
mircea_popescu: this is the important point of having defeated the usg flailing attempts to prevent actual fee market from developping : it will WORK. [15:37]
asciilifeform: other idea is to demonstrate that trb can work without riding on top of existing telecom structure. [15:38]
mircea_popescu: that's even later. [15:38]
asciilifeform: this means a node that can speak directly to another. [15:38]
asciilifeform: ( via radio, lan cable strung through window, whichever means. preferably transparently of layer.) [15:39]
asciilifeform: and incidentally, open fpga trivially yields sdr. [15:40]
deedbot: http://qntra.net/2017/08/amazon-fucks-retailers-panics-customers-with-recall-week-before-total-eclipse/ << Qntra - Amazon Fucks Retailers, Panics Customers With Recall Week Before Total Eclipse [16:25]
BingoBoingo: ^ Moar Inca Pls [16:26]
BingoBoingo: From the logs of the still suffering https://pbs.twimg.com/media/DHML2bPU0AEzaR2.jpg [16:27]
mircea_popescu: fun fact : watching the sun is not harmful to the eye. [16:30]
mircea_popescu: if it were, life would have evolved... A DIFFERENT FUCKING EYE [16:30]
mircea_popescu: wtf is wrong with people. [16:30]
BingoBoingo: Eh, there's probably a difference between eye taking in sunlight and focusing the lens apparatus of the eye directly at the sun. [16:33]
mircea_popescu: so what does the animal do, go blind every time it sees the sun ? [16:34]
mircea_popescu: yes, concave mirrors and lenses can blind you. but if the sun itself is a problem then you're a lifefrom from a different planet. [16:34]
mircea_popescu: somehow this has evolved from "im an old woman, nobody wants to pay attention to me, let me nag people" to "don't stare at the sun" to "omfg SUN IS DANGEROUS!" and it will never fucking end. [16:36]
BingoBoingo: Well, much of humanity is more stupid than the rest of the animals. Likely will continue staring well beyond point pain receptors yell to cease and decist. [16:36]
mircea_popescu: pupil closes reflexively. [16:36]
mircea_popescu: that's why the whole "doctor shining light in eyes" thing. [16:37]
mircea_popescu: if it doesn't close -- braindead. [16:37]
BingoBoingo: If we have learned anything from the Roger Ver-ified tribe, it is that there is a portion of the idiocy which will stare at the sun, for however long it takes the moon to transit. From this the nags and bags get their power. All it takes is one stuple to set all the nags off. [16:38]
mircea_popescu: yeah well. maybe some people actually should be blind. [16:39]
BingoBoingo: The nags are too deaf to hear that. Too much Maury/CNN/kennedy exposure. [16:40]
BingoBoingo: They can't get those language abilities back [16:40]
BingoBoingo: permabroke [16:40]
mircea_popescu: maybe should get some cnn protecvtors [16:40]
BingoBoingo: Too late [16:42]
mircea_popescu: for their daughters. who the fuck cares about the bags now. [16:42]
BingoBoingo: Biodiesel plant? [16:43]
mircea_popescu: maybe just "dun listen to old bags, they make you deaf and dumb" [16:43]
BingoBoingo: !!up scatha [16:52]
deedbot: scatha voiced for 30 minutes. [16:52]
shinohai: Gold: " The Rust Evangelism Strike Force recommends several resources for people who would like to poorly reimplement other unix tools." [16:53]
BingoBoingo: Umbrella saga continues www.jameslafond.com/article.php?id=8676 [16:54]
mircea_popescu: how the fuck does the police "decline" to patrol. [16:57]
mircea_popescu: this is some mindblowing usiana right there. [16:57]
BingoBoingo: Well, at this point Baltimore ~= Internment Camp, at the very least Ghetto in the 1940s sense [16:59]
mircea_popescu: the direct equivalent would be argentarded shopkeepers "declining to sell". [16:59]
mircea_popescu: happens all the time, yes. [16:59]
BingoBoingo: Oh, USian shopkeeps do that all the time too! [17:00]
mircea_popescu: im sure [17:01]
BingoBoingo: Hence walmart et al who are happy to sell [17:01]
mircea_popescu: except for sunglasses. [17:01]
BingoBoingo: Most of the grocery stores have the eclipse things here. Cardboard 3D glasses type things with ~shade 13 filter and lots of small print about "no more than 3 minutes continuous use" [17:02]
mircea_popescu: did i ever tell you the story of the progressive tea shop in timisoara ? [17:03]
BingoBoingo: Not to my recollection [17:05]
mircea_popescu: some douchebag opened a tea house, with a lot of verbiage about how relaxing and progressive and everything it is, and how it goes in other places and you know, brimming with the imbecile self-satisfaction of the nemtsov. [17:06]
mircea_popescu: tea was served in cups on trays. it's ok for tea to come in teabags, what. the trays also included alarm clocks set for whatever minutes a 10-bux-an-hour pinoy randomly put on the mass produced "fair trade" crappy teabags. 6 or 7 or 3 or whatever. [17:07]
mircea_popescu: consequently this idiot's teahouse was constantly ringing of alarms on multiple voices from multiple points. [17:07]
mircea_popescu: this is the typical end product of the imbecile "just the facts" "i am equpped to think for myself" ustard. [17:08]
mircea_popescu: some cheap plastic, a lot fo fine print and getting in everyone's way/ [17:08]
BingoBoingo: Fuck [17:09]
mircea_popescu: (for the record, the difference between the nemtsov and the pantsuit is that while they both represent the herdemocracy side, the pantsuit is ineffectual, lazy, effete citizen, whereas the nemtsov is hardy, insistent, ambitious orc born outside the walls.) [17:10]
BingoBoingo: In other social experiments on the substance of herdemocracy http://www.jameslafond.com/article.php?id=8674&pr=1 [17:14]
BingoBoingo: And the anthropological collection of Charlottesville stuff from the aquarium http://www.jameslafond.com/article.php?id=8661&pr=1 [17:22]
shinohai: https://steemit.com/cryptocurrency/@anonymint/re-anonymint-re-anonymint-re-anonymint-re-anonymint-shocking-crisis-coming-to-cryptocurrency-in-sept-20170814t065451341z <<< The expert on mircea_popescu has spoken [17:39]
mircea_popescu: what's it say ? [17:39]
shinohai: "I believe Mircea Popescu has gravely miscalculated. Either he has been coloring his own BTC within the danger period (i.e. being careful of the lineage of the BTC obtained in recent times), else his BTC is also susceptible to being stolen. Otherwise, he would need vast mining resources to maintain his own fork or he would need to penalize/bribe miners to not steal his BTC where it is so encumbered. He can [17:39]
shinohai: penalize by selling the fork and buying another honest one, but he will need sufficient mining resources and other significant BTC holders to enjoin him in this fight. Rather I think most astute BTC holders will realize this is the end of the paradigm and liquidate.' [17:39]
mircea_popescu: what is all this about ?! [17:40]
asciilifeform: it reads like barf [17:43]
asciilifeform: in the most narrowly technical sense of the word [17:43]
asciilifeform: half-digested, regurgitated matter [17:43]
shinohai: I honestly think anonymint uses crystal meth [17:44]
mircea_popescu: weird. anyway. [17:44]
asciilifeform: student exercise : write a generator of 'shocking crises coming to bitcoin!!!' crapola [17:45]
asciilifeform: betcha shannonizer could at least match quality of shinohai's link [17:46]
mircea_popescu: a better exercise would be if any/all these webexperts actually bothered to sit down and derive their conclusions from prime notions. [17:47]
shinohai: Some derp on tardstalk linked me to it in a PM ... wouldnt be surprised if he has spammed them to dpb too [17:47]
* mircea_popescu is persuaded theyd bit flip A FEW TIMES in the process. [17:47]
mircea_popescu: shinohai so tell him to come over an' explain himself ? [17:47]
shinohai: Sent him a msg, may be lulzy if he shows. [17:51]
BingoBoingo: Lol, steemit wants a PHONE NUMBER to post a comment! [17:51]
mircea_popescu: "innovative" sf webscum running out of shit to sell. [17:51]
mod6: 1900BUTTSEX [17:51]
shinohai: ^ [17:51]
danielpbarron: i did get a pm on tardtalk recently but i didn't log in to read it just saw the notification email [18:06]
shinohai: https://pbs.twimg.com/media/DHM3nOhWsAA9PFB.jpg <<< Well they aren't wrong [18:06]
trinque: lol! [18:07]
BingoBoingo: https://www.reddit.com/r/The_Donald/comments/6tme1k/jason_kessler_organizer_of_unite_the_right/ [18:10]
BingoBoingo: "According to the Southern Poverty Law Centre, (fuck them, but they've done good research into him), Jason Kessler was previously involved with the Occupy Movement and was an ardent Obama supporter through both terms." [18:11]
BingoBoingo: !~ticker --market all [19:13]
jhvh1: BingoBoingo: Bitstamp BTCUSD last: 4256.77, vol: 13883.50586255 | Bitfinex BTCUSD last: 4259.9, vol: 33364.94804248 | BTCChina BTCUSD last: 4307.377999, vol: 17941.84650000 | Kraken BTCUSD last: 4261.677, vol: 8705.5858828 | Volume-weighted last average: 4271.04889427 [19:13]
mircea_popescu: um. "The goal is to turn peaceful protests violent, and spark a Civil War." i don't get it, how is this a bad thing ? [19:37]
mircea_popescu: is the "i just wanted to" right opposed at the "i just wanted to left" removal as being too radical ? after all, they DO "just want to" undisturbed ? or what ? [19:39]
asciilifeform: 'can't fight enemy who has outposts in your head' or how did it go. [19:40]
asciilifeform: a111 broke?? [19:42]
asciilifeform: phf: ?? [19:43]
asciilifeform: or massive netsplit [19:48]
asciilifeform: asciilifeform is showing up in Framedragger's log but not phf's or ben_vulpes [19:49]
asciilifeform: which is odd, because i still see the bots ! [19:51]
shinohai: Annnnnnnnnnnnnnnd ... there's yer answer asciilifeform [19:57]
mod6: yikes [19:59]
asciilifeform: dafuq [20:00]
ben_vulpes: gracious [20:00]
mircea_popescu: http://trilema.com/2014/so-the-dollar-vigilante-scam-ring-is-going-to-jail/#comment-122625 << in other "trilema aeterna" lulz. [20:03]
trinque: lol potato and wire architecture not going so well for fleanode [20:04]
shinohai: !~weather [20:05]
jhvh1: stormy with a chance of packeting [20:05]
deedbot: http://www.contravex.com/2017/08/14/the-phantom-of-the-opera/ << » Contravex: A blog by Pete Dushenski - The Phantom of the Opera [20:10]
mod6: shinohai has it [20:15]
asciilifeform: http://btcbase.org/log/2017-08-14#1697917 << pupil only contracts if ~whole retina is illuminated. hence 'do not stare into laser with remaining good eye' [20:41]
a111: Logged on 2017-08-14 20:37 mircea_popescu: that's why the whole "doctor shining light in eyes" thing. [20:41]
asciilifeform: http://btcbase.org/log/2017-08-14#1697933 << pretty easily. asciilifeform had a bumper wreck in bmore once, passing cop declined to take report, demanded 'who was killed here? if nobody, i have no time for this, and no one else from station will come either' [20:43]
a111: Logged on 2017-08-14 20:57 mircea_popescu: how the fuck does the police "decline" to patrol. [20:43]
asciilifeform: ( why even asked? because insurance pays 0 w/out police docs ) [20:44]
asciilifeform: iirc the details of this tale are in last summer's l0gz [20:45]
mod6: wb [21:39]
mike_c: good evening [21:40]
mod6: how goes tonight? [21:40]
mike_c: smoothly :) [21:40]
mod6: heheh, good to hear, Sir. [21:40]
phf: was it netsplit? everything appears operational [21:42]
mod6: phf: fwiw, i don't see a111 in my current /names list. [21:51]
mod6: Guest41016 [~a111@unaffiliated/phf/bot/a111-] [21:52]
mod6: that's it ^ [21:52]
phf: aah [21:52]
phf: i see [21:52]
phf: for a second there i thought freenode finally jumped the shark [21:53]
mod6: heheh. [21:53]
asciilifeform: yeah i thought also. [21:53]
asciilifeform: in other enigmas : http://wotpaste.cascadianhacker.com/pastes/5q4lR/?raw=true [22:01]
asciilifeform: it is an implementation of http://btcbase.org/log/2017-08-14#1697730 [22:01]
a111: Logged on 2017-08-14 17:57 asciilifeform: if instead of 'mult of 64' we had 'powers of 2', we could dispense with the odd split in karatsuba [22:01]
asciilifeform: however, the expected speedup did NOT materialize !! [22:01]
asciilifeform: despite 50% reduction in temp space used by karatsuba mult and square [22:02]
asciilifeform: how this can be -- remains a puzzler. [22:02]
mod6: so you didn't get the 10%? [22:03]
asciilifeform: nope [22:04]
asciilifeform: ~no measurable speedup [22:04]
asciilifeform: which is surreal. [22:04]
mod6: hmm [22:04]
asciilifeform: i suspect that the reason is that the op fit in cache ANYWAY even prior [22:04]
asciilifeform: so reducing temp space, does 0. [22:05]
asciilifeform: this kind of optimization could be interesting if we were dealing in MB+ ffaism [22:05]
asciilifeform: but apparently does 0 for us in the expected application. [22:06]
asciilifeform: also imho the item above is LESS readable than the original. [22:07]
asciilifeform: so it turns out that imposing 'powers of 2' is NOT a win. [22:07]
asciilifeform: at least not from a mechanical pov. [22:07]
asciilifeform: http://wotpaste.cascadianhacker.com/pastes/YoOl3/?raw=true << 'classical' versions of all 3 routines above, for comparison. [22:09]
mod6: as it stands, at the moment, sounds like a mechanical push. with the ability to set the length to some power of two. i'd say it's a bit harder to follow, code-wise, with the recursive calls perhaps. [22:09]
mod6: but i've only scanned it once. so perhaps I shouldnt comment too much on that. [22:10]
asciilifeform: mod6: see earlier thread [22:10]
mod6: my ffa must be way old [22:11]
asciilifeform: mod6: almost certainly is [22:11]
asciilifeform: i abolished the 'record' thing [22:11]
asciilifeform: ( to enable building with all oopism banned ) [22:11]
mod6: ah [22:11]
asciilifeform: and to simplify reading. [22:12]
asciilifeform: ( no moar foo.Z , now just foo ) [22:12]
mod6: ah, gotcha [22:12]
asciilifeform: mod6: idea with this item, is that L is a power of 2 always. in 'classical' one, L can be anything (e.g. a 192-bit ffa ends up 3*64 on my box, i.e. L=3 ) [22:13]
asciilifeform: but as it happens, my hypothesis re 'this will speed up mult' is wholly false and the one where 'it will simplify program from reader pov' also, somewhat paradoxically, false. [22:14]
mod6: this might be extra-strength dumb, but... in your new power of 2 version, do you need to inline the Mul & Square of karatsuba? [22:15]
asciilifeform: 'this' being, if it wasn't obvious, the powerof2 constraint thing. [22:15]
asciilifeform: mod6: you can't inline a recursive invocation, wtf [22:15]
asciilifeform: ( picture this ) [22:16]
mod6: ah herp. ok. [22:16]
asciilifeform: gcc will give you a very special eggog, even [22:16]
asciilifeform: ^ for readers who wondered why karatsuba is the 1 routine in ffa ~not~ inlined... think. [22:17]
mod6: yeah, ok, so the compiler can't determine the max depth to unroll when recursive [22:20]
asciilifeform: gnat ain't supposed to unroll. at all. [22:20]
asciilifeform: ( in so far as i can tell, it indeed respects the standard, and preserves control flow as written ) [22:21]
asciilifeform: mod6: you can't inline a recursive call because this'd be logically equiv. to making the program infinitely long [22:22]
asciilifeform: i thought this was clear... [22:22]
mod6: that's what i was trying to get at. i suppose the compiler would just reject your request to inline then, or perhaps thats ctronic thinking. [22:22]
asciilifeform: it will [22:23]
asciilifeform: ( and it ain't an instance of 'compiler too smart for own good', either, but perfectly legitimate refusal to try to fill up the universe with your mistake ) [22:24]
asciilifeform: the inline thing may seem like premature optimization, but function calls in ada are quite expensive, because bounds checking. so it makes MASSIVE difference. [22:25]
asciilifeform: hence why everything that oughta be inlined, i empirically determined, and already is. [22:26]
mod6: werd. [22:26]
asciilifeform: ( the whole thing is quite compact, and misering on codesize loses massively moar than it wins here. ( [22:26]
asciilifeform: ) [22:26]
asciilifeform: this is perhaps the most pedantically massaged item asciilifeform ever wrote, this thing [22:27]
asciilifeform: ( and it ain't even over yet ) [22:28]
* asciilifeform tries to count how many 'improved' versions he wrote, and then discarded... [22:35]
BingoBoingo: <mircea_popescu> is the "i just wanted to" right opposed at the "i just wanted to left" removal as being too radical ? after all, they DO "just want to" undisturbed ? or what ? << It seems like there is a mass of confusion in them. [23:12]
mod6: http://btcbase.org/log/2017-08-14#1697732 << >> <+asciilifeform> despite 50% reduction in temp space used by karatsuba mult and square [23:56]
a111: Logged on 2017-08-14 17:58 asciilifeform: and by extension, with the temp buffers in same [23:56]
mod6: this is very interesting. [23:56]
mod6: <+asciilifeform> this kind of optimization could be interesting if we were dealing in MB+ ffaism << yeah, perhaps the sample size used was not enough to see the delta? [23:58]
Category: Logs
Comments feed : RSS 2.0. Leave your own comment below, or send a trackback.
Add your cents! »
    If this is your first comment, it will wait to be approved. This usually takes a few hours. Subsequent comments are not delayed.