tmsr - Log 2019-06-07

▁▁▁▁▄▁▁▁▁▁▁▁⏐︎▁▁▁▁▁▁▁█▂▆▁▁ 1394

feedbot: http://trilema.com/2019/the-state-of-the-vidya-address/ << Trilema -- The "State of the Vidya" Address.

asciilifeform: 'MERCENARIES TO KILL' << english-as-she-is-spoken(tm)(r) lel

asciilifeform: incidentally, re 'valve co.', and also elliotisms, didjaknow -- it appears to have been named for... an actual valve. as in, 1 of those nyc-style steam main valves, as big as a desk, it is ceremonially mounted in their house...

asciilifeform: asciilifeform's brother once had occasion to go there, and confirmed, it is there, and even oiled, turns..

asciilifeform: in re: those browser gamez -- the 1080 is 'shooting fly with cannon' for these, they'd happily run on e.g. rk (or prolly even 486..)

mircea_popescu: thing could mine more bitcoin than the damned things are worth.

asciilifeform: fwiw asciilifeform's 'i'ma build gamez box!11' adventure from 2y ago, ended very similarly to mircea_popescu's : that box is cold nao an' waiting for when asciilifeform next time needs to route a pcb

asciilifeform: 'why the fuck did i just burn 2k dubloon to build box on which nao playing 'amiga' emulator gamez'

asciilifeform: mircea_popescu: '90s, to asciilifeform , era synonymous with the troo game.

asciilifeform: no one ever said 'i wish i did not have this spare iron' lol

mircea_popescu: nicoleci, hey, make an account on http://trilema.com/2019/the-state-of-the-vidya-address/#comment-129346 forum and ask the guy whether he's amenable to sell out to us.

BingoBoingo: The aging Lafond met with the family (archived lest it disappear into "print") http://archive.is/QgrQg

mircea_popescu: (here god stands for lobbes an'... spyked doing it too iirc ?)

asciilifeform: meanwhile, thinking moar re this item : the two tests make roughly comparable demand on rng: 3582 * (2048 / 8) byte == 916992 bytes (single 2048 gcd) ; 3155 * (2048 / 8) == 807680 byte ; ☝︎

a111: Logged on 2019-06-06 15:46 asciilifeform: loox like diana_coman also used 2048b primes in test of diana_comantronic system ; 'A relatively short test run obtained 40 random primes of 2048 bits each in 13274 seconds in total (3.7 hours) meaning on average 331.85 seconds per prime (~6 minutes)' and diana_coman's 'winner' gets 16 shots; but caveats, a) we have diff iron

asciilifeform: given room temp fg @ 7.5 kByte/s , this is ~119 and ~105 sec. of fg , respectively, in avg case

asciilifeform: given that asciilifeform does not buffer fg ( exercise for reader: why not? ) these waits do not happen in parallel with the computation, and by all rights ought to expect that a live-fg test (not yet performed) can be expected to take longer by the above interval, vs. the bottled-entropy test pictured in yest. thrd.

asciilifeform: ( btw mircea_popescu in old thread correctly intuited that you Do Not Want to buffer an fg, but i dun think either of us ever wrote explicitly why )

asciilifeform: mircea_popescu: in retrospect, prolly we oughta have made'em daisy-chainable. w/out bufferism it is somewhat tricky to use multiple fg to get fatter stream.

mircea_popescu: you want industrial thoroughput for some kind of scientific megaapplication, use the crystals.

mircea_popescu: for the needs of desktop crypto this is perfectly calibrated.

asciilifeform: for systems like mircea_popescu's eulora rsa tho, possibly one'd want >7.5kB/s , but possibly also suffices.

mircea_popescu: yes well, just as soon as someone wants to do all-visible-celestial-bodies montecarloisims / novel protein folding etc -- can pay you to do the lyso array

asciilifeform: lyso-fg is about 20bux moar unit cost than classical. but i dun have the free hands atm.

asciilifeform: primary win imho from a hypothetical 'fast fg' isn't even any of this, but that it is faster to test

asciilifeform: ( i test mine when plugged in from shelf , can't speak for other folx )

asciilifeform: ... at any rate, 'peh' will continue to use the single-fg eater (as depicted in ch8, it hasn't changed) unless someone shows very good reason why it ought to be made moar complex and eat >1.

asciilifeform: ( and, readers will recall, it eats strictly on-demand, when '?' operator invoked, and uses no threads or any bufferism beyond 'fill this here fz from /dev/wherever' .

asciilifeform: since mircea_popescu mentioned 'lyso-fg' -- ftr if mircea_popescu ever comes to conclusion that we oughta bake one, it's all drawn up.

asciilifeform: ( the 1 annoying aspect of lysotronic fg as currently drawn, is that it gets the +45v bias voltage for the detector, from batteries, as asciilifeform does not know of a 'rng safe' method to generate it from +5 without oscillators )

asciilifeform: it needs veehehery small current, on the order of a coupla microamps

diana_coman: asciilifeform: my test was with bits directly read from fg, not buffered (if it wasn't clear); also fwiw further timings and tests are still on my list - it's just that my hands are/have been rather overflowing with eulora

asciilifeform: diana_coman: right, this was my inference from the orig article, your run time includes the wait for fg to emit

diana_coman: on the bright side, ffa is such a contrast to euloran client to make digesting FFA a break in itself

asciilifeform: ffa is much smaller , indeed, item than diana_coman's 3d lib dig

diana_coman: it's much more sanely structured so one CAN actually understand it, you know; that's the main thing; not to mention it's ...worth getting it into one's head to start with; anyway, unless I end up writing a blogpost on "why size is really not all re code", it's probably enough to say it's saner really.

asciilifeform: reminds me : current loc (sans blank lines) : 4010 (libff) , 2087 (peh)

asciilifeform: diana_coman: this was the design objective, yes. is meant for people to eat.

asciilifeform: ( re eulora, as i currently understand diana_coman's main work atm is an excavation, similar to trb circa '15, of heathen coad )

diana_coman: funnily enough (though quite usual in my xp) precisely after I complained re the left/right confusion, I had now the pleasant surprise to find out that meanwhile my brain sorted it really; I suspect it was more that initially I did not realise that there was a mismatch rather than the actual changing of convention; getting back to it was fine.

diana_coman: asciilifeform: bundled (excavation + replacement + change of quite fundamental stuff) so possibly yes.

diana_coman: just for any log readers around: this is not yet the 3d lib at all; it's JUST the planeshift client really, still.

diana_coman: btw asciilifeform ch14 and esp the 14Abis (the bounds proof for barrett's reduction) were a pleasure to read.

asciilifeform: ty diana_coman . happy to hear that it was edible ( iirc yours is the 1st report. )

asciilifeform: barrett is admittedly slightly tricky to 'load into head', but it's where we beat the living shit out of koch et al speedwise (for some reason they never bothered to bake a barrett, even tho method was known since '86