asciilifeform: carmichaels dun have much to do with m-r
asciilifeform: i recall digging for this, and instead finding conjecture that no canhaz
asciilifeform: nah i'd be satisfied with 'here's n, and here's 32 unequal m-r liars for n'
asciilifeform: mircea_popescu: if you recall where, plz link
asciilifeform: mircea_popescu: interestingly, iirc even to bake 32 liars (having control of the witnesses, presupposes) that 1) all diff 2) work for some n -- is open problem atm
asciilifeform: ( high in comparison with the naked 4**-x that is )
asciilifeform: or rad event in yer irons, also fairly high prob regardless of how you baked rng
asciilifeform: i.e. if you run somehow 1 per sec, will eat a coupla dozen moon-size asteroids prior to getting misfire..
asciilifeform: 4**(-32) ~= 5.4210e-20 , probability of fatal misfire for 32 rounds of m-r
asciilifeform: ( for n00bz -- prime gen parallelizes, obv., over as many cpu as you have. but ffa per se does not use threads, tho it can be run inside threads with no added headache )
asciilifeform: ( same params, i.e. pronounced prime if passes m-r with 32 rng-shat witnesses )
asciilifeform: fwiw 'speed of ffa' for applications involving modexp ( rsa keygen & enc/dec ) hinges ~entirely on speed of the multiplier unit in $iron .
asciilifeform: i expect that as moar folx eat ffa, we will have moar empirical figs to compare.
asciilifeform: ftr rk is effectively 2x, as measured, slower than this gauge box , on ffa.
asciilifeform deliberately uses oldest opteron in the torture room as 'standard ffa gauge' , if wasn't obv.
asciilifeform: ( the latter not used for any of the published tests, but to simply gauge the effect of 'modern' x86isms . fwiw does have constant-time mul etc )
asciilifeform: i also built at various points on a heathen i7 (2.8G) and there runs ~2x faster across the board, ftr
asciilifeform: ( mine, used as ref for all ffa figs, is opteron 2393SE @ 3.1G )
asciilifeform: that's substantially faster box than mine ftr
asciilifeform: diana_coman: no rush, we'll find out how compares on given $iron when you get around to 18 .
asciilifeform: aanyways i expect that when diana_coman gets to eating ch18, we can compare properly .
asciilifeform: ( in asciilifeform's demo, proggy halts when candidate passes 32 rnds )
asciilifeform: tho unless i misread, that ^ figure included 16 iterations thereof, rather than 1
asciilifeform: ( also had 'an average of 9.78 seconds per M-R run' , on again i dun presently know which irons )
asciilifeform: '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).'
asciilifeform: ( when 1st started ffa, was quite ready to live with 'keygen takes hour+' ftr )
asciilifeform: tbh i cannot however think of when or where i would say 'ugh, 10min is too long for baking a 4096b key'..
asciilifeform: ( naturally at no point will the algo complete faster than the time eaten by the given # of m-r rounds specified )
asciilifeform: to reduce the # of sad candidates that m-r ends up eating
asciilifeform: for thread-completeness, will add that it is possible to speed up the algo by giving not only 1 'primorial up to width' but it and then whichever number of product(n+1th-prime ... n+i-th prime) that yet again fit in the width, and so forth, for any desired # of small primes, for the initial sieve
asciilifeform: ( so i expect will be ~7 min. or so per 4096b key, on iron of this horse )
asciilifeform: granted for actual key will want also random e , and will reject some p,q,e triples based on the obvious rule
asciilifeform: mircea_popescu: afaik 'granger' is today an agricultural gear supplier in usa . ( named after 19th c union thing of some variety )
asciilifeform: any given piece of peh tape cannot refer to anything that lies to the right of it, ergo 1 walk suffices
asciilifeform: it oughta be , after study, entirely obvious from single eyeball walk why each 1 does what it does
asciilifeform: but imho 1st it oughta become 'second nature' for the n00b , to read the examples with naked eye
asciilifeform: at some pt i'ma publish the emacs majormode for editing pehs. it is ~much~ easier with colours etc
asciilifeform: Mocky: most battlefield pehtapes , i expect, will be ~much~ easier to read than these ( e.g. will include comments, an' meaningful names ) . these, are simply micro demos that fit in log line
