log☇︎
500+ entries in 0.052s
diana_coman: 2. FFA use/integration
feedbot: http://ossasepia.com/2019/07/29/overview-of-ffa-ch1-ch19/ << Ossa Sepia -- Overview of FFA Ch1 - Ch19
diana_coman: mp_en_viaje: I suppose there is also the same question applied potentially to the FG and FFA then; but fwiw I digested FFA as well as I could and I still think it is a gem, perhaps a pearl better put: quite the result of successive spinnings/polishings
diana_coman: well, there IS on one hand the FFA gem that has however come after the initial projection that he'll do consulting for smg and so I'll have what to use rather than write eucrypt with the salvaged-mpi (though he salvaged that at least)
mp_en_viaje: so no, i couldn't give less of a shit as things stand right now if ffa is "properly completed" or not ; nor if "more material" is made available as unqualified scribbles on trb code.
ossabot: Logged on 2019-10-08 21:58:50 asciilifeform: himself 'studying ffa' , load it again into head !
diana_coman: ftr I consider FFA mandatory for #o & rather entry level at that but so far people still need to excavate selves to get to that entry level even.
asciilifeform: i do not know whether diana_coman is still taking people into her school, but for a ffa n00b it is imho a very desirable school.
mod6: <+asciilifeform> mod6: if you start to study ffa, dun hesitate to ask asciilifeform to help as you go in. << will do!
asciilifeform himself 'studying ffa' , load it again into head !
asciilifeform: mod6: if you start to study ffa, dun hesitate to ask asciilifeform to help as you go in.
mod6: I have a whole list of things I wanna do; actually. Like, finally spend the time I need on all things FFA.
snsabot: Logged on 2019-09-25 22:26:36 asciilifeform: in related grrs, asciilifeform recently attempted backport of mircea_popescu's selector knob to own wp, but broke teeth cuz it dun know how to cross paragraphs, and i have '9000' multi-para selections, esp. in ffa series where coad
asciilifeform: in related grrs, asciilifeform recently attempted backport of mircea_popescu's selector knob to own wp, but broke teeth cuz it dun know how to cross paragraphs, and i have '9000' multi-para selections, esp. in ffa series where coad
asciilifeform: meanwhile, from the lulzcollector traps : usg nuke ministry's attempt at ffa. ( for bonus laffs, they apparently have a shithub, fulla misc. opensores rubbishes ...)
ossabot: Logged on 2019-09-14 02:24:17 asciilifeform: it was enheartening tho when diana_coman ate ffa 1-19 and commented. apparently it was edible by at least 1 . (even if took a maffs doctor )
ossabot: Logged on 2019-09-14 02:24:17 asciilifeform: it was enheartening tho when diana_coman ate ffa 1-19 and commented. apparently it was edible by at least 1 . (even if took a maffs doctor )
asciilifeform: it was enheartening tho when diana_coman ate ffa 1-19 and commented. apparently it was edible by at least 1 . (even if took a maffs doctor )
asciilifeform: diana_coman: primarily -- the intrinsic difficulty of reading. reader is stuck parsing out the ast using bare hands. it is possible to make the work easier, and i tried e.g. here . but remains painful for most folx.
asciilifeform: on ordinary x86 -- state-of-art today is bvt's inline asm where 0.25s per 4096bit modexp .
asciilifeform: http://logs.nosuchlabs.com/log/trilema/2019-09-02#1933202 << ffa test tapes already weigh 150+MB collectively
asciilifeform: e.g. ffa/peh dun use threads at all.
asciilifeform: ^ was actually the 1st nontrivial proggy asciilifeform wrote in ada, just prior to picking up ffa
asciilifeform: BingoBoingo: from this pt i'ma follow the prescription mp gave during last 'wai slow' thrd -- leave it alone, there aint much to be done re to speed it up ( as imho conclusively showed that the pipe is to blame ). will move on to other matters ( import of historic logs ; fix of reported catastrophic bugs, incl. the resync of numeration mp asked for ; and then to get back to ffa ! )
asciilifeform: iirc diana_coman already 1nce caught this and fixed, in ffa tree
asciilifeform: i suspect the glue for pg alone (i.e. if pg disgorges eggog, to produce proper ada-compat. exception, rather than crash), not to mention utf8ism, would weigh like ten ffa's
asciilifeform: this is the 'flip side of the medal' -- sometimes the given abstractions are not a good fit for $problem. picture e.g. attempt to ffa in commonlisp, where you'd have to bend over backwards to make sure you ~never~ cons
mp_en_viaje: only nowadays, with ffa and eucrypt & rest of stuff as they are to provide a basis, im starting talk of things such as http://trilema.com/2019/trilema-goes-dark/#comment-130709
asciilifeform: possibly the source of confusion for the drosophila was the fact of ffa series being tagged, among other strings, 'bitcoin' , lol
asciilifeform: arguably ffa series is moar primarily a handbook of how to build proper 'wide' arithmetizer, than pc proggy.
a111: Logged on 2019-07-29 09:44 feedbot: http://ossasepia.com/2019/07/29/overview-of-ffa-ch1-ch19/ << Ossa Sepia -- Overview of FFA Ch1 - Ch19
asciilifeform: http://btcbase.org/log/2019-07-29#1925423 << quite good summary! asciilifeform added nitpick . ☝︎
diana_coman: and yes, FFA is a gem.
mp_en_viaje: reading diana_coman 's executive summary really brings into focus the beauty of stan's ffa. fucking gem, what.
feedbot: http://ossasepia.com/2019/07/29/overview-of-ffa-ch1-ch19/ << Ossa Sepia -- Overview of FFA Ch1 - Ch19 ☟︎
asciilifeform: BingoBoingo: principal reason why i went awol from ffa to shit out this horror , was the http://btcbase.org/log/2019-07-19#1923477 burning q ☝︎
asciilifeform: in e.g. http://www.loper-os.org/pub/ffa/hypertext/ch19/fz_divis__adb.htm , links errywhere
a111: Logged on 2019-07-18 14:29 feedbot: http://bvt-trace.net/2019/07/unrolled-x86_64-assembly-multiplication-for-updated-to-ch-14-ffa/ << bvt's backtrace -- Unrolled x86_64 Assembly Multiplication for updated to Ch. 14 FFA
feedbot: http://bvt-trace.net/2019/07/unrolled-x86_64-assembly-multiplication-for-updated-to-ch-14-ffa/ << bvt's backtrace -- Unrolled x86_64 Assembly Multiplication for updated to Ch. 14 FFA ☟︎
bvt: hello. I will try to finish the ffa work from the workplan over next two days. i had some meatspace interference that stole a few work hours.
asciilifeform: in re ffa : ch20 and most of 21 written. but all of it paused atm , making detailed map/plan.
asciilifeform: will genesis like civilized people when myself have a free hand (atm do not, still writing piece re what remains to be done in ffa and how long takes etc )
a111: Logged on 2019-06-30 16:28 BingoBoingo: As it is, there's no mention of the MIPS experiment on your blog. Last post is FFA Chapter 19. I suspect if put together a weekly or every other weekly "This is what I've done, here's what I am chewing on, here is what has yet to be bit off, etc" would calm some nerves
BingoBoingo: As it is, there's no mention of the MIPS experiment on your blog. Last post is FFA Chapter 19. I suspect if put together a weekly or every other weekly "This is what I've done, here's what I am chewing on, here is what has yet to be bit off, etc" would calm some nerves ☟︎
asciilifeform: e.g. ffa , if counted from when asciilifeform first started the notebook and blew dust off knuth, is 3 y.o. nao.
a111: Logged on 2019-06-22 13:39 asciilifeform: http://btcbase.org/log/2019-06-22#1919162 << can't speak for other folx, but asciilifeform wrote 7 ch. of ffa thus far in '19 (featuring errything to do with primes, and all of peh, plus tutorials on the latter) and expect to wrap up 8th (# 20) before june is out.
a111: Logged on 2019-06-30 07:07 asciilifeform would like to bake new batch of sellable goodies , after we release peh/ffa
asciilifeform would like to bake new batch of sellable goodies , after we release peh/ffa ☟︎
asciilifeform: in the flagship branch, it comes into play as base case when 8 or fewer words in multiplicand
asciilifeform: ( possibly after ffa, we do a 'mips in 1 instruction a day' likbez series...? )
a111: Logged on 2019-06-23 18:45 feedbot: http://bvt-trace.net/2019/06/unrolled-assembly-squaring/ << bvt's backtrace -- Unrolled x86_64 Assembly Squaring for Ch. 12 FFA
asciilifeform: http://btcbase.org/log/2019-06-23#1919631 << iirc i pasted 100+ drafts of various ffa moving parts, before genesised any ☝︎
feedbot: http://bvt-trace.net/2019/06/unrolled-assembly-squaring/ << bvt's backtrace -- Unrolled x86_64 Assembly Squaring for Ch. 12 FFA ☟︎
mircea_popescu: asciilifeform, suppose fellow wants ffa. how does he pick the one he needs ?
mircea_popescu: both the mips and the x86 ffas pretend to ffa, do they ?
mircea_popescu: if tomorrow we have a ffa-for-pc and a ffa-for-mips alternative, there WILL BE AN IFDEF.
mircea_popescu: all this said, i don't believe if someone wrote an ebuilds for say ffa they'd thereby degrade ffa.
a111: Logged on 2019-06-22 16:41 bvt: mircea_popescu: i guess that there two components in your question. the easy is ffa: is agree that my ffa-related output is underwhelming for a lord, however 1) i don't expect that it will eat much more time (at ch.14 i'll see if there is really need for further asming, the anticipated answer is "not really") 2) i find speedup from asming generally useful, so i don't think the time i spent on it is wasted.
asciilifeform: http://btcbase.org/log/2019-06-22#1919279 << ftr i do not think it is waste. i expect that most 'realtime' applications (e.g. 'gossip') will want the asmistic ffa. ☝︎
bvt: after ffa I will have a look at other things (like ripping out kernel rng, having another look at gnat-arm64 internals, as it seems there is no ongoing work on this front atm). i expect to get something useful as a result, and maintain it in long term.
bvt: mircea_popescu: i guess that there two components in your question. the easy is ffa: is agree that my ffa-related output is underwhelming for a lord, however 1) i don't expect that it will eat much more time (at ch.14 i'll see if there is really need for further asming, the anticipated answer is "not really") 2) i find speedup from asming generally useful, so i don't think the time i spent on it is wasted. ☟︎
a111: Logged on 2019-06-22 13:49 asciilifeform: http://btcbase.org/log/2019-06-22#1919184 << sad to hear that phf is in this condition. hey phf can you at least throw bring the ffa patches/sigs on btcbase up to date ? i think i asked 3x already... surely this takes less than 10min on your end
asciilifeform: http://btcbase.org/log/2019-06-22#1919166 << trinque sweated out a draft cuntoo, which sadly i have not had chance to test in anger. i have a physical box that is destined for it , when get chance, and also will be porting it to the sim-mips, ditto. but i promised to mircea_popescu not to undertake any elaborate works until ffa suitable for 'discard gpg' and extension to other (gossipd, trbi, what else is waiting on it) paths ☝︎
asciilifeform: http://btcbase.org/log/2019-06-22#1919184 << sad to hear that phf is in this condition. hey phf can you at least throw bring the ffa patches/sigs on btcbase up to date ? i think i asked 3x already... surely this takes less than 10min on your end ☝︎☟︎
a111: Logged on 2019-06-22 07:09 bvt: i do expect that 'saltmine season' is over now -- i've been working on asming karatsuba squaring (ffa ch.12b) over this week, post expected tomorrow.
asciilifeform: http://btcbase.org/log/2019-06-22#1919162 << can't speak for other folx, but asciilifeform wrote 7 ch. of ffa thus far in '19 (featuring errything to do with primes, and all of peh, plus tutorials on the latter) and expect to wrap up 8th (# 20) before june is out. ☝︎☟︎
mp_en_viaje: bvt, so what's your idea going forward anyway ? are you basically going to be doing some ffa as you fiat overlod permit now and again, so by the end of 2019 you can look at having completed chapter 22 or somesuch, and that'll have been it ?
bvt: i do expect that 'saltmine season' is over now -- i've been working on asming karatsuba squaring (ffa ch.12b) over this week, post expected tomorrow. ☟︎
asciilifeform: i.e. ffa-style mux for the sim-branch clockwork.
asciilifeform: stack is also statically allotted on warmup, a la ffa.
asciilifeform: ffa is much smaller , indeed, item than diana_coman's 3d lib dig
diana_coman: on the bright side, ffa is such a contrast to euloran client to make digesting FFA a break in itself
asciilifeform: mechanism, for thread-completeness.
a111: Logged on 2019-06-06 07:11 mircea_popescu: http://btcbase.org/log/2019-06-05#1917355 << i do not think data should be included with ffa ; while i do believe data should be included with bitcoin. the reason for the split is that i perceive a difference between design and implementation / algorithmics and code, so following. if you see peh as a sort of "packaging of this theory for consumption" then maybe worth including,
mircea_popescu: specifically with peh, as distinct from ffa.
mircea_popescu: http://btcbase.org/log/2019-06-05#1917355 << i do not think data should be included with ffa ; while i do believe data should be included with bitcoin. the reason for the split is that i perceive a difference between design and implementation / algorithmics and code, so following. if you see peh as a sort of "packaging of this theory for consumption" then maybe worth including, ☝︎☝︎☟︎
asciilifeform: mips specifically is compact enuff in its definition that one could write a demonstrably-inescapable sim for it. would weigh approx what peh (sans ffa) weighs, i suspect.
pehbot: asciilifeform: FFA Ver: 00000000000000000000000000000000000000000000000000000000000000FD
asciilifeform: !A V[FFA Ver: ]#
asciilifeform: http://btcbase.org/log/2019-05-29#1916180 << direct or inverse proportion ? in e.g. ffa, the # of 'can be changed an' still work' is ~0 ☝︎
a111: Logged on 2019-05-17 14:35 asciilifeform: http://btcbase.org/log/2019-05-17#1914368 << consider ada. ( ffa series imho makes a decent ada tutorial, and so does diana_coman's series )
diana_coman: asciilifeform: for one thing you already have skills he doesn't yet seem to have and for the other thing you constantly say it yourself that at times there is no juice left for ffa /similar.
asciilifeform: http://btcbase.org/log/2019-05-17#1914368 << consider ada. ( ffa series imho makes a decent ada tutorial, and so does diana_coman's series ) ☝︎☟︎
asciilifeform: stjohn_piano_2: this sorta thing is almost universal afflication of young folx. i can see why you choked on ffa tho, it has literally no 'parts that can be safely skipped'
asciilifeform: stjohn_piano_2: ever read ffa series ?
asciilifeform: amberglint: once i wrap up ffa, will proceed to bake a probe for the working machine, to get oscillogram of couple minutes of bootup/run ☝︎
asciilifeform: bvt: with straight (no asmisms) 64bit ffa, i found ~exactly 2x slower on x60 than on the 2393
asciilifeform: this is trivially verifiable experimentally : build for 32bit ( 'MachineBitness' in http://www.loper-os.org/pub/ffa/hypertext/ch18/iron__ads.htm ) and get ~exactly 2x slower.
asciilifeform: in the parlance of seymor cray et al, ffa is 'cpu-bound problem'
asciilifeform: speed of ram, interestingly, has ~0 effect on ffa timing, given as the working set ~always fits in cache
asciilifeform: imho you defo wouldn't want to ffa on a post-2013 amd, or any intels etc aha
asciilifeform: mp_en_viaje: pre-barrett ffa , when rsaing, spends most of wallclock time in fz_mod , is all. which dun multiply. in 14+ , i fully expect 2-3x boost from asmism (possibly higher, with massage w/ magnifying glass for pipelineism etc )
asciilifeform: i fully expect that the asmistic branch of ffa will be sewn out of bvt's item
feedbot: http://bvt-trace.net/2019/05/unrolled-assembly-comba/ << bvt's backtrace -- Unrolled x86_64 Assembly Comba for Ch. 11 FFA
asciilifeform: observe that (despite considerable sweat) asciilifeform was not even able to write ffa , on gnat, 'using only what is standard'
asciilifeform: this is sorta the basic idea in e.g. ffa -- 'what's the simplest mechanism that will actually suffice'
a111: Logged on 2019-04-23 23:06 asciilifeform would luvvv to read 'ffa-style' incarnation of such a work, where the chip is 'built up' from empty space in ~hour-long chunks. but prolly this won't exist until asciilifeform writes 1..
asciilifeform would luvvv to read 'ffa-style' incarnation of such a work, where the chip is 'built up' from empty space in ~hour-long chunks. but prolly this won't exist until asciilifeform writes 1.. ☟︎
mp_en_viaje: AND, with all the experience from ffa, you actually got what you need as a basis to actually make that megaspire work.
mp_en_viaje: alright, well, month's not bad -- by the time you're done with ffa you can start a mega-bolix series.