asciilifeform: 'ivory' + os, on other hand, was extremely polished product (unequaled dev environ, multi-year uptime, etc) , so my interest is specifically there.

asciilifeform: 'cadr' is quite well documented, but not particularly informative re 'ivory', which built after more than decade of refinement/simplification

asciilifeform: amberglint: it did not result in very useful picture ( the 'production' model of ivory has thick metallic lid which my tube does not penetrate. ) xray was bought to photograph the 'macivory' pcb tracks. but since this, i was given a pinout (not published yet, cuz again short on time)

asciilifeform: i have 5 samples of 'ivory' (incl. the 1 installed in working machine) , still not enuff to throw around without thinking ☟︎

asciilifeform: little flex pcb thing, inserts under the 'ivory' and brings the signals out to analyzer. ☟︎

asciilifeform: 1st time i booted up that 'ivory' bolix, it actually listed a coupla 'and here ecc triggered, corrected x to y' when revving up

asciilifeform: !#s bolix ivory

asciilifeform: in other noose, asciilifeform nao has what appears to be full pinout, timing, bus sequence, register, init magic, interrupts, etc. spec for bolix 'ivory' (won't, presently, say from where, unless source wants to be cited, plox to write in. ) ☟︎

asciilifeform: in unrelated 'for laters' noose, asciilifeform got hold of 2 moar 'ivory' cpus (brings total # of samples in asciilifeform's house o'horrors to 5 )

asciilifeform: z80 of course not only costs fiddycents, but ~still made~ , and so has been photographed errywhere, prolly even in zimbabwe. whereas there's maybe 50 'ivory' remaining total. ☟︎

asciilifeform: density of 'ivory' is around 4x that of z80.

asciilifeform: 'ivory' aint much transistorwise by current-day measures, but at ~370,000 is slightly moar than its contemporary i386, and defo bigger than anyffing afaik that amateurs anywhere reversed to date.

asciilifeform: ( and , for completeness, a second, http://www.nosuchlabs.com/pub/bolix/pic/ivory_top_a_super.jpg , http://www.nosuchlabs.com/pub/bolix/pic/ivory_top_b_super.jpg , whynot . again ~300MB ea.! )

asciilifeform: i'ma share a pair of example inputs: http://www.nosuchlabs.com/pub/bolix/pic/ivory_bottom_a_super.jpg http://www.nosuchlabs.com/pub/bolix/pic/ivory_bottom_b_super.jpg (warning: 300MB! each!) . ☟︎

asciilifeform: phf: so what was it, you promised 'i'ma only leak the bits re ivory' ?

asciilifeform: 'sapper errs 1ce', could take another 10y to get a 2nd 'ivory' if i kill this one.

asciilifeform: incidentally, i had potentially interesting notion: a modern-day digicam ccd, of e.g. 25 'megapixel', is just about dense enuff that one could take a meaningful xray of the ivory die, if one could be found that responds to xray.. ☟︎

asciilifeform: phf: i am thinking, instead of burning time on emulating 'nubus' and ancient crapple, when working (from mac-side pov) clone exists i'ma drive it directly via onboard ft2232hl or similar ( the mac-side src , i've had for yrs, and it's quite straightforward re what is done to the card; once i know what specifically happens on the card when nubus regs are written, may as well do it from pc directly to/from the ivory cpu via usb-to-5v-f

asciilifeform: pretty sure there's custom bolix ic other than ivory in xl

asciilifeform: 'ivory' board, however ( http://btcbase.org/log/2018-11-28#1875835 ) is moar or less exactly the correct item, it's a minimal set of support logic for the chip, and piggybacks on ancient 1980s crapple ( box that's been reversed nine ways to pluto and 100% emulatable today ) for i/o ☝︎

asciilifeform: pretty painful proposition, to pick up a $5k board simply to read off the GALs and xray the pcb. but unless somebody has one to lend for the purpose, i dun see how that dig will ever move beyond 'i have these here 2 ic's' without access to orig ivory pcb.

asciilifeform: !Q later tell phf loox like dks is selling another! 'ivory' on lulzbay >> https://archive.is/Bkm8E ☟︎

asciilifeform: that very 'ns' which amberglint cracked, and on which 'ivory' was baked, was 20k loc !!

asciilifeform: all that presently exists in the wild is 1) small number of actual 'ivory' machines/cards, where could put cloned ic 2) the software/os stack

asciilifeform: mircea_popescu: he does : http://www.loper-os.org/pub/bolix_ivory.jpg object dun look removable, but rather like it was inserted from other end

asciilifeform: 'Figure 24 show some statistics about Ivory. The size of the chip was about a centimeter-and-a-half on a side in the earliest version which was done using 2-micron CMOS technology. The first commercial version will be done in a 1.6 micron process and will be about a centimeter on a side, as is the TI chip. This makes it a very large, (but not impossibly large), chip to fabricate. There are about 255,000 actual transistors on the die.

asciilifeform: 'It is of tiie same complexity as an Intel 80386 or a Motorola 68030. Figure 21 is a photomicrograph of the chip (chip people always seem to need to show their chip pictures). The Ivory chip is more or less divided into three horizontal slices; the top is the datapath and stack cache, the middle is the control system, and the bottom is the memory interface. In the middle of the control system is a very large microstore, again compara

asciilifeform: http://www.loper-os.org/pub/bolix_ivory_pinside.jpg ☟︎

asciilifeform: >> http://www.loper-os.org/pub/bolix_ivory.jpg << subj

asciilifeform: 'ivory' is somewhat unique case, imho, i dun expect to ever run across another 1980s item that even begins to merit this treatment.

asciilifeform was never able to find a high-res photo of ivory-3. perhaps phf has ?

asciilifeform: https://archive.is/jvkUK << photo of ivory-2 . appears to contain 1 rom, prolly no moar than a 32kB, going by the pin count.

asciilifeform: phf: tho imho ideally, we'd still get a decap of #2 after #1 is happily in a ersatz-ivory board and booting the full archaeological soft stack

asciilifeform: ( for n00bz -- decapping, even if done perfectly, is irreversible process, we dun know how many layers even in ivory, and to take pic of n+1st layer gotta dissolve the n-th... )

asciilifeform: i thought it was specifically re ivory

asciilifeform: literally erry other ic on the 'ivory 3' board is stock ( save possibly for a GAL or 2 )

asciilifeform: i wouldn't consider ever to put floatism in a new design, but for compat with the vintage soft one would need it ( fortunately unlike 'ivory' the weitek is documented to death, i even have a copy of the datashit from ancient ru stash )

asciilifeform: even ivory had a buncha similar ugly, iirc it still used external weitek thing (same as in yours and mine 486) for arithmetics

asciilifeform: !Q later tell phf in bolix lulz : loox like dks relisted that 'ivory' for 3rd time nao... apparently not so many eager-worth-of-$5k archaeologists any moar

asciilifeform: but yes, i'm not esp surprised that the derps 'working' on 3600. given as 1) 'ivory' needs ~20-50 k $ in either chinese microscopist, or nixon's burglars , take yer pick 2) 3600 is massive forest of idjit hackolade, which these people seem to ~like~, the way scarab luvvs shit

asciilifeform: amberglint: iirc he found pinouts. not crown jewels, tho could be of use in ~very slowly~ getting sumthing useful from the 2 'ivory' crystals i traded a 3620 for

asciilifeform: mircea_popescu: nope, tho possibly phf might , he appeared during prev bolix/ivory thread

asciilifeform: amberglint: when i went to his house 10y ago, offered to him to make fpga-ized 'ivory' , 'pro bono', if he'd only cough up with what. answr was approx 'can't , unless my master permits, and he wouldn't' ☟︎☟︎

asciilifeform: i suppose ns pill will come in handy if anybody ever unearths the ivory src

asciilifeform: what i wanted re ns, is a) the 'ivory' src b) src for ns itself, to pillage the algos

asciilifeform: phf: in other lulz, yet another ivory on lulzbay

asciilifeform: i recall reading once about an american 'sinologist' type who started by collecting the pipes ( they were often quite ornate, ivory, rhino, exotic woods ) and ended up... reconstructing actual opium den, somewhere

asciilifeform: and iirc phf has a working ivory -- worth trying there also

asciilifeform: somebody , maybe same old man, has a bookcase full of typewritten design apocrypha, memoes, etc. from ivory group. he ain't sharing either.

asciilifeform: think, somebody, somewhere, has the source code to ns. and the ns netlist , with comments, to the ivory.

asciilifeform: i do not have, nor ever had, a working 'ivory' of any description.

asciilifeform: fromloper: i only have two 'ivory' chips, and ideally would like to leave one intact , for active test . iirc phf also has 2 to use.