asciilifeform: how the living fuck does one reverse ciphered blob of instructions for unknown instruction set, consisting mainly of writes to registers of unknown purpose (mostly dealing in such things as setting PLL frequency etc) without physical peek ?
asciilifeform: the odd thing is that these are ~all in the l0gz
asciilifeform: gabriel_laddel_p: you're missing a couplea zeros
asciilifeform: 'i forgot my hat on top of mt everest, how much would you suppose you must charge, asciilifeform , to fetch it for me ??'
asciilifeform: 'hmm was that 1 matrox in '98 any good? or was it a hercules..'
asciilifeform: gabriel_laddel_p: this would take a rather long time. and i'm not convinced that the result would be very interesting
asciilifeform: it uses conductive, rather than convective, cooling, however, could be problem in a hot dc
asciilifeform: you can fit iirc 8 of'em in a traditional 1u
asciilifeform: apu1 racks, i even have the mechanical bracket somewhere
asciilifeform: their approach to engineering is 'junkyard wars' -- 'look mother, what i made from dead possum and string!'
asciilifeform: trinque: such marvels as, e.g., 'web20', javascript, etc. came directly from allowing teenaged scum to have whatever involvement in design aside from cleaning the shop at closing time each day.
asciilifeform: not really pertinent to hardware thread imho.
asciilifeform: trinque: 'ban ip' in context referred to 'ban port', 'reject dns', or whatever other machinations.
asciilifeform: that's an ip stack knob, not nic knob
asciilifeform: but relates, because 'make isolating nic for x86' is not a solvable problem, it is a gabriel_laddelism -- for one thing, such a device is already available, it is called usb nic. but is hog-slow, for fundamental reasons (polled i/o)
asciilifeform: ditch the 'insulators'. asciilifeform asks for 1 thing and 1 thing only hardwarewise -- massive open fpga, coupled to dram slots.
asciilifeform: ( how the hell did anyone even port coreboot to apu2 ?? )
asciilifeform: which makes firmware dev problematic.
asciilifeform: trinque: right. apu2 has a fritzed cpu however.
asciilifeform: the unfortunate bit is that 1) afaik it does not exist . 2) the idiot pc architecture gives your nic - regardless of who made it -- access to ALL OF RAM
asciilifeform: trinque: one nice thing is that the apu1 had 3 pci-m holes. so if a proper nic existed, you could potentially install it.
asciilifeform: costs considerably more, and often hard to get ( the swiss d00dz make'em in batches, i had to wait ~month for mine back when i got it ) than konsoomer crud
asciilifeform: otherwise identical chipset, and same one as apu1
asciilifeform: and presoldered sage header. (doesn't come with a sage box, however, unlike 'gizmo1')
asciilifeform: 'gizmo2' is a slightly moar 'human' box, it has hdmi port even
asciilifeform: but it is not long for this world. i looked yesterday, it is still offered for sale, but who knows for how long.
asciilifeform: apu1 was the imho interesting box -- the first, afaik, and only truly usable workstation x64 with PUBLISHED mobo schematics
asciilifeform: trinque: apu2 will run a node. so would a slightly cheaper intel box, and with - turns out - same amount of 'nsa inside'.
asciilifeform: ( there isn't, i will add, a 'lowest common denominator' solution for those. )
asciilifeform: trinque: on some boxes, 'first boot' is trivial. on others, not. for instance, my 3ware raid boxen won't load even 1 block after boot loader without the correct mod.
asciilifeform: my contention is that this functionality should never have been modularized / seen as optional.
