--- Log opened Mon May 30 00:00:37 2022
00:19 -!- mirage3352367 [~mirage335@2600:1003:b132:fc2e:c310:e83:e468:d1b0] has joined #hplusroadmap
00:23 -!- mirage33523 [~mirage335@64.79.52.86] has quit [Ping timeout: 252 seconds]
01:22 -!- mirage335236795 [~mirage335@64.79.52.86] has joined #hplusroadmap
01:26 -!- mirage3352367 [~mirage335@2600:1003:b132:fc2e:c310:e83:e468:d1b0] has quit [Ping timeout: 252 seconds]
05:21 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::93] has joined #hplusroadmap
05:34 < muurkha> maybe more relevantly craig wright is the person who fraudulently claims to be satoshi nakamoto
05:35 < muurkha> maaku: oh, I see what you mean about mimblewimble
05:37 < muurkha> with respect to jake, jake basically screwed over a large fraction of the people he came into contact with.  sexually assaulting a couple of them was not the major thing that led those people to hate him, but it was more legible to the public than the other things he did
05:41 < muurkha> everyone who knows jake personally can tell you how he frequently told people what they wanted to hear instead of what was true
06:08 < muurkha> lkcl: last year you submitted a 180nm-fabbed LibreSOC prototype to Imec's shuttle service, right?  I'm curious to hear how well it worked, keeping in mind that often it takes several runs through an MPW service to get something that works at all
06:08 < lkcl> muurkha, LIP6's Professor Galayko has done some preliminary tests on the PLL and found it to be operational
06:09 < lkcl> which has additional side-implications: (1) there's no short-circuits (2) the IO pad cell design is working
06:10 < lkcl> he'll be designing a PCB onto which some 100nF decoupling capacitors can be mounted next.
06:10 < muurkha> oh cool!  but it sounds like it doesn't run instructions?
06:10 < muurkha> or you don't know yet?
06:11 < lkcl> we don't know yet because there's only one solderless socket available
06:12 < muurkha> that's very exciting!
06:13 < lkcl> yes. slow, but yes :)
06:15 < muurkha> yeah, shorter feedback loops are maybe the biggest advance that software brings
06:17 < lkcl> we implemented JTAG boundary-scan and also connected up DMI (Debug Memory Interface) and Wishbone read/write over JTAG
06:18 < lkcl> so it's possible to STOP, load-into-memory, STEP/START and then read/dump the regfile(s)
06:18 < lkcl> i added jtagremote to the verilator simulation and made a python unit test and command-line program that is compatible with it
06:19 < lkcl> and also did an openocd test with the jtagremote protocol
06:19 < muurkha> how much real estate did that cost?
06:19 < lkcl> then wired up an FT232 to an ECP5 FPGA and ran the same things
06:19 < lkcl> very little
06:19 -!- Molly_Lucy [~Molly_Luc@user/Molly-Lucy/x-8688804] has joined #hplusroadmap
06:19 < lkcl> some shift registers and a few MUXes per IO pad
06:20 < lkcl> the DMI interface is... 450 lines of nmigen?
06:20 < lkcl> https://git.libre-soc.org/?p=soc.git;a=blob;f=src/soc/debug/dmi.py;hb=HEAD
06:22 < lkcl> basically with the JTAG-to-verilator working and JTAG-to-ECP5 working it's brain-dead-simple to wire up to the actual ASIC and run the exact same unit tests
06:22 < lkcl> only thing is, openocd has no "machine" definition. it's raw JTAG at the moment.
06:22 < muurkha> a thousand transistors or so?
06:23 -!- oxphi [~oxphi@107.181.189.44] has quit [Quit: Client closed]
06:23 < lkcl> probably more like 5 at a guess
06:23 < muurkha> almost free
06:24 < lkcl> DMI has to hook into the register file read/write paths (its own dedicated regfile ports)
06:24 < muurkha> sure, slightly wider muxe
06:24 < muurkha> s
06:25 < muurkha> I guess it would be a painful cost on something like a GA144 in part because of the routing
06:27 < lkcl> you mean this? http://www.greenarraychips.com/home/documents/greg/GA144.htm
06:27 < muurkha> yeah
06:29 < lkcl> then yes and no
06:29 < lkcl> if every individual IC has JTAG then you can just use daisy-chaining
06:29 < muurkha> well, each of the 144 processor tiles on the chip is maybe 5000 transistors
06:30 < lkcl> JTAG has the ability to do "pass-thru" i.e. it was specifically designed for the purpose of testing an entire wafer of ICs long before the dies were cut out and packaged
06:30 < muurkha> and most of the registers aren't connected up to a register file mux; they're basically bidirectional shift registers
06:30 < muurkha> (I mean the architectural registers, not the words of RAM)
06:31 < lkcl> if there are 10 GPIOs you need a... 40-bit shift register and 60 MUXes
06:31 < lkcl> sorry, 2x 40-bit shift registers and 60 (or so) MUXes
06:31 < muurkha> so adding the ability to read those registers via JTAG would involve adding the ability to read the registers at all
06:32 < lkcl> via a straight serial shift operation. it's brain-dead simple
06:32 < lkcl> bit-level shifting on TCK.  TDI goes in on the left, TDO goes out on the right
06:32 < lkcl> chain to the next IC.
06:32 < muurkha> I think the GA144 has 25 GPIOs
06:33 < lkcl> almost laughably trivial and yet really quite powerful, you could have 1,000 ICs in a row and test them all via only 4 wires. ok 5 including GND
06:33 < muurkha> but sure, adding boundary scan to something like the GA144 wouldn't be hard or expensive at all
06:34 < muurkha> I meant adding the ability to do the more invasive kinds of tests you're talking about over JTAG, reading registers the processors don't have the ability to read themselves and that kind of thing
06:35 < muurkha> but for anything where the processors are hundreds of thousands of transistors or more, you might as well
06:35 -!- mirage335236795 [~mirage335@64.79.52.86] has quit [Quit: Client closed]
06:36 -!- mirage335236795 [~mirage335@64.79.52.86] has joined #hplusroadmap
06:36 < lkcl> adding boundary scan for only 25 GPIOs i'd estimate at 200-bits for shift registers (DFFs) and... 150 MUXes
06:37 < lkcl> 4x DFFs per GPIO
06:37 < lkcl> 6x MUXes per GPIO
06:37 < lkcl> 100 DFFs not 200.
06:41 < abecedarius> jrayhawk, i'll follow those links today
06:53 -!- oxphi [~oxphi@107.181.189.44] has joined #hplusroadmap
07:05 < kanzure> https://wiki.replimat.org/wiki/Main_Page
07:28 < fenn> needs more triangles
07:36 -!- oxphi [~oxphi@107.181.189.44] has quit [Quit: Client closed]
08:12 < kanzure> https://www.fastcompany.com/27333/trillion-dollar-vision-dee-hock
08:34 < fenn> tldr plz
08:38 < fenn> "Hock designed the organization (VISA) according to his philosophy: highly decentralized and highly collaborative. Authority, initiative, decision making, wealth — everything possible is pushed out to the periphery of the organization"
08:44 < kanzure> well they agreed to not compete on standardization of card formats and protocols, and then allowed competition on pricing, and then did trademark licensing to share a brand
08:44 < kanzure> was thinking of this because of some work i'm doing on hardware wallets at the moment
08:45 < kanzure> there's a number of crypto hardware wallets that have incompatible/different wire protocols but they all essentially do the same thing (kind of, plus/minus some specialization around setup and how users go about unlocking a hardware wallet)
08:46 < kanzure> and part of the reason why there is no standard protocol for this is because of the competition between altcoins, to the extent that trezor (etc) literally sells a 'bitcoin-only' version
08:46 < kanzure> (which is kind of dumb because obviously the bitcoin-only version has a much smaller marketsize))
08:48 < fenn> "chaordic organizations"
09:02 < kanzure> https://github.com/BlockchainCommons/Airgapped-Wallet-Community/discussions/26
09:08 < docl> I have a plan, which I've been assured is impossible (by multiple smart people) but don't know enough to trust such assurances, to retrieve information traveling away at the speed of light. basically assumes every relevant bit is being massively duplicated in photonic form and by carefuly monotoring of every speck of dust in a given radius you could eventually infer a detailed 3d image of the location 
09:08 < docl> at the center. you would need incredibly good characterization of every photon source in the vicinity (IOW the visible astronomical universe) to characterize and filter out the noise. but since it doesn't require running physics backwards, it seems like less of a theoretical challenge
09:13 < muurkha> docl: maybe with enough redundancy in the photonic representation it would work
09:13 < muurkha> but does the infrared from the outside of a cremation furnace have that redundancy?  maybe it has the opposite
09:14 < docl> basically, when a photon originating from Earth hits a dust speck 100 ly away there will be a change to its velocity, etc. any incoming photons from that speck will give you some basis to infer that. so it's a kind of quasi-reflection (i.e. not the same photon, but information about it similarly limited to the speed of light)
09:15 < muurkha> like, LUKS tries to ensure secure erasure of disk encryption keys by storing them in multiple blocks of random data, the last of which is the XOR of the actual (password-encrypted) keys with all the other blocks of data
09:15 < muurkha> so that if any one of the 50 or so copies gets corrupted, the keys are lost
09:15 < docl> I'm not betting on the infrared of a cremation furnace, more the light passing through you before you die and during times leading up to your death. higher temp stuff is going to be a lot harder to unravel
09:16 < muurkha> hmm, that seems more plausible
09:16 < muurkha> by looking at reflections from dust specks 100 light-years out we could form a sort of hologram of the earth 200 years ago
09:16 < muurkha> see where people lived, what they looked like walking around
09:17 < muurkha> probably still impossible but less obviously so
09:17 < docl> yeah that's my thinking
09:17 < kanzure> there are various proposals for using (substantially more local, eg orbital) dust clouds for high resolution astronomical observations
09:17 < kanzure> https://diyhpl.us/~bryan/papers2/space/Orbiting%20rainbows%20-%20optical%20manipulation%20of%20aerosols%20and%20the%20beginnings%20of%20future%20space%20construction%20-%20NASA.pdf
09:17 < docl> lets me not give up hope for the people who are dead-dead in the null hypothesis
09:18 < docl> there would also be various distict phenomena to look at, like neutrino paths deviating slightly as they move past a given atom (which is sub-lightspeed)
09:18 < kanzure> https://diyhpl.us/~bryan/papers2/space/Self-deployed%20extremely%20large%20low%20mass%20space%20structures%20-%202007.pdf
09:19 < docl> the fixed speed of light seems like it sorts the data in a way that might be implausible with neutrino sources though (since afact different neutrinos from a given source might move different speeds)
09:21 < kanzure> there are also proposals for using large objects as remote lenses (such as planets, stars, and galaxies)
09:22 < docl> part of why I'm so interested in making a dyson swarm soon
09:25 < docl> might be possible to image 100y but no further back or something like that. crazy thing is if you do unlock one point in time at high enough res to download a mind from, it probably includes every living person (and every secret, including bitcoin keys etc) from then, as long as it's not blocked by something high temperature
09:28 < docl> just being behind a lot of material (e.g. underground) might also be hard to unravel. but then again it may still only be a matter of collating enough data from enough dust specks
09:30 < muurkha> emulating somebody else's mind is something people do all the time, it's just a matter of the degree of fidelity
09:31 < muurkha> presumably the hardware for it evolved in order to facilitate social interactions: apes who could predict other apes' behavior under different circumstances could have desirable rather than undesirable interactions with them
09:32 < docl> yeah, but doing so based on observation of their neural structures is imo a bit different from doing it based on behaviors
09:32 < muurkha> right, but we're talking about doing it based on behaviors that can be seen from 100 light-years away, aren't we?
09:33 < muurkha> and presumably something superhuman could do a better job of it, figuring out passphrases from strange turns of phrase, anomalous hesitations, strangely avoided turns of phrase, etc.
09:33 < muurkha> but we don't normally say that Robert Graves downloaded Claudius's mind
09:33 < docl> well we aren't limited to visible spectra so neurons/fine structures might be imageable
09:35 < muurkha> what, like submillimeter wave emissions?
09:36 < docl> yeah pretty much every kind. even gravity is going to show up if you pull enough data together
09:37 < muurkha> my thought is that generally light that can penetrate 5 mm of skin and skull isn't going to show details of other similar tissues that are much finer than 5 mm, and similarly for 10 mm, 15 mm, etc.
09:38 < docl> plenty of gamma rays and x-rays passing through in aggregate
09:39 < muurkha> well, you can definitely get a neuron-scale image of somebody's brain if you shoot enough X-rays through it from different angles
09:41 < muurkha> usually the total radiation dose involved is high enough to be fatal, so there may *not* be plenty in aggregate; there may be far too little
09:41 < docl> I guess it's a question of how many are needed given the constraints of the scenario
09:42 < muurkha> but I don't actually know what the quantum efficiency of our current X-ray detectors are
10:13 -!- spaceangel [~spaceange@ip-78-102-216-202.net.upcbroadband.cz] has joined #hplusroadmap
10:30 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Ping timeout: 276 seconds]
10:31 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
11:48 < docl> was wondering if kamato hongo's weird 96-hour sleep cycle (48 hours awake / 48 asleep) was related to her longevity but apparently it started after a hip surgery at 110 and she died at 116 so probably not
11:49 < docl> if there's a connection, the change to her sleep cycle was likely a product of being old rather than the long sleep being a longevity contributor
11:51 -!- catalase [catalase@stable.bnc.by.bnc4you.xyz] has quit [Quit: Stable ZNC by #bnc4you]
11:51 -!- lkcl [lkcl@we.will.rock.you.bnc4you.xyz] has quit [Quit: BNC by #bnc4you]
12:00 -!- catalase [catalase@stable.bnc.by.bnc4you.xyz] has joined #hplusroadmap
12:04 -!- lkcl [lkcl@we.will.rock.you.bnc4you.xyz] has joined #hplusroadmap
12:06 < maaku> docl: sorry your idea is way below the noise threshold
12:06 < maaku> those grains of dust are being hit with photons from literally everything else in the universe
12:07 < maaku> you physically can't tease out just the contribution from one vector
12:12 < docl> most of the other things in the universe are a lot further away, which means the vector the photons can follow is very narrow / point-like compared to the ones from earth. my intuition (could be flawed) is that this makes it possible to characterize and account for them
12:17 -!- oxphi [~oxphi@107.181.189.44] has joined #hplusroadmap
12:25 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
12:26 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
14:35 < kanzure> .tw https://twitter.com/abecedarius/status/1531360434082693122
14:36 < saxo> @kanzure @perrymetzger There was an amusing bit in the Cyberiad on the legendary palefaces creating the robots, and how absurd it'd be to believe it. (@abecedarius, in reply to tw:1531259571251052544)
14:51 < kanzure> always weird running into carl feynman https://www.aleph.se/Trans/Words/h.html
15:06 -!- spaceangel [~spaceange@ip-78-102-216-202.net.upcbroadband.cz] has quit [Remote host closed the connection]
15:34 < kanzure> "No, my well-wishing friend, there was a planet once there, a fairly civilized planet as planets go, flowing with the usual quantity of blood, sweat and tears. Well, one morning we dropped eight hundred million transistorized Universal Wish Granters on that planet, [and they blew up their planet]. [..] Our ancestors, true, turned our sun into a cube and made a box of our planet, arranging its ...
15:34 < kanzure> ...mountains in a monogram, but that was only to test their Gnostotron. They could have just as easily assembled the stars in a checkerboard, exitinguished half the heavens and lit up the other half, constructed beings peopled with lesser beings, giants whose thoughts would be the intricate dance of a million pygmies, and they could have redesigned the galaxies, revised the laws of time and ...
15:34 < kanzure> ...space--but tell me, what sense would there have been to any of this? Would the universe be a better place if stars were triangular, or coments went around on wheels?"
15:48 < kanzure> sounds like an excuse to be lazy to me
15:57 < kanzure> either hanuman's character had underdeveloped motives or requiem for homo sapiens was really just an elaborate spoof on the principle question presented by blade runner
16:04 -!- srk- [~sorki@user/srk] has joined #hplusroadmap
16:06 -!- srk [~sorki@user/srk] has quit [Ping timeout: 255 seconds]
16:06 -!- srk- is now known as srk
16:09 -!- srk- [~sorki@user/srk] has joined #hplusroadmap
16:13 -!- srk [~sorki@user/srk] has quit [Ping timeout: 258 seconds]
16:13 -!- srk- is now known as srk
16:37 < abecedarius> the chapter on Trurl's Electronic Bard is fun reading too in these days of gpt3 etc.
19:15 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::93] has quit [Quit: Leaving]
20:01 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Ping timeout: 240 seconds]
22:10 < jrayhawk> https://astralcodexten.substack.com/p/a-guide-to-asking-robots-to-design
23:38 -!- mirage3352367959 [~mirage335@64.79.52.86] has joined #hplusroadmap
23:40 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
23:42 -!- mirage335236795 [~mirage335@64.79.52.86] has quit [Ping timeout: 252 seconds]
23:44 -!- mirage3352367959 [~mirage335@64.79.52.86] has quit [Ping timeout: 252 seconds]
--- Log closed Tue May 31 00:00:38 2022