2025-03-13 11:51:16 <GeDaMo      > https://www.tomshardware.com/tech-industry/the-worlds-smallest-microcontroller-measures-just-1-38-mm2-and-costs-20-cents
2025-03-13 12:15:04 <veltas      > noice
2025-03-13 12:15:44 <veltas      > Could run Forth on that, for sure
2025-03-13 16:54:48 <user51      > veltas: Yeah, that makes sense, but branching words would need to keep their depth, probably on the cf/rstack.
2025-03-13 16:57:33 <user51      > veltas: re: words like IF during interpretation.
2025-03-13 16:58:06 <veltas      > Been trying to work with a simpler editor https://pastebin.com/raw/9ekEWkJe
2025-03-13 17:00:49 <veltas      > user51: [IF] only uses a stack while it's skipping content, so the rest of the time there
2025-03-13 17:00:52 <veltas      > 's no stack
2025-03-13 17:01:29 <veltas      > [THEN] is a no-op and [ELSE] just skips to the next [THEN] with [IF][THEN] nesting, so also a stack while running
2025-03-13 17:01:34 <veltas      > But not otherwise
2025-03-13 17:02:19 <veltas      > I'm not really keen on [IF] [THEN] but it's in the ANS standard so I can refer to it
2025-03-13 17:03:17 <veltas      > The old-school way to conditionally skip stuff is to conditionally LOAD blocks, and segregate different versions in different blocks
2025-03-13 17:17:29 <user51      > Darn crash.
2025-03-13 17:17:49 <user51      > I'm not really doing anything old school, just exploring what I'm interested in.
2025-03-13 18:21:47 <veltas      > user51: https://forth-standard.org/standard/tools/BracketELSE has a reference implementation
2025-03-13 18:28:03 <veltas      > I'm just saying because I am interested in classic Forth techniques, especially as they tend to result in smaller implementations
2025-03-13 18:28:06 <veltas      > Well sometimes anyway
2025-03-13 18:28:39 <veltas      > Also because it was the 'classic' Forths that were used when Forth had a proper niche and serious market share, but since e.g. the ANS standard it's had less and less
2025-03-13 19:22:51 <veltas      > Yeah this block editor in less than a block is definitely ripped off of crc's basic editor
2025-03-13 20:19:36 <veltas      > It is already surprisingly functional
2025-03-13 20:42:39 <veltas      > Of course Forth isn't for bootstrapping
2025-03-13 20:42:43 <veltas      > Forth is the destination
2025-03-13 21:52:16 <xentrac     > not sure if *serious* market share
2025-03-13 22:00:22 <xentrac     > GeDaMo: there are definitely microcontrollers that are cheaper than that, like https://www.lcsc.com/product-detail/Microcontroller-Units-MCUs-MPUs-SOCs_PADAUK-Tech-PMC251-S14_C317604.html which costs half as much and has two hardware threads
2025-03-13 22:00:34 <xentrac     > a bit harder to run Forth on
2025-03-13 22:00:45 <GeDaMo      > I thin kit was more about the size, the price was just mentioned
2025-03-13 22:01:37 <GeDaMo      > https://hackaday.com/2019/04/26/making-a-three-cent-microcontroller-useful/
2025-03-13 22:01:58 <xentrac     > a single-threaded version is 2.71¢: https://www.lcsc.com/product-detail/Microcontrollers-MCU-MPU-SOC_PADAUK-Tech-PMS150C-U06_C168658.html
2025-03-13 22:03:36 <xentrac     > I've seen a couple of microcontrollers on LCSC that are under 2¢, like the NY8A051 (which is not an 8051 despite the name) and the Cypress CY8C4045FNI, but they're out of stock now, and who knows if that's forever
2025-03-13 22:04:14 <xentrac     > it's true that I don't know of any microcontrollers that are smaller than a WLCSP-8
2025-03-13 22:06:56 <xentrac     > you could target-compile a Forth for any of these, but none of them have enough RAM to hold much of a program.  TI's has 1KiB of RAM
2025-03-13 22:07:33 <xentrac     > so you aren't going to do much interactive interpretation
2025-03-13 22:08:23 <xentrac     > the PY32 has enough RAM, though, and like the TI chip, it's a 32-bit ARM: https://www.lcsc.com/product-detail/Microcontroller-Units-MCUs-MPUs-SOCs_PUYA-PY32F002AL15S6TU_C5292060.html
2025-03-13 22:09:38 <xentrac     > 8.92¢ and I think 3KiB of SRAM, though the datasheet actually says "最大 3Kbytes SRAM", where the Chinese means "maximum", so it's probably a datasheet for multiple models
2025-03-13 22:14:44 <xentrac     > yeah, I'm not finding anything as small as WLCSP-8 on LCSC
2025-03-13 22:25:30 <GreaseMonkey> IIRC the Padauk ones are largely a shuffled-around-for-legal-reasons PIC
2025-03-13 22:26:01 <GreaseMonkey> 3 KB RAM sounds kinda tight for ARM
2025-03-13 22:26:09 <xentrac     > they definitely have a lot of PIC-like characteristics
2025-03-13 22:26:19 <xentrac     > you'd put the ARM code in the Flash, not the RAM
2025-03-13 22:26:36 <xentrac     > which is 20KiB
2025-03-13 22:26:59 <xentrac     > I don't think the reason Padauk shuffled around PIC stuff was legal
2025-03-13 22:27:01 <GreaseMonkey> ...then again the RISC-V CH32V003 which competes in that space is 2 KB RAM, 16 KB flash
2025-03-13 22:27:19 <xentrac     > their datasheets have diagrams that seem to have been copied from Microchip datasheets
2025-03-13 22:27:53 <xentrac     > so they aren't too concerned about copyright law, and copyright law doesn't cover instruction sets or similar functional things
2025-03-13 22:28:49 <xentrac     > to stop competitors from copying that kind of stuff you need patents, and I'm pretty sure all of Microchip's patents that could cover the original PICs expired before Padauk launched their line
2025-03-13 22:59:29 <veltas      > xentrac: Forth was more serious in the late 70's and 80's
2025-03-13 22:59:48 <veltas      > It was never a major popular language though
2025-03-13 23:00:12 <xentrac     > yeah, I feel like Forth is still a serious language
2025-03-13 23:00:15 <veltas      > But I mean it was the first high level language in space, as far as we know.  Standard in astronomy
2025-03-13 23:00:27 <xentrac     > standard?
2025-03-13 23:00:30 <veltas      > Used in medical equipment, minicomputers
2025-03-13 23:00:40 <veltas      > Big and tiny applications
2025-03-13 23:00:48 <xentrac     > yeah, used in a lot of different places!
2025-03-13 23:00:49 <veltas      > It's barely used now compared to how it was back then
2025-03-13 23:01:18 <xentrac     > yeah, it's definitely much less used now
2025-03-13 23:02:34 <veltas      > There was at least one BYTE special on it, where Chuck Moore wrote a pretty terrible article about lots of stuff that most people wouldn't care to hear about
2025-03-13 23:02:55 <veltas      > But really that matches the language itself, because inherently it's not going to be appealing to most people
2025-03-13 23:04:32 <veltas      > Forth shined when the software ecosystem was a lot less developed, and the hardware landscape was more diverse
2025-03-13 23:04:38 <veltas      > And also more memory constrained
2025-03-13 23:05:42 <xentrac     > looks like this issue of BYTE: https://archive.org/details/byte-magazine-1980-08
2025-03-13 23:05:52 <xentrac     > I really loved BYTE's covers
2025-03-13 23:05:53 <veltas      > Yeah
2025-03-13 23:06:07 <veltas      > He proudly declared in this article that Forths can't be portable, and that they have to be rewritten for each arch
2025-03-13 23:06:24 <veltas      > I mean even in 1980 it was obvious that that was not a good software strategy
2025-03-13 23:06:30 <veltas      > And I think he was decidely proven wrong
2025-03-13 23:06:32 <xentrac     > 
ACTION downloads a 200-megabyte PDF like a fool

2025-03-13 23:07:15 <veltas      > Also talked about being a difficult person etc, I mean if you've got a figurehead like this then things won't go well
2025-03-13 23:07:20 <xentrac     > I don't know, I feel like there are a lot of software companies that ran aground on trying to make software portable around that time
2025-03-13 23:07:26 <veltas      > The writing was on the wall in 1980
2025-03-13 23:07:55 <xentrac     > like, everybody who tried to ship products on the UCSD P-system came to regret it
2025-03-13 23:08:12 <xentrac     > I ran the P-system myself
2025-03-13 23:08:30 <xentrac     > I remember it as being painfully slow
2025-03-13 23:08:46 <veltas      > I just mean in the sense that standard programming languages, or mostly-standard languages, dominated the 80's, 90's, ... need I go on?
2025-03-13 23:08:56 <xentrac     > yeah, they did
2025-03-13 23:09:26 <veltas      > Likewise ultimately operating systems abstracted the hardware for programs
2025-03-13 23:09:33 <xentrac     > but in 01980 the only way to ship software of salable quality for microcomputers was by writing it in assemby
2025-03-13 23:09:36 <xentrac     > *assembly
2025-03-13 23:09:51 <xentrac     > I was looking at MacPaint the other day, which was from 01984
2025-03-13 23:09:55 <veltas      > Exactly, in 1980 the limitations of the time were good for Forth
2025-03-13 23:10:01 <xentrac     > it's almost half written in assembly, not counting QuickDraw
2025-03-13 23:10:52 <veltas      > But anyone who thought that this was a matter of *principle* was left by the wayside, including Forth Inc.
2025-03-13 23:10:52 <xentrac     > which was written entirely in assembly by the same author and which in some sense contained most of MacDraw's code
2025-03-13 23:11:02 <xentrac     > and if she wasn't running Forth, Inc., in 01980, she certainly was by 01984
2025-03-13 23:11:26 <xentrac     > well, let's say 01985.  But I think 01984 or earlier
2025-03-13 23:11:34 <veltas      > It's a weird situation that historically one of the worst things for Forth was the person who made it
2025-03-13 23:12:02 <xentrac     > he's a weird person, but maybe it took a weird person to invent Forth
2025-03-13 23:12:04 <xentrac     > or discover it
2025-03-13 23:12:06 <veltas      > I think Chuck really was just not a businessman, he never really got that
2025-03-13 23:12:38 <xentrac     > at the Silicon Valley Forth year-end fireside last year, he talked a fair bit about Forth and GreenArrays
2025-03-13 23:12:45 <veltas      > Chuck was a really solid engineer at some point, but he got to a stage where he also had to be a leader and a businessman, and that he couldn't do
2025-03-13 23:13:04 <xentrac     > but also he talked about the virtues of using a metal walking stick when you go hiking in the mountains
2025-03-13 23:13:13 <xentrac     > because it electrically earths you
2025-03-13 23:13:31 <veltas      > That probably has more to do with him being 86 than anything else
2025-03-13 23:13:50 <xentrac     > see https://old.reddit.com/r/Earthing/
2025-03-13 23:13:57 <xentrac     > it's a whole meme
2025-03-13 23:14:05 <veltas      > Yeah I can't take that seriously
2025-03-13 23:14:24 <veltas      > ColorForth has some interesting stuff in it, and cool ideas, and I know crc found it inspirational
2025-03-13 23:14:24 <xentrac     > yeah, it doesn't seem plausible to me
2025-03-13 23:14:41 <xentrac     > I was having lunch with Andy Glew a few years back
2025-03-13 23:14:44 <veltas      > But the greenarrays stuff is pretty crazy
2025-03-13 23:15:01 <xentrac     > the guy who designed MMX, you might know of him
2025-03-13 23:15:12 <xentrac     > I asked him what he thought of Chuck Moore
2025-03-13 23:15:32 <veltas      > I don't know him, but still interesting
2025-03-13 23:15:34 <xentrac     > at first he thought I was talking about his ex-boss at AMD
2025-03-13 23:15:40 <veltas      > lol
2025-03-13 23:15:47 <xentrac     > but when I clarified, he exclaimed:
2025-03-13 23:15:59 <xentrac     > "Oh, HIM!  In my DREAMS I could do what he does."
2025-03-13 23:16:50 <veltas      > Chuck was a fantastic engineer at some point....
2025-03-13 23:16:59 <xentrac     > the GreenArrays and IntellaSys chips do actually work in practice, unlike the F21/i21
2025-03-13 23:17:38 <xentrac     > but to me this doesn't look like "Chuck was a fantastic engineer at some point, but stopped being any good later"
2025-03-13 23:17:50 <veltas      > ga144 is totally useless
2025-03-13 23:18:15 <xentrac     > to me it looks like "Chuck spent his entire career doing things that were at the limit of being feasible, so they often failed"
2025-03-13 23:18:19 <veltas      > I mean holistically, obviously you *can* use it, but I don't see any place where it's the best value
2025-03-13 23:19:04 <xentrac     > yeah, I haven't seen anybody using it successfully, but it does seem to do what the datasheet says and have a working, usable instruction set
2025-03-13 23:19:09 <veltas      > My analysis here is more specific and less vague, I think at some point the equation went against him and he doubled down on reinventing the wheel
2025-03-13 23:19:41 <veltas      > Like e.g. GA144 is not a good business idea, I wouldn't have invested, even though I find the creator and concept both fascinating and want it to work
2025-03-13 23:19:42 <xentrac     > reinventing the wheel was already, as you pointed out, often a poor cost/benefit tradeoff in 01980
2025-03-13 23:19:52 <veltas      > Even more so now
2025-03-13 23:20:18 <xentrac     > I think the GA144 was a much better idea when Intellasys was first working on it, 20 years ago, before GreenArrays was founded
2025-03-13 23:20:40 <veltas      > Okay but still there's a whole 20 years to figure it out
2025-03-13 23:21:01 <xentrac     > most of those were spent in lawsuits between Chuck and Leckrone
2025-03-13 23:21:28 <xentrac     > Chuck's wife died before his final victory
2025-03-13 23:21:45 <veltas      > Yeah I read what Chuck wrote about her, very sad
2025-03-13 23:22:43 <xentrac     > yeah
2025-03-13 23:23:44 <veltas      > I'd be less critical if this was a personal endeavour, but when you try and actually get people involved in a business and drive them over a cliff it's a bit different
2025-03-13 23:25:03 <veltas      > Cheaper than ever to do a Forth chip ASIC if someone wants to do that for their amusement
2025-03-13 23:25:19 <xentrac     > well, enormously more expensive than it would have been six months ago
2025-03-13 23:25:39 <veltas      > But ... Forth is easy to port, why make a chip *for* it?
2025-03-13 23:25:40 <xentrac     > and probably more expensive than when iTV was trying to do it in the 90s
2025-03-13 23:26:28 <xentrac     > when you could do a MOSIS run for US$2000
2025-03-13 23:26:44 <xentrac     > as for why, well, it's aesthetically appealing to not have an extra unnecessary instruction set underneath Forth
2025-03-13 23:27:52 <veltas      > Honestly, writing a Forth in a 'Forth-friendly instruction set' is not really that much nicer
2025-03-13 23:29:14 <veltas      > I just think there's a point in saying Forth is good at porting to different archs, so why bother making an arch for it
2025-03-13 23:29:22 <veltas      > Given making an arch is already really hard to do proper
2025-03-13 23:30:54 <xentrac     > the IHP Tiny Tapeout alpha-testing run had a bunch of different CPUs but no Forth CPUs: https://tinytapeout.com/runs/ttihp0p2/
2025-03-13 23:31:28 <xentrac     > I think there's a creditable argument in favor of RISC rather than MISC
2025-03-13 23:31:31 <veltas      > Yeah that's something I'm thinking of
2025-03-13 23:31:41 <veltas      > I mean, in discussion, I don't plan to do this
2025-03-13 23:32:04 <veltas      > I think RISC, MISC, whatever.  Forth will drive any of it.
2025-03-13 23:32:23 <xentrac     > have you seen Chuck Thacker's "A Tiny Computer"?  https://www.cl.cam.ac.uk/teaching/1112/ECAD+Arch/files/Thacker-A_Tiny_Computer-3.pdf
2025-03-13 23:32:29 <xentrac     > It's about a page and a half of Verilog.
2025-03-13 23:32:57 <veltas      > Yeah it's easy to make a bad simple CPU
2025-03-13 23:33:06 <xentrac     > I don't think it's bad, though.
2025-03-13 23:33:53 <veltas      > I don't really know what the point is, business wise
2025-03-13 23:34:06 <veltas      > Obviously it's someone's hobby, or educational, fair enough
2025-03-13 23:35:01 <xentrac     > Well, take a look at the link I posted above.
2025-03-13 23:35:08 <veltas      > Yeah I did
2025-03-13 23:35:20 <veltas      > What about it?
2025-03-13 23:35:29 <xentrac     > It's not someone's hobby; it's educational.
2025-03-13 23:36:58 <veltas      > I said "hobby, or educational"
2025-03-13 23:37:28 <veltas      > The author makes some bold claims in there and they're yet to be proven, good luck to his idea, meanwhile I'm not in CPU research or the CPU business
2025-03-13 23:37:33 <veltas      > Better men than me have tried
2025-03-13 23:38:06 <xentrac     > Oh, he was, though.  He was one of the top computer architecture researchers in history: https://en.wikipedia.org/wiki/Charles_P._Thacker
2025-03-13 23:38:36 <veltas      > This was written in 2007
2025-03-13 23:38:46 <xentrac     > Yeah.  he's dead now.
2025-03-13 23:39:50 <veltas      > Well it sounds like he was proposing these ideas for CPUs in FPGAs, I don't know what if any of that was applied to FPGAs
2025-03-13 23:40:29 <xentrac     > in the link I gave above, he described the results of applying it to FPGAs; he got it to run at 66 MHz in 200 LUTs on a Virtex.
2025-03-13 23:41:08 <veltas      > Sorry I don't find this that interesting, honestly
2025-03-13 23:41:09 <xentrac     > Historically it seems like a MISC design like the MuP21 takes less transistors to reach a given level of performance and convenience than a RISC design like the ARM2, but arguably the RISC approach is conceptually simpler, and it can run at a lower clock speed
2025-03-13 23:41:56 <veltas      > Is English a second language for you?
2025-03-13 23:42:06 <xentrac     > Haha, no, why?
2025-03-13 23:42:27 <xentrac     > I'm not sure if the same relation holds on an FPGA substrate.
2025-03-13 23:42:29 <veltas      > It feels like there is a communication issue here
2025-03-13 23:42:44 <veltas      > But could just be culture/dialect etc rather than language
2025-03-13 23:43:17 <veltas      > But the way I feel talking to you is quite similar to experiences I've had exclusively with people who are ESL
2025-03-13 23:44:03 <xentrac     > Yeah, I had that feeling to a very peculiar extent with our conversation the other day
2025-03-13 23:44:13 <veltas      > But you seem to like inventing conventions and all that so maybe that extends beyond intentional quirks
2025-03-13 23:44:39 <xentrac     > I was pretty puzzled at how hard it was for you to understand what I was saying
2025-03-13 23:45:09 <xentrac     > but I concluded that you were probably feeling defensive, which interfered with communication
2025-03-13 23:45:25 <veltas      > I assure you I wasn't
2025-03-13 23:45:56 <xentrac     > But I'm sure you've encountered the idea that there is an objective reality before, for example
2025-03-13 23:46:00 <veltas      > I was engaging in a confrontational way because sometimes that's good for people, but obviously failed
2025-03-13 23:46:10 <veltas      > It's a risk
2025-03-13 23:46:29 <xentrac     > even if you don't agree with it personally
2025-03-13 23:46:43 <veltas      > I do believe in an objective reality
2025-03-13 23:49:27 <veltas      > Well I was arguing in good faith, openly, and attempting to do so constructively
2025-03-13 23:50:52 <veltas      > I think re-reading I did misunderstand you a bit, but I think you are not as good at 'reading' people as you think maybe, because you have guessed my state quite a bit and been far off the mark
2025-03-13 23:50:52 <veltas      > I do apologise though if you think I was too much
2025-03-13 23:51:49 <veltas      > Well it's a skill to be able to understand what people misunderstand.  Sometimes I pull it off, which is always good for working with customer support cases
2025-03-13 23:51:52 <veltas      > Not always