2022-03-30 00:17:31 <KipIngram   > Just whipped this out:
2022-03-30 00:17:33 <KipIngram   > : body  base.b @ + ;
2022-03-30 00:17:36 <KipIngram   > : header  base.h @ + ;
2022-03-30 00:17:38 <KipIngram   > : name  lfa nfa count x:7F and type space ;
2022-03-30 00:17:40 <KipIngram   > : (see)  4 + .h@ (;) .=; header name me ;
2022-03-30 00:17:42 <KipIngram   > : see  { find 4 + h@ body 4 - (see) 0 } ;
2022-03-30 00:17:44 <KipIngram   > Sample operation:
2022-03-30 00:17:53 <KipIngram   > see (num) frame { 0 pre digits post (lit) (;) }  ok
2022-03-30 00:18:10 <KipIngram   > (From frame on is the output)
2022-03-30 00:18:38 <KipIngram   > That doesn't look quite right, but it's getting pretty close.
2022-03-30 00:19:14 <KipIngram   > Oh, actually it is - the assembly definition is this:
2022-03-30 00:19:58 <KipIngram   >                 run     _frame, frame, zero, pre, digits
2022-03-30 00:20:01 <KipIngram   >                 run     post, plit, 4+do, unframe, dosem
2022-03-30 00:20:43 <KipIngram   > It terminates on finding (;), which isn't going to work in all cases.  Doesn't print that one, though.
2022-03-30 00:29:38 <KipIngram   > Hmmm.  I see some odd looking things I'll need to check into.
2022-03-30 00:29:51 <dave0       > hey KipIngram
2022-03-30 00:29:57 <KipIngram   > hey dave0
2022-03-30 00:30:02 <KipIngram   > Good day?
2022-03-30 00:30:20 <dave0       > yeah it was alright.. there was less people and traffic today, it was good
2022-03-30 00:30:24 <dave0       > how are you?
2022-03-30 00:30:36 <dave0       > i don't like crowds
2022-03-30 00:30:37 <KipIngram   > I'm good.  Pretty laid back work day.
2022-03-30 00:30:42 <dave0       > cool cool
2022-03-30 00:30:50 <KipIngram   > Me either, at least not too much.
2022-03-30 00:55:34 <KipIngram   > Ah - that output actually does make sense.  It just looked a little off because my SEE is too simple - I'm not checking for inline arguments.
2022-03-30 00:55:44 <KipIngram   > So it tried to print those as word names.
2022-03-30 00:56:41 <KipIngram   > The words that have inline arguments all have a bit in their header set, so it will be easy to handle that.
2022-03-30 00:57:22 <KipIngram   > that ME <;; above really ought to be ME -36
2022-03-30 03:04:11 <veltas      > dave0: What do you do where there's lots of people?
2022-03-30 03:07:45 <dave0       > veltas: i try to avoid it.. i walk on quiet roads, take the less common routes
2022-03-30 03:08:58 <dave0       > the hardest part to avoid is if i have to get something from the grocery store.. then it's all business.. straight to the ATM, straight to the shop, leave out the back door and stick to the side streets
2022-03-30 03:09:41 <dave0       > 3pm is an annoying time, that's when all the school kids finish for the day
2022-03-30 03:10:18 <dave0       > and more traffic when the soccer moms pick up their kids, and the school buses on every street
2022-03-30 03:14:09 <veltas      > Oh it's just on your commute then?
2022-03-30 03:17:26 <dave0       > yup
2022-03-30 03:17:37 <dave0       > at home i'm by myself, which i like
2022-03-30 03:18:34 <veltas      > I'm the total opposite in that regard lol
2022-03-30 03:19:24 <veltas      > Too used to open offices
2022-03-30 03:26:02 <dave0       > veltas: don't you like a little bit of "me" time?
2022-03-30 09:43:24 <KipIngram   > I definitely know what you mean about "me time."
2022-03-30 09:43:50 <KipIngram   > I prefer my family's company over a crowd of strangers.
2022-03-30 10:13:17 <KipIngram   > So, I enjoyed that GA144 video I linked yesterday.  It's a bit dry - he just walks through what all the cores are doing, and you can't really "track it all" in real time, but I think it would be useful if you sat down to try to program that thing.
2022-03-30 10:13:47 <KipIngram   > There's more on the dev system here:
2022-03-30 10:13:50 <KipIngram   > http://www.etherforth.org/
2022-03-30 10:15:10 <KipIngram   > Looks like the whole thing can be downloaded from there.
2022-03-30 10:15:34 <KipIngram   > I'd love to play with one of those, but they want $500 for the eval kit, and that's kind of "ugh."
2022-03-30 10:15:54 <KipIngram   > It's hard for me to believe there's $500 of parts on that board, given the GA144 itself is just $20.
2022-03-30 10:16:33 <KipIngram   > I've never been terribly pleased with chip vendors who also try to make their development stuff a profit center - you'd think the right strategy would be to price those things as low as possible to try to encourage more demand for the chip.
2022-03-30 10:17:00 <KipIngram   > But I guess when you're struggling to keep an enterprise afloat you do whatever you can.
2022-03-30 10:17:39 <KipIngram   > And, to be fair, I imagine there are a lot of people like me, who just want to play with the thing.  Selling me an eval kit at rock bottom price won't pump up their chip volume at all.  So I kind of see both sides.
2022-03-30 10:18:23 <KipIngram   > Anyway, it looks like programming in that environment is very "different" from what I imagine most of us are familiar with.
2022-03-30 10:18:56 <KipIngram   > It's more like you're designing a circuit, and you're just shaping what the components in that circuit do with little bits of code.
2022-03-30 10:28:32 <MrMobius    > I think some companies lose money on eval boards but more than make up for it when someone who got an eval chip uses it for a product
2022-03-30 10:28:49 <MrMobius    > like TI selling its MSP430 launchpad for $4.30 for a while
2022-03-30 10:30:08 <MrMobius    > also im laughing thinking of them asking $500 for that thing. the guy who runs it gives talks on forth day at svfig and keeps mentioning they have 0 clients. dunno how $500 eval boards will help considering a $20 stm32 board is about as good as my first pc
2022-03-30 10:31:02 <MrMobius    > I would love to know though what the efficiency of that chip is. I know they talked a lot about it being clockless and its low energy consumption. it would be neat to put it in terms of competing microcontrollers
2022-03-30 10:31:21 <MrMobius    > unfortunately my pile of unused "treasures" is already too big to add that to it
2022-03-30 10:35:16 <KipIngram   > Right - that's how I'd approach it.  I'd want to get the eval kits into as many hands as possible.
2022-03-30 10:36:28 <KipIngram   > My impression is that Chuck was really focused on energy consumption.  He talked about it in a video I saw - I think dave0 linked it yesterday.
2022-03-30 10:36:36 <KipIngram   > 7 micro-joules per instruction.
2022-03-30 10:36:49 <KipIngram   > And the static power was down in the nanowatts, I think.
2022-03-30 10:36:55 <KipIngram   > Per core.
2022-03-30 10:37:25 <KipIngram   > Yeah - I'd love to own one or two of those boards, but oh well.
2022-03-30 10:40:42 <MrMobius    > scmartboard was selling the chip and adapter board for $40 at one point
2022-03-30 10:45:45 <lagash      > speaking of low power consumption, anyone heard of ONiO Zero? I've been following RISC-V lately and their chips are interesting.. https://www.onio.com/
2022-03-30 10:47:36 <boru        > That website is obnoxious.
2022-03-30 10:48:12 <KipIngram   > Yeah - they did seem to sort of go whole hog on the "bells and whistles" flash.
2022-03-30 10:48:15 <boru        > I've got some gd32vf boards knocking around here. Fun to poke around with, but the RISC-V foundation don't really care about their embedded subset.
2022-03-30 10:48:47 <KipIngram   > Oh man - it IS obnoxious.
2022-03-30 10:48:48 <boru        > Have they even officially standardised the embedded set ISA yet? Last time I looked/asked about it, they didn't care and said they were focused on their big iron stuff.
2022-03-30 10:50:16 <boru        > JFC, how do you scroll on this website?
2022-03-30 10:50:41 <boru        > They've somehow hijacked your mouse and make the website scroll at 0.25 speed or some such.
2022-03-30 10:51:04 <KipIngram   > And all those huge useless pictures.
2022-03-30 10:51:48 <boru        > For a company concerned with the environment, their website sure doesn't do anything towards reducing power usage.
2022-03-30 10:52:59 <KipIngram   > So that SEE word was easy to write, but it's rather limited.  In addition to the obvious things needed (checking for more than just ; to terminate on, detecting offsets and not treating them like words, etc.) I think a good see would scan the whole dictionary structure and use the pointers in the word headers as boundaries too, or at least notify when a word executed through into the next word's territory.
2022-03-30 10:53:41 <KipIngram   > E.g., if I SEE WARM, I get a nice batch of output, but it doesn't give me any indication that somewhere in there I've moved into QUIT.
2022-03-30 10:54:23 <KipIngram   > The only way it could know that would be by seeing that QUIT pointed into the middle, and that could be any word, so it would need to be aware of the whole dictionary.
2022-03-30 10:55:47 <KipIngram   > Ack.  That website spun up my fans.  Buh bye.
2022-03-30 10:56:12 <KipIngram   > That's just misguided website design.
2022-03-30 10:56:23 <lagash      > Yeah sorry, you can just search for the chip on your favorite electronics site or whatnot
2022-03-30 10:56:39 <lagash      > I hate the design too. Their first iteration wasn't too bad.
2022-03-30 10:56:42 <KipIngram   > :-)  No worries - I am curious so I will.
2022-03-30 10:56:53 <KipIngram   > Someone wanted to show off all their magic skills.
2022-03-30 10:57:20 <KipIngram   > I guess I get it - everyone likes to show what they can do.
2022-03-30 10:57:35 <lagash      > I feel they're trying to impress someone, but it ain't us :D
2022-03-30 10:57:43 <KipIngram   > Right.  :-)
2022-03-30 10:58:13 <boru        > I think it's more of the usual trend of use-web-technologies-for-everything-including-things-its-not-suitable-for that plagues the www these days.
2022-03-30 11:03:38 <KipIngram   > You know, just making the GA144 so it powers up ready to be loaded with code looks like it took some thought.
2022-03-30 11:04:36 <KipIngram   > From what I can tell that's done by having the cores power up ready to "execute a port."  So I guess they're waiting for data to show up on that port, which they will treat as an instruction.
2022-03-30 11:05:24 <KipIngram   > I guess you bring up one on the periphery, and then guide it through bringing up its neighbors, and so on.
2022-03-30 12:00:47 <KipIngram   > The explnaations on that etherforth site are pretty good.
2022-03-30 12:00:53 <KipIngram   > All the details.
2022-03-30 13:23:36 <KipIngram   > This is damn odd.  My f2xm1 floating point instruction seems to be having no effect.
2022-03-30 13:24:38 <KipIngram   > The confersion routine is calculating log2(10^exponent) just fine.  Correct result.  The next step is f2xm1 to get 10^exp - 1, and then add 1 to get 10^exp.  But that f2xm1 behaves as a "no op."
2022-03-30 13:26:29 <KipIngram   > Oh, I see the problem.
2022-03-30 13:26:47 <KipIngram   > That f2xm1 instruction requires an argument between -1 and 1.
2022-03-30 13:26:52 <KipIngram   > Ok.
2022-03-30 14:13:06 <KipIngram   > So I need to separate my log2(10^exp) result into integer and fractional parts, run f2xm1 on the fraction, and then multiply by 2^(integer part).
2022-03-30 16:57:00 <KipIngram   > Well, that was annoying to have to take care of.  You've got to watch that integer value because you always shift 1 left |integer| times, but if integer is negative then you fdiv with it instead of fmul with it.
2022-03-30 16:57:06 <KipIngram   > But, it seems to be working fully now.
2022-03-30 16:58:45 <KipIngram   > This is good for checking stuff like that, btw:
2022-03-30 16:58:47 <KipIngram   > https://baseconvert.com/ieee-754-floating-point