Shop OBEX P1 Docs P2 Docs Learn Events
Lament - Page 14 — Parallax Forums

Lament

1910111214

Comments

  • cgraceycgracey Posts: 14,133
    edited 2016-05-16 18:57
    dr hydra wrote: »
    Great discussion...

    I use to be a big supporter of freezing the design and getting it out in the wild....however, these last design changes, increasing HUB memory and executing directly from HUB with little performance lost, are two awesome features and game changers. Unbound by COG memory limits puts the P2 in a new league. Great feature! 200MIPS per COG, 256K HUB, eight multiprocessors (not limited to 2K), video capabilities that don't require a PHD in signal processing...and...and no interrupts. I am happy with the design...and the wait is going to be worth it

    Well, it's a little different than that:

    1) 80 MIPS per cog at 160MHz
    2) 512KB hub RAM with 32-bits-per-clock throughput to every cog
    3) 16 cogs
    4) Interrupts, if you want to use them
    5) Pipelined CORDIC available to all cogs
    6) 64 smart pins that do all kinds of things

    P.S. Woops! I think I just responded to a message from a few years back.
  • Software isn't like hardware....software *NEVER* seems to get finished! :D

    Personally having the attention span of a gnat with ADHD, I applaud Chip's determination to better our determinism. Thank you for sharing the good, bad and ugly of this process and sticking with it!
  • Heater.Heater. Posts: 21,230
    T Chap,
    My only lament is Leon's many years of persistent antagonism of Parallax and forum members by promoting other competing products.
    I don't blame Leon for drawing attention to the XMOS. They were both multi-core chips built from similar design ideals.

    In fact the XMOS has been, and as far as I know still is, the only device to compare the Propeller to.

  • jmgjmg Posts: 15,140
    Heater. wrote: »
    I don't blame Leon for drawing attention to the XMOS. They were both multi-core chips built from similar design ideals.

    In fact the XMOS has been, and as far as I know still is, the only device to compare the Propeller to.

    Yes, it is the closest comparable part. Some others take cores to the extreme, but they are not Microcontroller-area devices.

    Digikey shows 276 items under XMOS Embedded Microcontrollers, and they start from $2.85/1k, and go to $30+/1k

    If I restrict to a more P2-like TQFP128/QFN124 then it starts at $6.75/1k
  • LeonLeon Posts: 7,620
    edited 2016-05-17 07:07
    There is also the Parallella:

    https://www.parallella.org/

    http://www.adapteva.com/

    It comes on a neat little $99 PCB with a 16-core Epiphany RISC SOC and Xilinx Zynq FPGA (dual core ARM A9).

    The Epiphany chip is described here:

    http://www.adapteva.com/epiphanyiii/

    The architecture might be of interest to those people who have been discussing the P2 architecture. On-chip operation is deterministic.

    They were having financing problems, but raised $3.6M from Ericsson and a venture capital firm three years ago.

    Like the Propeller and XMOS chips it is largely the brainchild of one person, Andreas Oloffson.
  • ErNaErNa Posts: 1,738
    What makes the propeller great is not the chip itself, it is Chip. I over my livetime had never else the experience of someone realizing what must be done against the mainstream in such an extent. And we all could not only watch this process, but also we could take part in it. Either by using the chip, as with P1 or by contributing to P2 in many ways. There is, as I know and believe, no precendence.
    Parallel processing is a paradigma and the P1 was the first chip in the hand of amateurs that allowed to create parallel software for under 10$. And there is no need to study computer science to do so.
    Clearly, others follow this way, just because this is a natural way. But the transputer failed and XMOS is no sky rocket. If we look for performance / cost, the Propeller is the winner, that is sure.
    Now it is our task to make the P2 a success. We need a wall breaking application. Something that makes use of the unique features, not found in any other chip. We do not know, what this will be, so we have to search into all directions, open or closed, doesn't matter.
    Critics makes sense, can be constructive. That is the path we should go. I remember this: if you want to have a perfect president, start your campain!
  • Erna

    The Propeller is just one notion of a parallel processing chip, outside of the Xmos line no one else has adopted the idea of using a bunch of homogeneous cores for processing. Xmos has had a good bit of success with their product line in various niches so it tells me that going that route for Parallax may pay off depending on how they play it.

    Adapteva is another player in this, but they don't seem to have gotten any design wins. It also looks that for their new designs they are going to RISC-V.

    Now the big players like Infineon and Freescale went a different way for multicores. They usually have 1-2 powerful cores(like a ARM M4 or PPC), a massive array of peripherals in H/W and several "minion" controllers somewhat like the Cog to handle the grunt work. It's parallel processing of a kind.

    And they sell millions of them.

    The point is, the P-2 is entering crowded market that isn't easily wowed. It will take some work on the part of Parallax to impress potential mid to high volume customers.




  • jmgjmg Posts: 15,140
    rod1963 wrote: »
    The point is, the P-2 is entering crowded market that isn't easily wowed. It will take some work on the part of Parallax to impress potential mid to high volume customers.

    I don't think they need 'high volume' customers.
    eg Analog devices have quite a range of Analog Front Ends that target Medical Imaging.
    That is not a high volume market, but neither it is a price paranoid market.

    P2 should do very well in the instrumentation and measuring markets, and the Sea-of-USB ability is looking pretty unique.


  • cgraceycgracey Posts: 14,133
    edited 2016-05-17 09:29
    ErNa wrote: »
    ...I over my livetime had never else the experience of someone realizing what must be done against the mainstream in such an extent...

    Well, that's very flattering, ErNa.

    What is mainstream and impeding us all is the central banking model that the whole world is caught up in.

    I've been thinking about these things for years and Ken hates for me to even bring it up, but I think honest money is something everyone needs to understand the concept of. The education system and the media keep everybody in the dark about how things actually work today, how they've worked in the past, and how they are supposed to work.

    Our kids go to this little elementary school that only has 20 students in it. It's run by a man who taught locally before starting his own school when the first of his (now seven) children got to be first-grade age. He's really versed in a lot of subjects and teaches grammar from books written in the 1950's.

    Anyway, he runs a central bank model of school money he calls "squid". These are paper notes that he prints on his computer and then gives out to the kids. They buy and sell things all year from each other using "squid". Every year, he winds up printing lots of squid that he gives to some kids for doing special things. Sometimes, his own kids get easy squid for maybe finding his toenail clipper, in a pinch. The short of it is that those close to the central bank receive a disproportionate amount of money and every year the system suffers big inflation, punishing the prudent. This never gets explained. It just happens. I had the same experience with my economics teacher in the 7th grade. At the beginning of the year, students brought cupcakes in on Fridays and sold them for $15 "Dudabrandt" dollars (that was the name of their central bank). We had huge inflation, too, and at the end of the year, cupcakes were $500 Dudabrandt dollars. In desperation, the teacher once offered $200k Dudabrandt dollars to any student who would tell him who stole the fundraiser checks. At the year-end class auction, where everybody bid on stuff with the Dudabrandt dollars, those that diligently sold cupcakes got nothing, while kids involved in the central-bank shenanigans bid huge amounts for everything. No mention was ever made of the inflation effect and the political money vectoring. Oh, well. Just life, we were to suppose.

    Anyway, honest money is something that needs an expenditure of work/energy to come into existence. Just printing it isn't fair to the majority who must use it. Plus, they are disenfranchised from being able to produce any of their own. Gold and silver had to be dug out of the ground and refined, while Bitcoins take computations to 'mine'. I say the effects in America of bad money are so great that many people don't even have an inkling of what work is, anymore, as they are paid NOT to work.

    So, I was thinking if there could be some kind of honest money system introduced at our kids' school which could, at least, compete with the teacher's problematic central banking system. In the real world, central bankers always declare everything honest to be illegal. If they can't print it, it's not money. Tonight, I got the teacher to agree to let me come out next week and do a presentation on what money is, has been, and should be. I'll introduce a system that the kids could adopt to have honest money, which can't be inflated by those in charge (or weaponized, as in our sad case).

    We live in an area where a river runs by and there are lots of smooth rocks in the dirt, even a mile from the river. It occurred to me that rocks of a very certain size are rare and take effort to discover. They are also durable and handy to count. So, I got a 1.25" and a 1.125" drill bit and made some templates to hand out to the kids, as part of the presentation on money. They can go find rocks that fit through the bigger hole at all angles, but not through the smaller hole at any angle, and introduce that "money" right into the system. Everyone has a template and they can verify, for themselves, the authenticity of each stone. Nobody can magically make lots of them. Power to the Kids!

    rock1.jpg
    rock2.jpg

    Of course, easy central bank squid might seem more attractive, as no work is involved, there's money at the outset, and you can buy cookies right away, even though you'll be doomed to poverty all year, while only the pets will get rich.
    1328 x 747 - 297K
    747 x 1328 - 210K
  • Heater.Heater. Posts: 21,230
    Leon,
    There is also the Parallella:....It comes on a neat little $99 PCB with a 16-core Epiphany RISC SOC and Xilinx Zynq FPGA (dual core ARM A9).
    I know. I have got one on my desk here. I backed the Parallella kickstarter.

    I was never inclined to make comparisons between the Parallella Epiphany chip and the Propeller. Apart from having multiple cores they don't really have much in common. They built for entirely different purposes. The Propeller is a micro-controller, the Epiphany is a high speed floating point number crunching engine a co-processor.

    Sadly I have never done anything with my Parallella board. It runs damn hot and requires a heatsink and fan. That's for the on board ARM chip not he Epiphany by the way. Having had such a headache making a parallel FFT for the Prop I started
    to think making use of all those floating point cores was gong to be tough programming for me.

    One day may be.
  • Heater.Heater. Posts: 21,230
    rod1963,
    Adapteva is another player in this, but they don't seem to have gotten any design wins. It also looks that for their new designs they are going to RISC-V.
    The Epiphany chip is not usable as stand alone device. It's a maths co-processor.

    Currently it's supplied on a board with a Zynq SOC so that you can use the Epiphany from Linux. The Zynq is an ARM processor.

    The idea with going RISC-V is to replace the ARM. Use that open sourced architecture rather than the proprietary ARM. It's not to replace the Epiphany.

    One could imagine Parallella building a chip that includes both the RISC-V as the Linux running engine and an array of Epiphany processors for the number crunching work.
  • LeonLeon Posts: 7,620
    edited 2016-05-17 12:34
    I got one of the early boards, but I've never been able to get it to work. By the time I tried it the warranty period had expired (I think it was four weeks). I've just tried it again - no luck.

    Like Heater, I also backed the Kickstarter.

    I've just ordered one of the latest boards from D-K:

    https://www.digikey.co.uk/product-detail/en/adapteva-inc/P1601-DK02/1554-1001-ND/5018662

    Hopefully, I'll have more luck with this one. At any rate, I'll have a more sensible warranty.

    Here are some projects:

    https://github.com/parallella/parallella-examples/blob/master/README.md

  • Heater. wrote: »
    The Epiphany chip is not usable as stand alone device. It's a maths co-processor.

    I'm sure you've thought about pairing the Epiphany with the P2. Would that be a good fit? Or is it too dissimilar, too much overlap of functionality, etc?

  • LeonLeon Posts: 7,620
    edited 2016-05-17 12:59
    Something running Linux would also be required, and the development software would have to be ported to it. I think the Epiphany SDK already does this, running on a PC. There is an ARM version, as well.
  • Heater.Heater. Posts: 21,230
    Seairth,
    I'm sure you've thought about pairing the Epiphany with the P2. Would that be a good fit? Or is it too dissimilar, too much overlap of functionality, etc?
    That's an idea that's crazy enough to look into.

    I have no idea how the Epiphany hardware interface looks, in terms of actual signals on pins. How is program code loaded into the thing? How is data put in and results got out?

    One could imagine an Ephiphany in some P 2 application that is dedicated to being super fast FFT or whatever. One develops the FFT code for the Epiphany and provides the resulting binary blob to the Propeller application to be loaded to the Epiphany at start up.

    I wonder how hard that would be to do.

    My gut feeling is that even the P2 could not exchange data with the Epiphany fast enough to make good use of it.
  • Chip,

    Nice Walnut!!

    I like your proposal because you can get in early as the provider of the "official" measuring device (hole boards) and either provide them directly for a fee or possibly more a lucrative arrangement of licensing them. Of course, you should get one of your friends in government to act as the regulatory arm to make sure all holes start out the same and remain the same over time. Perhaps an annual inspection/licensing process here??

    You know, a pocket full of those rocks look heavy! Plus, prices in the economy are trending toward fractional rocks. How about I set up a rock box where people can put all their heavy rocks and I can give them some sort of script and/or fractional units (pebbles, marbles, etc.).

    Sounds like we can have this centralized in a couple years!!

    This all came about as I pondered if centralization was just the inevitable evolution of any economic system. It all gets above my ability to ponder rather quickly.
  • Heater. wrote: »
    Seairth,
    I'm sure you've thought about pairing the Epiphany with the P2. Would that be a good fit? Or is it too dissimilar, too much overlap of functionality, etc?
    That's an idea that's crazy enough to look into.

    I have no idea how the Epiphany hardware interface looks, in terms of actual signals on pins. How is program code loaded into the thing? How is data put in and results got out?

    One could imagine an Ephiphany in some P 2 application that is dedicated to being super fast FFT or whatever. One develops the FFT code for the Epiphany and provides the resulting binary blob to the Propeller application to be loaded to the Epiphany at start up.

    I wonder how hard that would be to do.

    My gut feeling is that even the P2 could not exchange data with the Epiphany fast enough to make good use of it.
    Check that link to example code that Leon posted. There is already an FFT written for the Epiphany processor. Just need to do the P2 side now.

  • rjo__rjo__ Posts: 2,114
    Central banks don't just print money... it actually would be preferable if they did. What our central banks do is create a dollar of debt for every dollar the money supply increases.

    There is a fixed relationship between the national debt and the total supply of real dollars. So, when our representatives vote to increase the national debt, what they are really voting for is an increase in the supply of money. In the previous decade or two, this increase seems to be used largely to float insolvent institutions, which are all based on debt. These institutions produce nothing but debt and then decide who to share that debt with. It's a problem.

    When there was a real relationship between the value of a dollar and the cost of a fixed commodity(gold)... one could easily determine how much the supply needed to increase based upon the total mass of the chosen commodity. The chosen commodity was arbitrary, but you could rationalize the process... make it predictable and fair, but not efficient or sustainable.

    We tend to dismiss communism, but one of the interesting things about communism is that it had no real need for a central banking system... they had one, but they didn't really use it the way we do now. When more money was needed at some level in the economy, the money was "created" and directed at that level of the economy. As always, corruption invariably attended the creation and distribution. Communism tore itself apart, largely because of this "unsolved" problem and the corruption it created. "Communism" failed long before the Soviet Union fell apart.

    Our system can fail for the same reason, corruption. When those who master the system are more engaged with systematic corruption than they are with keeping the system working, it will fail. We are always close.

    What we could do is to decide that new dollars should be create by a heterogeneous mechanism, where-in the usual debt based system continues to operate, but a variable portion of new dollars is determined by the tax base and distributed to the creators of the tax base. So, the more taxes you pay, the greater you benefit by future monetary expansions.

    In this way, during healthy economic times, when you pay taxes, you can expect to get part or all of that money returned when the economy grows beyond the money supply.

    We can argue about the dynamics and the implementation, but we can't argue about the problem... it is there and it will go away when we figure out how to make it go away.





  • LeonLeon Posts: 7,620
    I've just noticed that I downloaded the wrong binary file - I need the one including the display. I'm downloading it and might get my board working with that.
  • The Parallella development boards could be fun to play with along with the P1V. You could have a combined Linux/Parallela/Propeller 1 system. Linux for all the things it does well, parallella for fast maths and the P1 for deterministic soft peripherals. I'm not sure what problem this solution is looking for. A robot with complex motion needing some heavy duty, fast math?

    The offerings compared:
    parallella_table.jpg

    Not saying anyone needs anything else to distract them......
    500 x 314 - 48K
  • LeonLeon Posts: 7,620
    edited 2016-05-18 17:02
    I've got my Parallella Kickstart board working at last! One of the problems was the PS: I was using the one supplied with my smartphone which should have been OK as it is rated at 2A, but when I used another 2A supply and the correct .img file on the SD card it booted OK. I then had a problem getting SSH to work via an Ethernet connection to my router with PuTTY on my laptop (I couldn't find the board's IP address. Thanks to some help from a couple of other users on the Parallella forum I got that to work, and am now logged on to Ubuntu Linux running on the ARM cores in the Zynq FPGA.

    Now to try out some programs on the 16-core Epiphany device.
  • Heater.Heater. Posts: 21,230
    Hey, Leon, you inspire me to try and fire up the old Parallella again.

    It will be good to have something to do over the summer instead of lamenting away....
  • LeonLeon Posts: 7,620
    edited 2016-05-18 19:03
    I haven't been able to get the display version working. Anyway, the headless version is more useful for playing around with software.

    The only text editor I've been able to find is vi/vim. It's years since I've used vi - it takes some getting used to, although it is very powerful. Tried a web search and found nano, which is also installed. It's very easy to use.
  • Heater. wrote: »
    It will be good to have something to do over the summer instead of lamenting away....
    Yup! There are innumerable alternatives to lamenting away. :)

  • Ken GraceyKen Gracey Posts: 7,386
    edited 2016-05-18 21:08
    cgracey wrote: »
    I've been thinking about these things for years and Ken hates for me to even bring it up, but I think honest money is something everyone needs to understand the concept of. The education system and the media keep everybody in the dark about how things actually work today, how they've worked in the past, and how they are supposed to work.

    Hate - that's not the right description. A waste of time discussing on these forums when you could be sleeping - probably so!

    Talk about lament. I might call you later and play a lamentful clarinet tune in consideration of this thread. I've got an especially sad tune from the late 1800s that makes me feel the same way when I read some of this thread.

    Back to work, Chip! Finish up! No more features: no LUT, no mega-flexible-I/O monsters, just a simple P2! Get 'er done!

    Ken Gracey
  • cgraceycgracey Posts: 14,133
    Ken Gracey wrote: »
    cgracey wrote: »
    I've been thinking about these things for years and Ken hates for me to even bring it up, but I think honest money is something everyone needs to understand the concept of. The education system and the media keep everybody in the dark about how things actually work today, how they've worked in the past, and how they are supposed to work.

    Hate - that's not the right description. A waste of time discussing on these forums when you could be sleeping - probably so!

    Talk about lament. I might call you later and play a lamentful clarinet tune in consideration of this thread. I've got an especially sad tune from the late 1800s that makes me feel the same way when I read some of this thread.

    Back to work, Chip! Finish up! No more features: no LUT, no mega-flexible-I/O monsters, just a simple P2! Get 'er done!

    Ken Gracey

    Yes, we languish under a bad money system and continue to do so because it's considered a waste of time to think about. All the important thinking's already been done for us.
  • jmgjmg Posts: 15,140
    Ken Gracey wrote: »
    Talk about lament. I might call you later and play a lamentful clarinet tune in consideration of this thread. I've got an especially sad tune from the late 1800s that makes me feel the same way when I read some of this thread.

    Lamenting is easy, anyone can do it.

    To cheer yourself up, pop over to the USB testing thread
    http://forums.parallax.com/discussion/163830/usb-testing/p1

    or, the FastSpin Compiler for P2
    http://forums.parallax.com/discussion/164187/fastspin-compiler-for-p2/p1

    or the P2 Simulator..
    http://forums.parallax.com/discussion/161755/p2-simulator/p1

    The parallel testing and development seems to be percolating along nicely :)
  • More importantly, will it be possible to purchase P2 chips with Californian river pebbles?
  • KeithEKeithE Posts: 957
    edited 2016-05-18 23:55
    The latest Time has an article on Financialization by the author of "Makers and Takers: The Rise of Finance and the Fall of American Business".

    Some of their bullet points:

    "· Thanks to 40 years of policy changes and bad decisions, only about 15 % of all the money in our market system actually ends up in the real economy – the rest stays within the closed loop of finance itself.
    · The financial sector takes a quarter of all corporate profits in this country while creating only 4 % of American jobs.
    · The tax code continues to favor debt over equity, making it easier for companies to hoard cash overseas rather than reinvest it on our shores.
    · Our biggest and most profitable corporations are investing more money in stock buybacks than in research and innovation.
    · And, still, the majority of the financial regulations promised after the 2008 meltdown have yet come to pass, thanks to cozy relationship between our lawmakers and the country’s wealthiest financiers."
  • cgraceycgracey Posts: 14,133
    edited 2016-05-19 01:06
    KeithE wrote: »
    The latest Time has an article on Financialization by the author of "Makers and Takers: The Rise of Finance and the Fall of American Business".

    Some of their bullet points:

    "· Thanks to 40 years of policy changes and bad decisions, only about 15 % of all the money in our market system actually ends up in the real economy – the rest stays within the closed loop of finance itself.
    · The financial sector takes a quarter of all corporate profits in this country while creating only 4 % of American jobs.
    · The tax code continues to favor debt over equity, making it easier for companies to hoard cash overseas rather than reinvest it on our shores.
    · Our biggest and most profitable corporations are investing more money in stock buybacks than in research and innovation.
    · And, still, the majority of the financial regulations promised after the 2008 meltdown have yet come to pass, thanks to cozy relationship between our lawmakers and the country’s wealthiest financiers."

    And all this is enabled/aided/abetted by central banking's debt "money", which system is never pointed as the problem it truly is, but always protected and only spoken of as if it's some well-intended fixture necessitated by the existence of the economy.
Sign In or Register to comment.