Whilst the Prop II piggy back for the Pi might be Pi specific. The general idea is not. This should also work with Beagle Boards, IGEPs and so on. There is a huge array of such boards on the market already and more comming every day. Prices getting lower and lower.
Actually, the BBB looks good to me. I don't have one of those either though. Too busy working with Propeller boards!
Is 250K enough sales for someone manufacturing a micro? Maybe it is. I don't know, but I'm guessing not. And I'm also guessing that Parallax Semiconductor was not founded to tap the likes of the Arduino market, but something rather more vast....
Thing is, the Arduino is a phenomenon.* I'm not so much impressed by where it is today as where it is likely to lead down the road. Today's hobbyists and little kiddies at play will be tomorrow's designers and consumers. The -duino suffix will be recognized and reassuring to people all around the world and apparently blessed because of its magical "open-sourcedness". The Google gurus are estimating 5 billion people getting on-line in the next decade or so thanks to mobile devices, and, if that's true, there will be lots of "hobbyists" out there looking for simple-to-use ready-made microcontroller gadgets to help them solve their various problems. That's potentially an F-load of units. Furthermore, I doubt the Arduino is going to stand still. These simple-minded units will always be available, but brainacs are going to ride the Arduino coat-tails and make systems that are far more sophisticated. It's really a question of who jumps on the bandwagon and who doesn't. I'm afraid if the Propeller doesn't find a spot on that wagon, it's going to find itself neither here nor there - not accepted as a "hobbyist" chip (because it'll be considered too complex compared to the Arduino) and yet not something that can compete with ARMs or whatever. Not that I really know what I'm talking about.
I think the market for the Prop2 lies in connecting the real world with the virtual. Whether it is telepresence, augmented reality, sensor input and output, or any other ideas I can't fathom at the moment.
I think the market for the Prop2 lies in connecting the real world with the virtual. Whether it is telepresence, augmented reality, sensor input and output, or any other ideas I can't fathom at the moment.
Agreed. And it seems like all those analog pins will have something to do with it.
Who else is bold enough to offer 80+ non-multiplexed ADCs on a microcontroller?
I do wonder though how much memory could be available by limiting the analog.
Arduino isn't going away and it's growing, you already can find its development system on PIC32's and ARM's and 16 bit microchip offerings. It's a game changer in that it made microcontrollers accessible to mere normals/artist crowd and those who want more than a BS2 can offer. The P2 should be a choice for those who want to play with the Arduino.
Okay, Parallax doesn't sell the million P2's its hoping for. But it's better than blowing off the market then wondering why the hobbyist world
passed them by as the Arduino starts munching on BS2 sales.
As for the commercial market, there are lots of niches but I don't see a "killer app" for it. I think it's 10 years to late for that. PLC's, instrumentation controllers, testers, maybe part of a low end entertainment system is where I think the P2 will shine.
Agreed. And it seems like all those analog pins will have something to do with it.
Who else is bold enough to offer 80+ non-multiplexed ADCs on a microcontroller?
I do wonder though how much memory could be available by limiting the analog.
Make that 32 ADCs. The ADCs are implemented as part of the counters, with 4 per COG. The DACs are represented similarly, there are 4 DAC drivers per COG, for a total of 32.
However, 32 ADCs and 32DACs are possible at once, plus the remaining 32 digital I/O, so it really doesn't represent much of a limitation, other than requiring coordination among all the COGs.
@jazzed, There was some discussion early on of the tradeoffs involved between memory and other features of the P2. I don't think the ADC circuitry uses much chip area and, particularly if you look at the chip layout, the memory arrays are the major blocks after the cogs themselves. The small amount of area saved by leaving out the ADCs would not buy enough locations to make a difference. One of the reasons for having all those ADCs is the desire to make all the I/O pins and cogs work the same. Any cog or I/O pin can do all the functions of any other cog or I/O pin. They're all programmed the same. The only distinction is that the I/O pins are grouped into ports and each port can be powered separately (but doesn't have to be). Using the same design for all of the cogs and I/O pins also simplifies the overall design and testing.
My understanding from talking with Chip is that you could "on the fly" switch which pins are driven from the cog as DACs, and get 8 (or more) per cog at lower rates. So in theory you could drive all the pins as DACs. I believe this kind of "switching" will work with the ADC mode as well, but Chip would have to confirm.
My understanding from talking with Chip is that you could "on the fly" switch which pins are driven from the cog as DACs, and get 8 (or more) per cog at lower rates. So in theory you could drive all the pins as DACs. I believe this kind of "switching" will work with the ADC mode as well, but Chip would have to confirm.
There may always be something faster, bigger, or more popular than the Prop2, but what about the value of being able to craft code that does things exactly as you want them done, on time, and accurately? That may be the Prop2's value in the market. I don't see anyone else making those kinds of things any easier - just the opposite. An FPGA gives the raw capability to do any digital thing quickly and accurately, but software is much more efficient for forming up functions than writing HDL code.
To me the holy grail in bit-banging space would be the combination of an FPGA and a CPU. From the FPGA guys you get 2 ends of the spectrum, big ARMs or PowerPC in the big FPGAs, and compiled pico/micro Blaze in the smaller FPGAs. From the FPGA vendors and their perspective this all makes sense. What I need is just a little FPGA on top of an ARM CPU (or pick your flavor).
The kinds of applications for this are often some arcane serial protocol.encoding, that is easy to do on the fly with a little logic, but a real pain to do in software (includes bit-stuffing, 4b6b type, manchester encoding and the like. And to do this at high rates of speed. You might be able to bit bang 10 Mbit Ethernet, or USB 1.1, but without some hardware assistance you won't get to 100 Mbit or 480 Mbit USB.
Now I don't know if there are patent issues or not, but so far no one has really mated small FPGA peripherals with an embedded type ARM (M0/M3).
NXP has a new very interesting peripheral called an SGPIO which is basically a bank of configurable shift registers, that can spit out video/ i2s, or take in video, i2s or various other streams. Now if they could add just a little gating to do encode/decode you could do anything.
To me the holy grail in bit-banging space would be the combination of an FPGA and a CPU. From the FPGA guys you get 2 ends of the spectrum, big ARMs or PowerPC in the big FPGAs, and compiled pico/micro Blaze in the smaller FPGAs. From the FPGA vendors and their perspective this all makes sense. What I need is just a little FPGA on top of an ARM CPU (or pick your flavor).
The kinds of applications for this are often some arcane serial protocol.encoding, that is easy to do on the fly with a little logic, but a real pain to do in software (includes bit-stuffing, 4b6b type, manchester encoding and the like. And to do this at high rates of speed. You might be able to bit bang 10 Mbit Ethernet, or USB 1.1, but without some hardware assistance you won't get to 100 Mbit or 480 Mbit USB.
Now I don't know if there are patent issues or not, but so far no one has really mated small FPGA peripherals with an embedded type ARM (M0/M3).
NXP has a new very interesting peripheral called an SGPIO which is basically a bank of configurable shift registers, that can spit out video/ i2s, or take in video, i2s or various other streams. Now if they could add just a little gating to do encode/decode you could do anything.
There may always be something faster, bigger, or more popular than the Prop2, but what about the value of being able to craft code that does things exactly as you want them done, on time, and accurately?
I could not agree more, and I think that's the benefit that Parallax has to drive home again and again and again to potential customers. regardless of their industry sector. The P1, combined with Spin and PASM is amazingly adept for rapid application development (RAD). And the lack of interrupts makes it possible to craft code that "just works" with a minimum of nasty surprises. I'm sure the P2 will shine in that area as well. But Parallax still has to find volume customers who appreciate the benefits of RAD in order for the P2 to be a success.
There may always be something faster, bigger, or more popular than the Prop2, but what about the value of being able to craft code that does things exactly as you want them done, on time, and accurately? That may be the Prop2's value in the market. I don't see anyone else making those kinds of things any easier - just the opposite. An FPGA gives the raw capability to do any digital thing quickly and accurately, but software is much more efficient for forming up functions than writing HDL code.
I couldn't agree more! This is the Prop's strength, and should be marketed as such. That said, though, when it comes to high-volume ("killer") apps engineering time and effort take a back seat to unit cost. It's brutal and ruthless, but that's just how it is. Which is why I and others here think that the Propeller's best chances for commercial viability are to be found in niche applications, lower quantities per design win but where ease of use, flexibility, etc. are given more weight. The bigger apps may come along later -- but it's not a good idea to bank on those being there from the very start.
I want to see a Prop II piggy back board for the Raspi. Do I have to do this myself?:)
An RPi pairing certainly makes sense. A BeagleBone Black somewhat less so, since it has PRUs available to do much of the work one would presumably want to add a Propeller for.
Is 250K enough sales for someone manufacturing a micro? Maybe it is. I don't know, but I'm guessing not.
-Phil
No, it's not enough in terms of chip sales. A minimum of 2 million units/year is a necessary number of chips Parallax must sell to run a viable chip design/fabrication business. This is a volume target with a primary focus of design-in for rapid prototype to production customers, but the number could be decreased by an increased focus on education and hobby uses and selling more of the associated hardware these customers purchase.
We plan on selling the number of chips required to make the business productive, at least a few million a year. But there will be a place for every kind of customer [as there is today] even those who need two chips.
I chose a while ago to sit out of P2 until I can physically hold one, but I still have been monitoring this thread and a few others. This thread has a wealth of valuable points to consider as the P2 gets closer to launch. While many are talking about software and features for marketing the P2, I have a few thoughts on the hardware platform for it:
Design a solid development board that is basic, yet provides a true P2 core system for projects.
Include appropriate power supplies (5, 3.3, 1.8)
Include standard options like SD. (maybe preload an SD bootloader)
Onboard USB is a must.
Design it to fit an off the shelf enclosure. Pick one from Polycase, PacTec, or whomever, but give it a home on day one. Polycase has many cases that use the same base, but have different height lids to give you options for example.
Standardize your user header pinouts. This is #1 in my opinion. If the Arduino layout can have a fundamental design flaw that is ignored because it was kept as a standard, that's proof a standard is critical.
Keep the board size minimal
Do not put a large proto area on the board, but rather have a proto board that drops on available (like the quickstart proto board)
Andy, we'll release our plans for hardware really soon. I think you'll be happy with what we're doing - it's along the lines you're thinking and leaves plenty of opportunities for other people who want to build their own boards. Chip will talk about it on Saturday, too, and maybe show the design in process.
I'm a Lisp fan from way back and would have to make a Propeller Lisp that uses that architecture. I suspect the audience for that sort of thing could be counted on one hand though. :-)
Hi David, I would be one of those ;.) - I started with LISP on a CD6600 Mainframe in 1982 and did my PhD on a Symbolics 3620 Lisp-Machine - with HW GC 8MB RAM and 2 gigantic 150MB disks. But you are probably right - there are only a few of us.
I would like to second WBA Consulting's emphasis on a solid dev board.
The QuickStart was long overdue and a major boon to my company. Since I/O daughterboards are a lot easier to design than SMT processor boards, QS has made it possible for us to fab low-count and one-off commercial products that would have been prohibitively expensive otherwise. Something like QS will be even more important to P2 with its odder power supply requirements making basic board layout more complex. It's a neat trick that you can power a DIP40 P1 from a couple of batteries and alligator clips but P2 will not be amenable to that..
Again, the emphasis on early Parallax boards like the Demoboard and Protoboard was education, but QS can be the heart of an actual field deployed device with a daughterboard instead of the hand assembly labor implied by the Protoboard. Shields are a large part of what drove the Arduino market. A naked Arduino doesn't actually do much useful, but odds are somebody has a shield available that's at least close to what you might want to use it for. Making the PropDuino was probably a good idea but it's still me-tooism. P2 will be so far advanced over Arduino in pin count and capability that making an Arduino shield carrier for it would be just silly. It has a chance to define the new really more powerful standard for such things. And to do it with standard connectors.
Is 250K enough sales for someone manufacturing a micro? Maybe it is. I don't know, but I'm guessing not. And I'm also guessing that Parallax Semiconductor was not founded to tap the likes of the Arduino market, but something rather more vast.
-Phil
It's not. But it is a start. Add that to the RPi 1M, and the new BBB which should also sell like hotcakes.
These are the markets we can help with, and they will aid Parallax.
But the real market for P2 (and P1) for Parallax is to find niche areas where reasonable volumes can be achieved with good returns for their manufacturers. The key here is going to be the flexibility of the P2 (only 1 chip in the stable that does most things) and fast development/productivity. I haven't yet seen those niche or killer products mentioned, but keep the discussion going and we may indeed find some.
Here's a viewpoint I have, which will certainly be controversial with David, but I assure you, there is a rationale.
You'll be happy to know that I've given up on this argument. Spin is nice, C is nice, Lisp is nice, Snobol is nice, everyone has their favorite language and it is probably impossible to convince most people to consider another.
but more like the Project Board... if you have to stick with single holes, believing that we are somehow going to wire-wrap components that refuse to be wire-wrapped... then
please perforate the center section so we can punch it out:) I like the straight line of pin-outs. Maybe in groups? A single header is great for the dedicated few... for the rest of us, easily
available groups of straight lines seem pretty appealing:)
I have a Raspberry Pi... really lame. Probably a good idea at the time. Really a great way to market a cheap board. IMHO waste of time to hook a P2 to it. The worlds greatest micro-controller hooked up to the world's cheapest computer? Why?
When I last looked, they aren't making circuit boards in most parts of the
world... and the one's that want to already have a computer much better than a Raspberry PI. Connectivity is a foregone conclusion:)
How are we doing with Spin2? I like C... never used it. Don't want to use it. Can't see myself using it. Great for business, UUUU...universities and dedicated programmers... not so good for normals.
Spin and PASM are perfect. SPASM2 will be perfecter(sic). I say SPASM because I am hoping there will be more links between Spin2 and PASM2.
Tomorrow gets a little closer all the time.
For me... tomorrow looks a lot like Saturday... I can't be there... again. And I won't even make the live broadcast... but I trust I will be able to crank my computer up and watch it all starting Saturday night
Parallax really isn't like any other company... it's better:)
Comments
I'm hoping that Parallax knows how to avoid mines by now (not gold mines).
Or answer the question why there is no 'one click' debugger in the IDE like elsewhere.
Thing is, the Arduino is a phenomenon.* I'm not so much impressed by where it is today as where it is likely to lead down the road. Today's hobbyists and little kiddies at play will be tomorrow's designers and consumers. The -duino suffix will be recognized and reassuring to people all around the world and apparently blessed because of its magical "open-sourcedness". The Google gurus are estimating 5 billion people getting on-line in the next decade or so thanks to mobile devices, and, if that's true, there will be lots of "hobbyists" out there looking for simple-to-use ready-made microcontroller gadgets to help them solve their various problems. That's potentially an F-load of units. Furthermore, I doubt the Arduino is going to stand still. These simple-minded units will always be available, but brainacs are going to ride the Arduino coat-tails and make systems that are far more sophisticated. It's really a question of who jumps on the bandwagon and who doesn't. I'm afraid if the Propeller doesn't find a spot on that wagon, it's going to find itself neither here nor there - not accepted as a "hobbyist" chip (because it'll be considered too complex compared to the Arduino) and yet not something that can compete with ARMs or whatever. Not that I really know what I'm talking about.
*I can't believe it's me typing these words.
Agreed. And it seems like all those analog pins will have something to do with it.
Who else is bold enough to offer 80+ non-multiplexed ADCs on a microcontroller?
I do wonder though how much memory could be available by limiting the analog.
Arduino isn't going away and it's growing, you already can find its development system on PIC32's and ARM's and 16 bit microchip offerings. It's a game changer in that it made microcontrollers accessible to mere normals/artist crowd and those who want more than a BS2 can offer. The P2 should be a choice for those who want to play with the Arduino.
Okay, Parallax doesn't sell the million P2's its hoping for. But it's better than blowing off the market then wondering why the hobbyist world
passed them by as the Arduino starts munching on BS2 sales.
As for the commercial market, there are lots of niches but I don't see a "killer app" for it. I think it's 10 years to late for that. PLC's, instrumentation controllers, testers, maybe part of a low end entertainment system is where I think the P2 will shine.
Make that 32 ADCs. The ADCs are implemented as part of the counters, with 4 per COG. The DACs are represented similarly, there are 4 DAC drivers per COG, for a total of 32.
However, 32 ADCs and 32DACs are possible at once, plus the remaining 32 digital I/O, so it really doesn't represent much of a limitation, other than requiring coordination among all the COGs.
Yes, it would work.
The kinds of applications for this are often some arcane serial protocol.encoding, that is easy to do on the fly with a little logic, but a real pain to do in software (includes bit-stuffing, 4b6b type, manchester encoding and the like. And to do this at high rates of speed. You might be able to bit bang 10 Mbit Ethernet, or USB 1.1, but without some hardware assistance you won't get to 100 Mbit or 480 Mbit USB.
Now I don't know if there are patent issues or not, but so far no one has really mated small FPGA peripherals with an embedded type ARM (M0/M3).
NXP has a new very interesting peripheral called an SGPIO which is basically a bank of configurable shift registers, that can spit out video/ i2s, or take in video, i2s or various other streams. Now if they could add just a little gating to do encode/decode you could do anything.
The kinds of applications for this are often some arcane serial protocol.encoding, that is easy to do on the fly with a little logic, but a real pain to do in software (includes bit-stuffing, 4b6b type, manchester encoding and the like. And to do this at high rates of speed. You might be able to bit bang 10 Mbit Ethernet, or USB 1.1, but without some hardware assistance you won't get to 100 Mbit or 480 Mbit USB.
Now I don't know if there are patent issues or not, but so far no one has really mated small FPGA peripherals with an embedded type ARM (M0/M3).
NXP has a new very interesting peripheral called an SGPIO which is basically a bank of configurable shift registers, that can spit out video/ i2s, or take in video, i2s or various other streams. Now if they could add just a little gating to do encode/decode you could do anything.
-Phil
No, it's not enough in terms of chip sales. A minimum of 2 million units/year is a necessary number of chips Parallax must sell to run a viable chip design/fabrication business. This is a volume target with a primary focus of design-in for rapid prototype to production customers, but the number could be decreased by an increased focus on education and hobby uses and selling more of the associated hardware these customers purchase.
We plan on selling the number of chips required to make the business productive, at least a few million a year. But there will be a place for every kind of customer [as there is today] even those who need two chips.
- Design a solid development board that is basic, yet provides a true P2 core system for projects.
- Include appropriate power supplies (5, 3.3, 1.8)
- Include standard options like SD. (maybe preload an SD bootloader)
- Onboard USB is a must.
- Design it to fit an off the shelf enclosure. Pick one from Polycase, PacTec, or whomever, but give it a home on day one. Polycase has many cases that use the same base, but have different height lids to give you options for example.
- Standardize your user header pinouts. This is #1 in my opinion. If the Arduino layout can have a fundamental design flaw that is ignored because it was kept as a standard, that's proof a standard is critical.
- Keep the board size minimal
- Do not put a large proto area on the board, but rather have a proto board that drops on available (like the quickstart proto board)
anyhow, that's my 141 guinea francsExactly my idea, that's why I'm working on something like this.
But, I think development software is key too... revolutionary development software, at least for the hobbyist world.
I was just thinking that a hard-hat might be good stencil for P2...
The QuickStart was long overdue and a major boon to my company. Since I/O daughterboards are a lot easier to design than SMT processor boards, QS has made it possible for us to fab low-count and one-off commercial products that would have been prohibitively expensive otherwise. Something like QS will be even more important to P2 with its odder power supply requirements making basic board layout more complex. It's a neat trick that you can power a DIP40 P1 from a couple of batteries and alligator clips but P2 will not be amenable to that..
Again, the emphasis on early Parallax boards like the Demoboard and Protoboard was education, but QS can be the heart of an actual field deployed device with a daughterboard instead of the hand assembly labor implied by the Protoboard. Shields are a large part of what drove the Arduino market. A naked Arduino doesn't actually do much useful, but odds are somebody has a shield available that's at least close to what you might want to use it for. Making the PropDuino was probably a good idea but it's still me-tooism. P2 will be so far advanced over Arduino in pin count and capability that making an Arduino shield carrier for it would be just silly. It has a chance to define the new really more powerful standard for such things. And to do it with standard connectors.
I favor an swept 8 bladed propeller like they use on TurboProp engine or submarines.
These are the markets we can help with, and they will aid Parallax.
But the real market for P2 (and P1) for Parallax is to find niche areas where reasonable volumes can be achieved with good returns for their manufacturers. The key here is going to be the flexibility of the P2 (only 1 chip in the stable that does most things) and fast development/productivity. I haven't yet seen those niche or killer products mentioned, but keep the discussion going and we may indeed find some.
Great idea. that would make an excellent logo.
but more like the Project Board... if you have to stick with single holes, believing that we are somehow going to wire-wrap components that refuse to be wire-wrapped... then
please perforate the center section so we can punch it out:) I like the straight line of pin-outs. Maybe in groups? A single header is great for the dedicated few... for the rest of us, easily
available groups of straight lines seem pretty appealing:)
I have a Raspberry Pi... really lame. Probably a good idea at the time. Really a great way to market a cheap board. IMHO waste of time to hook a P2 to it. The worlds greatest micro-controller hooked up to the world's cheapest computer? Why?
When I last looked, they aren't making circuit boards in most parts of the
world... and the one's that want to already have a computer much better than a Raspberry PI. Connectivity is a foregone conclusion:)
How are we doing with Spin2? I like C... never used it. Don't want to use it. Can't see myself using it. Great for business, UUUU...universities and dedicated programmers... not so good for normals.
Spin and PASM are perfect. SPASM2 will be perfecter(sic). I say SPASM because I am hoping there will be more links between Spin2 and PASM2.
Tomorrow gets a little closer all the time.
For me... tomorrow looks a lot like Saturday... I can't be there... again. And I won't even make the live broadcast... but I trust I will be able to crank my computer up and watch it all starting Saturday night
Parallax really isn't like any other company... it's better:)