Yeah we all know you despise BASIC and let everyone know you do every chance you get. But it's clear a lot of folks still use it considering Parallax sells a lot of BS2's. Three million by their count That's very good by any measure.
Fact is a lot of little apps don't need Linux or a PI. That's why PicAxes still sell and Mbed and the Teensy are doing well. Throwing a Pi and Linux at simple embedded projects is both ignorant and inelegant.
And the Prop's selling points haven't landed it much of anything except a small group of techies after 8 years. That is no success. I think it's clear Parallax isn't ever going to be a player in the semi-conductor field. It's strength is in supporting education and hobbyist products. Which I think is great because few cater to hobbyists or educators.
We have had that discussion here many times. I do not recall that ever being the goal. There are some nice goals on the table though. Perhaps you should ask Ken about them. I and others have.
Comparing price and specs on the other bazillion devices, all adopting each others features is pretty much useless in the context of this discussion.
BASIC is just great BTW. I think you will find Heater would express a much more complex opinion of it, given a different discussion context.
At present, it has little relevance to Propellers and why they are the way they are and what value that all has.
The little device with 14 vectors has them as a simple and standard way to share the single CPU among peripheral events.
One can easily end up servicing an interrupt only to then service another one, etc... pile it all on the stack and hope it all gets done in time. Each vector provides a simple hardware tag to get this done.
Should Parallax have a near miss, they might just choke up their cash flow and have to lay off a lot of people. It isn't enough to just ponder that or that feature available on another chip. There is a business that has to go forward in good order.
Release the FPGA image for the Propeller 2 ASAP seems like a good first step. It would help us all understand what to expect.
A Spartan-3 with 100 pins single quantity is $6.44. ...
Parallax is now selling DE0 FPGA boards, maybe they should spend the effort on making the programming more DIY friendly if Verilog is the issue.
Parallax do have a larger FPGA board coming for P2, and P1V has been well received and has good education applications.
FPGAs are getting more flexible all the time, tho the bottom-part of the price curve needs to be watched - there is a creeping trend to BGA only, and you do not always see the newest families lowering the entry price.
Yeah we all know you despise BASIC and let everyone know you do every chance you get. But it's clear a lot of folks still use it considering Parallax sells a lot of BS2's. Three million by their count That's very good by any measure
Fair enough. Putting my opinions of various programming languages aside for a minute we can take my statement as " I don't see how providing language system XXXX on a PIC32 or whatever is going to get much traction." and I still assert that.
When the BASIC STAMP was in it's hey day, the 1990's, it was a different world. Micro-controllers were not so readily available to you average Joe, worse still the compilers, programmers, IDEs, were not so readily available, often expensive or limited. Even documentation was a problem, this was before the internet arrived. In that environment the BASIC STAMP was do doubt brilliant and a God send to any one who wanted to dabble in such things. It was a great innovation by Parallax and an opportunity to ship chips with a huge mark up for their closed source language system. This achievement is acknowledged by the creators of the Arduino who were themselves inspired by the STAMP.
What I am saying is that the world has changed. The internet is here. Open source software is here. The Arduino and Pi and a dozen others are here. MCU's abound, they are cents a piece, the tools and information required to use them is open and free. How can the STAMP model make an impression today? Shipping somebody else chip, like a PIC32, is not a money maker and nobody want's to pay for the software, nobody want's to develop for closed source proprietary systems.
A product needs a unique selling point, like making multi-tasking really easy, like being deterministic in it's timing, like the Propeller !
That of course puts the Propeller into a niche market. Especially since it's architecture is so weird and it's programmed in a language nobody has ever heard of.
Fact is a lot of little apps don't need Linux or a PI. That's why PicAxes still sell and Mbed and the Teensy are doing well. Throwing a Pi and Linux at simple embedded projects is both ignorant and inelegant
Very true. I don't believe I have ever said a full up Linux running system is the answer to every problem. It's clearly not suitable for many situations.
My hope is that the PII builds on the PI's unique selling points, with it's language neutrality and greatly expanded capacity, memory, speed, pins, it's ability to run C/C++ easily and so on.
potatohead,
Was the goal ever to be a player?
Hmmm...what was all that "Parallax Semiconductor" thing about a while back?
What I am saying is that the world has changed. The internet is here. Open source software is here.
No doubt this is true and I think the influence of OpenSource is a bit like a double edge sword. On the one hand it increases access to existing technology but on the other hand it takes away cashflow from those innovators who will invest in future innovations. The result is the existing holders of technologies ( the existing big guns ) win and OpenSource builds a moat around their existing products by denying cashflow to potential innovative competitors ( parallax's of the world )
I think much of where parallax finds itself comes down to this paradox.
A product needs a unique selling point, like making multi-tasking really easy, like being deterministic in it's timing, like the Propeller !
I have been developing a hardware companion to my VirtualBreadboard software which I have mentioned in earlier posts. I chose the propeller but I stopped when I saw the new MAX10 FPGA's. But having tried now to get up to speed with the MAX10 I just don't have the time to sink into the learning curve to get the result with the MAX10 so actually I have come back to the propeller and am getting ready to roll out the product. As I have mentioned in the past the propeller is ideal for my application as a Soft-peripheral co-processor.
That of course puts the Propeller into a niche market. Especially since it's architecture is so weird and it's programmed in a language nobody has ever heard of.
I agree. My own application is very unique but here is where I can from personal experience relate to where the propeller has a place
* Require FPGA type flexibility with regular circuit reconfiguration
* Small Engineering Shops who don't have FPGA design experts on hand or budget to outsource to one
* Designing medium volume products without the budget to engage a full FPGA solution but propeller is 'good-enough' at least as an interim step
Unique is good but also it means you don't get selected for mainstream education. For example I don't see much motivation for wide adoption of SPIN as a language to teach. However, propeller ASM stands out as a unique parallel processing language which could conceivably find a place as an introductory course on parallel processing concepts, trade-offs and practical solutions.
This is far different than being the low-hanging-fruit solution that the BASIC STAMP found itself when first introduced.
So really I see the propeller as a BootStrap FPGA device. Perhaps parallax can gear up as a FGPA solutions provider using the Propeller as the fast time to market first product and then provide expertise for getting second iteration products to market based on FPGA solutions that incorporate their propeller IP.
That would be a service I would use.
As for the P2 - I think the problem is it cuts into the space where you should be using an FPGA. The issue is it's to generic in this space and becomes impossible to create a specification for a concrete ASIC because you can literally do anything. I think I understand where they want(ed) to go - get the community to help them define this specification. However it's two impossible experiments at once - an experiment for creating an impossible specification and an experiment of managing a community to create a consensus on an impossible specification.
I think parallax should give up on the P2 right now and keep the IP blocks created this far to be building blocks for the FPGA service mentioned above.
The end result is to use the propeller as time to market first generation proving products service to help select projects and generate income for second generation full FPGA products supported by parallax with P2 function blocks for accelerating development.
Just a humble suggestion for a impossibly competitive economic world.
XMOS devices have been marketed as an FPGA replacement ever since they were launched, and can be programmed in XC. I can't see Propeller devices which have to be programmed in assembler competing very successfully in that market segment.
Re: Open source / closed source...hmmm...OK, I'll bite.
I don't see any "double edge sword" or paradox here. Keeping your ideas, skills, talents, effort (call it "innovation") locked up in secret silos for which you license use of inscrutable binary versions is not good for the overall welfare of the human race. Neither is it good for technological progress.
Yes it takes away cash flow from some. So what? If your creation, for which you are charging rent, is implemented by a bunch of nerdy geeks that make it Free or Open Software that is just tough. Clearly you secret source is not so special after all. (See BASIC STAMP vs Arduino, or Espruino or MicroPython or ....)
A case from history: When Galileo was messing with lenses and making a telescope the purest, clearest glass in the world came from Venice. The secrets of their glass making were closely guarded, so much so that the glass makers had to live on the island of Murano and were not allowed to leave! How much better it is for the world that the secret got out and we could all do that?
Turns out Open Source is good for profit motivated corporations as well. If that software is not your main business but you need it anyway it's better to develop it and open source it and have other corporations, who can also make use of it, collaborate and contribute to the development. Obvious cases here are the Linux kernel and the LLVM compiler. There are a ton of others. People like Google contribute a lot to these things because they need it and it's more efficient have to other companies pool their resources with you.
Parallax can also benefit from Free and Open Source software, prop-gcc for example.
I don't see any "moat" around the Free and Open Source world. I see it expanding at an ever increasing rate!
Re: Propeller / FPGA.
I agree, the Prop can fit application where doing it in software on a regular single CPU MCU is complex or impossible but doing it with an FPGA is way more expensive and complex to develop.
I sometimes imagine "grown up" FPGAs where the logic blocks are actually processors, hundreds, thousands of them. The interconnect matrix is configurable. Devices like the Propeller and the XMOS are small steps in that direction.
Re: Spin
Well, yeah, nobody teaching programming to beginners will want to be using that, given that it is only usable on a Propeller. No body teaching algorithms and so on will want to be using Spin, given that there are so many other languages that can express the ideas far more clearly.
Luckily, Spin is so simple that anyone who can program in anything else should be able to get the hang of it in short order.
I have never understood the idea that the P2 or P1 could compete in the low end FPGA world. You can build a P2 with an FPGA, does not mean you can build an FPGA with a P2.
The P2 is struggling to get released in 180 nm technology, FPGAs are using 20 nm technology these days. Now if you could build a P* in 20 nm you might have something, but that is way beyond the cash and human resources of Parallax, and trying to bootstrap to it is crazy when it takes 10 years to do an iteration (remember Moore's Law doubling every 18 months).
There have been numerous FPGAs with embedded processor cores, none have really seemed to take off.
To me the better solution to the low end smart circuit is a small FPGA paired with a small ARM, both you can now buy for a total less than $10 in small quantities. Is that beyond the DIY capability? I really don't think so in an open source world, where shared peripheral code in the FPGA would be available.
Parallax still has its education expertise, it should use the current off the shelf technology to update it.
I agree, the Prop can fit application where doing it in software on a regular single CPU MCU is complex or impossible but doing it with an FPGA is way more expensive and complex to develop.
Along those lines, why not use the existing P1 verilog description to make smaller faster Props in 90nm technology? Different package sizes/pin counts could be offered. Reduced cog count models could also be offered. Seems like development costs would be low and heat generation not a problem.
I'd flip at the opportunity to buy a small and fast 2-4 cog Prop.
IMHO, the only problem with P1 was Code protection. Industry isn't interested if somebody vad copy their product. So P2 have more potential with code protection. Theoretically if P2 cost the same price as Small Arm+CANbus than theoretically every modern car could have many P2's inside. Another P1 problem is "useless". OKey, You tech students basic electronics and programming with P1 and...? Nothing more. Engineers can't use P1 in commercial appication (in general), so they migrate to ARM, AVR, etc. So, I student start with Mainline platform, than he will never migrate to P1 (or P2). P2 invasion must not just start, but must to continue. That is what I expect from P2.
Nothing personal JDat, but the P2 doing CAN can not compete on price with existing solutions. A quick search of Digikey shows 2500 ARM variants with CAN and code protection available starting at a price less than $3. And all of those solutions are single chip, no external Flash needed.
Too much of what I have seen here is pie in the sky thinking, and with another deadline missed, I think it is time for a reality check. Again I think Parallax should give Chip a budget to finish the P2, if that gets exceeded then maybe it is time to punt.
But in the meantime Parallax should update their offerings, or they will get further behind.
Brucee! It's OK. Cars and CAN was like an example. The idea: P1 can't be on professional market. P2 should!
8 years in past... How looked arduino on 2006? How looked linux on ARM? How looked Raspberry PI style devices? Hol looked embedded development on linux? Remember WindowsCE? Everything is changing. P2 need to be... You know... It's no only about the silicon P2. It's about market niche.
I have never understood the idea that the P2 or P1 could compete in the low end FPGA world. You can build a P2 with an FPGA, does not mean you can build an FPGA with a P2.
It's simple. We start from the idea that any logical functionality you can imagine can be implemented as software and run on a CPU, or it can be implemented as a mass of logic gates in circuit designed to do whatever it is. Software and a sea of gates are logically equivalent.
Of course a logic gate implementation soon becomes huge, complex, power hungry and inflexible. So we don't do that, we make a CPU and write software to generate instructions for it.
On the other hand software is slow. So when we have identified commonly used, standard, useful functionality we bake that into silicon to get the speed up. Like UARTs, USB, video, etc, etc.
With that in mind, we might find there are tasks for which a traditional single CPU machine is too slow, but a FPGA is to expensive, complex and power hungry. Enter the Propeller and the XMOS devices. Speed and flexibility for that middle ground.
...the better solution to the low end smart circuit is a small FPGA paired with a small ARM...
But I think the P2 has to have a target market, which the P1 never had. What we use to call it at HP was a solution looking for a problem.
The market has changed and while the P2 has been a blank sheet of paper, it is all things to all people. But something in the mid-teens dollar wise has a tremendous amount of competition out there.
I understand the concept of fast deterministic bit banging, but 80 MHz doesn't cut it. Sure you can do a UART, but that technology is ancient history. Can it do USB 2.0? Not what I have seen so far.
It can do low res video, but that is not real unique anymore. The P2 doesn't have the horsepower or memory to handle a JPG decode let alone an MPG decode.
I have a Zynq too, but those are pretty pricey for a low end project, but I also have lots of ARMs and lots of small Spartan devices sitting in my bin that can do all that.
...the P2 doing CAN [or the function of you choice] can not compete on price with existing solutions. A quick search of Digikey shows 2500 ARM variants with CAN [or the function of you choice] and code protection available starting at a price less than $3. And all of those solutions are single chip, no external Flash needed.
I think you have hit the nail on the head there. There is no point in Parallax trying to compete with those devices. The market is flooded, the profit margins are slim, there are entrenched channels filling those needs. Parallax could never hope to play that game.
If Parallax wants to make a chip, which Chip clearly does, it had better be something unique, weird and wonderful. Flexible and easy to use. Does that sound like a Propeller?
That might only attract 0.00001% of the global MCU market. So be it.
And what's with the code protection thing? I have been involved in projects using microprocessors and MCU in embedded systems for three decades and not one of them used any code protection. Sounds like an unnecessary distraction.
As for USB and such, well, that's what FTDI is for
And video. I'd drop that. Makes no sense unless you can do HDMI and have a bucket of memory.
That might only attract 0.00001% of the global MCU market.
This number need only keep Parallax sustainable and meet it's owners financial goals. Remember that when framing this stuff in terms of "the market." (not directed at anyone in particular)
I disagree on analog video. Streaming signals in the P1 is basically limited to video and a few niche uses people found. In the P2, this is going to be a more generalized thing with a lot more use.
But, let's discuss video alone for a moment. Despite many claims, analog signals remain in wide use. They will continue to remain in wide use for quite a while yet. HDMI is pretty nice, but it also places considerable demands on anything using it too. Those range from IP / compliance concerns, to baseline memory / throughput requirements that are often quite expensive relative to to the tasks at hand. The case for analog means is made right there. It's not a universal case, just one that clearly justifies analog as an option.
One nice niche for these chips may well be industrial controls, and that's either refurbishing / replacing existing ones, or making new ones. Video requirements for these vary considerably and analog capable hardware is inexpensive and flexible. Not to mention getting a pixel perfect, frame accurate display is expensive on HDMI, and largely trivial on analog. How one chooses to use memory varies considerably too. It's possible to toss a full buffer up there, or use a few K to dynamically display things too. All of that can be packed into a little library for others to just use as well, meaning the lower level complexity does not have to be a significant consideration.
Finally, all the P2 is supposed to be doing is offering VGA support at pixel clocks that meet the HDTV standards. The latter is implied, but very clearly an appropriate design goal, as we've seen it in the images offered to date. That, plus a more generalized and bi-directional ability to stream signals, is very, very useful, given all the analog support the chip will have due to the smart pin type design. The more power hungry and esoteric / games related hardware won't come along, and that's a great decision. In trade, we get 16 COGS and a target clock of 200Mhz, which is more than enough to support doing a video signal on anything you can name in software. Perfect. That's how Props should be.
The idea that useful video requires a "bucket of memory" and HDMI is complete bunk.
Code protection is just one of those things... I know I won't need / use it, unless required to. And that's the thing right there. It's a compliance / qualifier for many contracts / projects. If the P2 has it, great! More doors open. Given the ongoing escalation of IP considerations, having that option is going to make a lot of sense. None of us has to go there, of course, and a lot of us won't. But it's there for those who will or have to. Fine by me.
That same IP escalation speaks highly of analog means and methods too. Want to build it to do whatever you want? Really? That's not HDMI. No joke. There are baseline requirements associated with the license to use the tech. Bury that in the chip, and that chip gets encumbered with all those things and who wants that? Better to let somebody else manage that stuff, and just send a nice, open signal to that package and do what it allows you to do.
And what's with the code protection thing? I have been involved in projects using microprocessors and MCU in embedded systems for three decades and not one of them used any code protection. Sounds like an unnecessary distraction.
Sorry, i mean something like in AVR/PIC/etc what prevent copy/paste cloning for devices. P1 for teaching P2 for teaching and beyond... It's no necessary to get in to mainstream market. It's Okay if Parallax will get 0.00001% of market. It's better than 0.0000001%.
Despite many claims, analog signals remain in wide use.
As it happens today I took the last two devices in the house that can accept good old analogue video to the recycling centre. TVs that is. That leaves a couple of monitors that still have VGA inputs. My guess is that in a year or so there will be no such things available anymore.
The idea that useful video requires a "bucket of memory" and HDMI is complete bunk.
Having worked on projects using gigantic, vector graphic displays, that could show complex maps and text and so on, in colour no less, all the while using only 10s of K of memory I have to agree with you. They were radar displays back in the day. Think original Asteroids consoles on steroids:)
Meanwhile we are getting stuck, with a processor at one end, a display at the other, and some money grabbing rent seekers in the middle http://www.hdmi.org/manufacturer/terms.aspx how we ever let that happen I have no idea.
Unless of course we can drive whatever parallel signalling it is that nice LCD panels use now a days...
Sorry, i mean something like in AVR/PIC/etc what prevent copy/paste cloning for devices. P1 for teaching P2 for teaching and beyond... It's no necessary to get in to mainstream market. It's Okay if Parallax will get 0.00001% of market. It's better than 0.0000001%.
I don't know how big the market is, but let's say it's a billion-dollar market. If Parallax gets 0.00001% of the market that's is 10**9/10**7, which is only $100. I doubt if Parallax would be OK with only making $100 per year on the P2. Even if it were a 100 billion-dollar market that still only be $10,000 per year. That's hardly worth the effort.
I don't know how big the market is, but let's say it's a billion-dollar market. If Parallax gets 0.00001% of the market that's is 10**9/10**7, which is only $100. I doubt if Parallax would be OK with only making $100 per year on the P2. Even if it were a 100 billion-dollar market that still only be $10,000 per year. That's hardly worth the effort.
I wonder what percentage of the current micro controller market the P1 has? Maybe the P2 could have that same size piece. Would that be sufficient? To some extent, it seems like the P2 is really Chip's design for himself and, if other people like it, they are welcome to use it. Isn't that essentially what the P1 was? It seems quite a few people wanted to use the P1. If similar numbers want to use the P2, will that be sufficient? If not, how many more are needed? 10x the P1 sales?
It is also worth noting the education opportunity. P2 may well expand the practical scope.
I suppose but I still have my doubts that it will be adopted widely for education unless it is also accepted by industry. Why teach people how to program a unique architecture if it doesn't get used for practical applications? I suppose it might be used in a survey course like Lisp and Snobol used to be just to show that there are other ways of doing things. Will that be enough?
I wonder what percentage of the current micro controller market the P1 has? Maybe the P2 could have that same size piece. Would that be sufficient? To some extent, it seems like the P2 is really Chip's design for himself and, if other people like it, they are welcome to use it. Isn't that essentially what the P1 was? It seems quite a few people wanted to use the P1. If similar numbers want to use the P2, will that be sufficient? If not, how many more are needed? 10x the P1 sales?
I'm not sure what Parallax's business goals are, but I think they would be happy if they could recoup their investment in P2 development and pay their employee's salaries. I'm guessing they've spent a million dollars on developing the P2, plus maybe another million on salaries for people that have worked on the P2 and the 1-2-3 FPGA board. So they need to recoup at least 2 million dollars just for P2 development. If we divide that over a 5-year period that $200,000 per year. However, that's minor compared to paying the salaries for 40+ employees. With overhead and other expenses we're probably talking in the range of at least $4million/year.
Let's say that the P2 and products based on the P2 will be responsible for 25% of Parallax's revenue. And assume that Parallax has a 40% margin. So Parallax will need to gross $2.5 million in sales just on P2 products, or $10 million over all. My concern for Parallax is that if they don't deliver the P2 soon they will get to the point where they have to start cutting back on expenses, which normally means cutting back on the number of employees. So it's really in Parallax's best interest that they get the P2 out ASAP.
I have never understood the idea that the P2 or P1 could compete in the low end FPGA world. You can build a P2 with an FPGA, does not mean you can build an FPGA with a P2.
True, but the point is not really to expect a P2 to swallow HDL, but to be able to meet the same specs, in some projects (not all).
That means apps like multi channel PWM, or instrumentation timing, could use a P2/P1 that would otherwise have to use a FPGA
There have been numerous FPGAs with embedded processor cores, none have really seemed to take off.
To me the better solution to the low end smart circuit is a small FPGA paired with a small ARM
FPGA and MPU are well established now, but they have had a checkered past.
The pairing of FPGA & smaller MCUs (AVR,8051) was proven to miss critical mass, and even ARM pairings took a few starts to get going - Altera had ARM, then swung to NIOS and now finally back to ARM again....
It is certainly more flexible to use a small MCU and a Small FPGA, which I think is why finding a niche for FPGA+ARM has taken a while.
On-chip MPU makes sense where you need the higher bandwidth connections, but those are more complex systems.
Packaging is another reason for low volume designers to favour Small MCU and Small FPGA/CPLD.
My concern for Parallax is that if they don't deliver the P2 soon they will get to the point where they have to start cutting back on expenses, which normally means cutting back on the number of employees. So it's really in Parallax's best interest that they get the P2 out ASAP.
I don't know what to think about this. If their goal is to attract a much larger customer base for the P2, it might not matter much how frustrated we P1 users get waiting. Most of their customers for P2 won't be current P1 customers anyway. If that's the case, making sure that the design is what they really want might be more important than getting it out "real soon now". On the other hand, if they expect current P1 customers to promote the new P2 and grow the market that way then getting P2 out before all of us get frustrated and drift away will be important. It's probably actually a combination of both and I certainly won't claim to be able to predict which will be more important.
Comments
I'm hoping to get an 8 core 64 bit ARM soon https://www.96boards.org/products/ce/hikey/
Propellers they are not.
Yeah we all know you despise BASIC and let everyone know you do every chance you get. But it's clear a lot of folks still use it considering Parallax sells a lot of BS2's. Three million by their count That's very good by any measure.
Fact is a lot of little apps don't need Linux or a PI. That's why PicAxes still sell and Mbed and the Teensy are doing well. Throwing a Pi and Linux at simple embedded projects is both ignorant and inelegant.
And the Prop's selling points haven't landed it much of anything except a small group of techies after 8 years. That is no success. I think it's clear Parallax isn't ever going to be a player in the semi-conductor field. It's strength is in supporting education and hobbyist products. Which I think is great because few cater to hobbyists or educators.
We have had that discussion here many times. I do not recall that ever being the goal. There are some nice goals on the table though. Perhaps you should ask Ken about them. I and others have.
Comparing price and specs on the other bazillion devices, all adopting each others features is pretty much useless in the context of this discussion.
BASIC is just great BTW. I think you will find Heater would express a much more complex opinion of it, given a different discussion context.
At present, it has little relevance to Propellers and why they are the way they are and what value that all has.
The little device with 14 vectors has them as a simple and standard way to share the single CPU among peripheral events.
One can easily end up servicing an interrupt only to then service another one, etc... pile it all on the stack and hope it all gets done in time. Each vector provides a simple hardware tag to get this done.
Totally agree. Price and specs simply don't capture the most outstanding aspects of the Propeller.
Should Parallax have a near miss, they might just choke up their cash flow and have to lay off a lot of people. It isn't enough to just ponder that or that feature available on another chip. There is a business that has to go forward in good order.
Release the FPGA image for the Propeller 2 ASAP seems like a good first step. It would help us all understand what to expect.
Parallax do have a larger FPGA board coming for P2, and P1V has been well received and has good education applications.
FPGAs are getting more flexible all the time, tho the bottom-part of the price curve needs to be watched - there is a creeping trend to BGA only, and you do not always see the newest families lowering the entry price.
When the BASIC STAMP was in it's hey day, the 1990's, it was a different world. Micro-controllers were not so readily available to you average Joe, worse still the compilers, programmers, IDEs, were not so readily available, often expensive or limited. Even documentation was a problem, this was before the internet arrived. In that environment the BASIC STAMP was do doubt brilliant and a God send to any one who wanted to dabble in such things. It was a great innovation by Parallax and an opportunity to ship chips with a huge mark up for their closed source language system. This achievement is acknowledged by the creators of the Arduino who were themselves inspired by the STAMP.
What I am saying is that the world has changed. The internet is here. Open source software is here. The Arduino and Pi and a dozen others are here. MCU's abound, they are cents a piece, the tools and information required to use them is open and free. How can the STAMP model make an impression today? Shipping somebody else chip, like a PIC32, is not a money maker and nobody want's to pay for the software, nobody want's to develop for closed source proprietary systems.
A product needs a unique selling point, like making multi-tasking really easy, like being deterministic in it's timing, like the Propeller !
That of course puts the Propeller into a niche market. Especially since it's architecture is so weird and it's programmed in a language nobody has ever heard of. Very true. I don't believe I have ever said a full up Linux running system is the answer to every problem. It's clearly not suitable for many situations.
My hope is that the PII builds on the PI's unique selling points, with it's language neutrality and greatly expanded capacity, memory, speed, pins, it's ability to run C/C++ easily and so on.
potatohead, Hmmm...what was all that "Parallax Semiconductor" thing about a while back?
No doubt this is true and I think the influence of OpenSource is a bit like a double edge sword. On the one hand it increases access to existing technology but on the other hand it takes away cashflow from those innovators who will invest in future innovations. The result is the existing holders of technologies ( the existing big guns ) win and OpenSource builds a moat around their existing products by denying cashflow to potential innovative competitors ( parallax's of the world )
I think much of where parallax finds itself comes down to this paradox.
I have been developing a hardware companion to my VirtualBreadboard software which I have mentioned in earlier posts. I chose the propeller but I stopped when I saw the new MAX10 FPGA's. But having tried now to get up to speed with the MAX10 I just don't have the time to sink into the learning curve to get the result with the MAX10 so actually I have come back to the propeller and am getting ready to roll out the product. As I have mentioned in the past the propeller is ideal for my application as a Soft-peripheral co-processor.
I agree. My own application is very unique but here is where I can from personal experience relate to where the propeller has a place
* Require FPGA type flexibility with regular circuit reconfiguration
* Small Engineering Shops who don't have FPGA design experts on hand or budget to outsource to one
* Designing medium volume products without the budget to engage a full FPGA solution but propeller is 'good-enough' at least as an interim step
Unique is good but also it means you don't get selected for mainstream education. For example I don't see much motivation for wide adoption of SPIN as a language to teach. However, propeller ASM stands out as a unique parallel processing language which could conceivably find a place as an introductory course on parallel processing concepts, trade-offs and practical solutions.
This is far different than being the low-hanging-fruit solution that the BASIC STAMP found itself when first introduced.
So really I see the propeller as a BootStrap FPGA device. Perhaps parallax can gear up as a FGPA solutions provider using the Propeller as the fast time to market first product and then provide expertise for getting second iteration products to market based on FPGA solutions that incorporate their propeller IP.
That would be a service I would use.
As for the P2 - I think the problem is it cuts into the space where you should be using an FPGA. The issue is it's to generic in this space and becomes impossible to create a specification for a concrete ASIC because you can literally do anything. I think I understand where they want(ed) to go - get the community to help them define this specification. However it's two impossible experiments at once - an experiment for creating an impossible specification and an experiment of managing a community to create a consensus on an impossible specification.
I think parallax should give up on the P2 right now and keep the IP blocks created this far to be building blocks for the FPGA service mentioned above.
The end result is to use the propeller as time to market first generation proving products service to help select projects and generate income for second generation full FPGA products supported by parallax with P2 function blocks for accelerating development.
Just a humble suggestion for a impossibly competitive economic world.
Check out what OzPropDev has been doing to make getting P1's (P1V's) going with Max10 (and other) hardware:-
http://forums.parallax.com/showthread.php/160327-Custom-P1-Verilog-code-generator
Re: Open source / closed source...hmmm...OK, I'll bite.
I don't see any "double edge sword" or paradox here. Keeping your ideas, skills, talents, effort (call it "innovation") locked up in secret silos for which you license use of inscrutable binary versions is not good for the overall welfare of the human race. Neither is it good for technological progress.
Yes it takes away cash flow from some. So what? If your creation, for which you are charging rent, is implemented by a bunch of nerdy geeks that make it Free or Open Software that is just tough. Clearly you secret source is not so special after all. (See BASIC STAMP vs Arduino, or Espruino or MicroPython or ....)
A case from history: When Galileo was messing with lenses and making a telescope the purest, clearest glass in the world came from Venice. The secrets of their glass making were closely guarded, so much so that the glass makers had to live on the island of Murano and were not allowed to leave! How much better it is for the world that the secret got out and we could all do that?
Turns out Open Source is good for profit motivated corporations as well. If that software is not your main business but you need it anyway it's better to develop it and open source it and have other corporations, who can also make use of it, collaborate and contribute to the development. Obvious cases here are the Linux kernel and the LLVM compiler. There are a ton of others. People like Google contribute a lot to these things because they need it and it's more efficient have to other companies pool their resources with you.
Parallax can also benefit from Free and Open Source software, prop-gcc for example.
I don't see any "moat" around the Free and Open Source world. I see it expanding at an ever increasing rate!
Re: Propeller / FPGA.
I agree, the Prop can fit application where doing it in software on a regular single CPU MCU is complex or impossible but doing it with an FPGA is way more expensive and complex to develop.
I sometimes imagine "grown up" FPGAs where the logic blocks are actually processors, hundreds, thousands of them. The interconnect matrix is configurable. Devices like the Propeller and the XMOS are small steps in that direction.
Re: Spin
Well, yeah, nobody teaching programming to beginners will want to be using that, given that it is only usable on a Propeller. No body teaching algorithms and so on will want to be using Spin, given that there are so many other languages that can express the ideas far more clearly.
Luckily, Spin is so simple that anyone who can program in anything else should be able to get the hang of it in short order.
The P2 is struggling to get released in 180 nm technology, FPGAs are using 20 nm technology these days. Now if you could build a P* in 20 nm you might have something, but that is way beyond the cash and human resources of Parallax, and trying to bootstrap to it is crazy when it takes 10 years to do an iteration (remember Moore's Law doubling every 18 months).
There have been numerous FPGAs with embedded processor cores, none have really seemed to take off.
To me the better solution to the low end smart circuit is a small FPGA paired with a small ARM, both you can now buy for a total less than $10 in small quantities. Is that beyond the DIY capability? I really don't think so in an open source world, where shared peripheral code in the FPGA would be available.
Parallax still has its education expertise, it should use the current off the shelf technology to update it.
Along those lines, why not use the existing P1 verilog description to make smaller faster Props in 90nm technology? Different package sizes/pin counts could be offered. Reduced cog count models could also be offered. Seems like development costs would be low and heat generation not a problem.
I'd flip at the opportunity to buy a small and fast 2-4 cog Prop.
Too much of what I have seen here is pie in the sky thinking, and with another deadline missed, I think it is time for a reality check. Again I think Parallax should give Chip a budget to finish the P2, if that gets exceeded then maybe it is time to punt.
But in the meantime Parallax should update their offerings, or they will get further behind.
8 years in past... How looked arduino on 2006? How looked linux on ARM? How looked Raspberry PI style devices? Hol looked embedded development on linux? Remember WindowsCE? Everything is changing. P2 need to be... You know... It's no only about the silicon P2. It's about market niche.
Of course a logic gate implementation soon becomes huge, complex, power hungry and inflexible. So we don't do that, we make a CPU and write software to generate instructions for it.
On the other hand software is slow. So when we have identified commonly used, standard, useful functionality we bake that into silicon to get the speed up. Like UARTs, USB, video, etc, etc.
With that in mind, we might find there are tasks for which a traditional single CPU machine is too slow, but a FPGA is to expensive, complex and power hungry. Enter the Propeller and the XMOS devices. Speed and flexibility for that middle ground. I believe that is called a "zynq": http://www.xilinx.com/products/silicon-devices/soc/zynq-7000.html
Neat idea. I have one on my desk. Problem is we now have the complexity of creating software for the ARM and HDL for the FPGA.
But I think the P2 has to have a target market, which the P1 never had. What we use to call it at HP was a solution looking for a problem.
The market has changed and while the P2 has been a blank sheet of paper, it is all things to all people. But something in the mid-teens dollar wise has a tremendous amount of competition out there.
It can do low res video, but that is not real unique anymore. The P2 doesn't have the horsepower or memory to handle a JPG decode let alone an MPG decode.
I have a Zynq too, but those are pretty pricey for a low end project, but I also have lots of ARMs and lots of small Spartan devices sitting in my bin that can do all that.
If Parallax wants to make a chip, which Chip clearly does, it had better be something unique, weird and wonderful. Flexible and easy to use. Does that sound like a Propeller?
That might only attract 0.00001% of the global MCU market. So be it.
And what's with the code protection thing? I have been involved in projects using microprocessors and MCU in embedded systems for three decades and not one of them used any code protection. Sounds like an unnecessary distraction.
As for USB and such, well, that's what FTDI is for
And video. I'd drop that. Makes no sense unless you can do HDMI and have a bucket of memory.
This number need only keep Parallax sustainable and meet it's owners financial goals. Remember that when framing this stuff in terms of "the market." (not directed at anyone in particular)
I disagree on analog video. Streaming signals in the P1 is basically limited to video and a few niche uses people found. In the P2, this is going to be a more generalized thing with a lot more use.
But, let's discuss video alone for a moment. Despite many claims, analog signals remain in wide use. They will continue to remain in wide use for quite a while yet. HDMI is pretty nice, but it also places considerable demands on anything using it too. Those range from IP / compliance concerns, to baseline memory / throughput requirements that are often quite expensive relative to to the tasks at hand. The case for analog means is made right there. It's not a universal case, just one that clearly justifies analog as an option.
One nice niche for these chips may well be industrial controls, and that's either refurbishing / replacing existing ones, or making new ones. Video requirements for these vary considerably and analog capable hardware is inexpensive and flexible. Not to mention getting a pixel perfect, frame accurate display is expensive on HDMI, and largely trivial on analog. How one chooses to use memory varies considerably too. It's possible to toss a full buffer up there, or use a few K to dynamically display things too. All of that can be packed into a little library for others to just use as well, meaning the lower level complexity does not have to be a significant consideration.
Finally, all the P2 is supposed to be doing is offering VGA support at pixel clocks that meet the HDTV standards. The latter is implied, but very clearly an appropriate design goal, as we've seen it in the images offered to date. That, plus a more generalized and bi-directional ability to stream signals, is very, very useful, given all the analog support the chip will have due to the smart pin type design. The more power hungry and esoteric / games related hardware won't come along, and that's a great decision. In trade, we get 16 COGS and a target clock of 200Mhz, which is more than enough to support doing a video signal on anything you can name in software. Perfect. That's how Props should be.
The idea that useful video requires a "bucket of memory" and HDMI is complete bunk.
Code protection is just one of those things... I know I won't need / use it, unless required to. And that's the thing right there. It's a compliance / qualifier for many contracts / projects. If the P2 has it, great! More doors open. Given the ongoing escalation of IP considerations, having that option is going to make a lot of sense. None of us has to go there, of course, and a lot of us won't. But it's there for those who will or have to. Fine by me.
That same IP escalation speaks highly of analog means and methods too. Want to build it to do whatever you want? Really? That's not HDMI. No joke. There are baseline requirements associated with the license to use the tech. Bury that in the chip, and that chip gets encumbered with all those things and who wants that? Better to let somebody else manage that stuff, and just send a nice, open signal to that package and do what it allows you to do.
Sorry, i mean something like in AVR/PIC/etc what prevent copy/paste cloning for devices. P1 for teaching P2 for teaching and beyond... It's no necessary to get in to mainstream market. It's Okay if Parallax will get 0.00001% of market. It's better than 0.0000001%.
Meanwhile we are getting stuck, with a processor at one end, a display at the other, and some money grabbing rent seekers in the middle http://www.hdmi.org/manufacturer/terms.aspx how we ever let that happen I have no idea.
Unless of course we can drive whatever parallel signalling it is that nice LCD panels use now a days...
Let's say that the P2 and products based on the P2 will be responsible for 25% of Parallax's revenue. And assume that Parallax has a 40% margin. So Parallax will need to gross $2.5 million in sales just on P2 products, or $10 million over all. My concern for Parallax is that if they don't deliver the P2 soon they will get to the point where they have to start cutting back on expenses, which normally means cutting back on the number of employees. So it's really in Parallax's best interest that they get the P2 out ASAP.
That means apps like multi channel PWM, or instrumentation timing, could use a P2/P1 that would otherwise have to use a FPGA
FPGA and MPU are well established now, but they have had a checkered past.
The pairing of FPGA & smaller MCUs (AVR,8051) was proven to miss critical mass, and even ARM pairings took a few starts to get going - Altera had ARM, then swung to NIOS and now finally back to ARM again....
It is certainly more flexible to use a small MCU and a Small FPGA, which I think is why finding a niche for FPGA+ARM has taken a while.
On-chip MPU makes sense where you need the higher bandwidth connections, but those are more complex systems.
Packaging is another reason for low volume designers to favour Small MCU and Small FPGA/CPLD.