What we really need now is a new FPGA image with enough to prove the current basic design...
* 16 Cogs 2KB / 512 Longs (less special registers ~16) Dual Port Ram
* 512KB hub (testing can use less)
* 80-100MHz Instructions
* 64 I/O with ADC & DAC (less is fine for testing)
We don't need quad hub, hubexec, hubslots, maths and cordic to start testing. Any of these may be added while we are testing, as can certain other instructions.
Only Chip and Ken can decide what extras go in. All we can do is flesh out proposals - they can be ignored totally if that's what they want.
Personally I have not bothered much with exercising PII on my nano board. Things were just changing too much all the time.
I'm genuinely interested in what prevents us all from doing more testing with what we already have, while we wait for a major release. We have the proposed P2 instuction set, which is a subset of the current instructions. It'll probably have similar execution speed (effective MIPS) on the DE0. The bonus is we'll get more P16X64A cogs into a DE0
From my point of view, in use the changes were relatively minor, the most significant was probably renaming the AUX memory and associated pointers. Many of the other changes were new additions (serial, or hub execution, for instance) and therefore didn't impact existing code.
I'd be curious what OzPropDev's take on it is since he kept his invaders ticking along between releases.
So what factors prevent people from diving into testing stuff with the DE0?
Heater, I am not arguing against much of what you are saying in that larger context, however I am going to continue to play Devil's Advocate for some reasons...
But isn't that the point? It's the short term that needs serving. Not some far distant future. There are customers clamouring at the door now for the PII as far as I can gather from Ken's statements. They might well start drifting away.
Anyway, here and now, this moment is the "long term". It is the end of the long term that was envisaged starting 8 years ago!
As many of said Parallax and the Prop cannot compete with the likes of the ARM. Better not to even go that way. They have to provide something unique. So far the Props Unique Selling Points (USP) have been it's simplicity and it's simplicity and most of all it's simplicity.
That simplicity and flexibility should not be overlooked or underestimated.
Unfortunately, I am not sure where I've read from Ken that there are either NEW customers clamoring for a new Prop. I think I've read that many EXISTING customers would greatly like to see a more powerful chip soon rather than later, however not sure Parallax is in any imminent danger of losing them while it sorts out the current SNAFU.
The biggest disagreement I would have with your comments above, is this false belief that somehow Parallax needs to avoid advancing to avoid competing with the likes of ARM, etc.
The real truth is, ARM is a juggernaut that is relentlessly sweeping everyone away on both the high-end, and the low-end that Parallax competes in. Not to say it is unstoppable, however as it relentlessly improves speed, h/w peripheral on-chip, lowers die area, power requirements, and price, it is going to roll through the Parallax space without even noticing or even being aware of the space. Parallax is going to be in a bind in just a couple of short years I believe, once ARM really starts dual/quad Coring even their 'small' chips, and adding simple 720P+ video/lcd as is happening on the $20-30 variants now.
Assume most of Parallax's volume customers quickly migrate to the new P2 simply for future-proofing.
Sales of P2 will go to n-thousand/month, and P1 will drop by a similar margin.
Bottom line, unless Parallax can keep a higher margin for the P2 than the P1 (which is reported to be almost all gravy), chances are high that there isn't going to be ANY great revenue increase for Parallax.
Just a revenue now split between 2 product lines.
Somehow, the majority of people on the forum seem to think that simply by pulling the P2 out sooner, Parallax is somehow going to have some fountain of NEW money pouring in enough to keep the party going ad infinatum.
Must be that New Math I keep hearing about.
So, assume the P2 comes out at Christmas. The 6-12 months after that you'll see demand for the P2, and proportionate drop in P1. Financially, Parallax should break even or come out ahead, though they could also end up with lower total profit as margins may not be as high on the P2 as it is proportionally to the P1.
How easy it will be for Parallax to come up with the additional $100-200K+ to respin the P2 with hubexec, etc is known only to Parallax. Will it take 1 year, 2, 3 ? During that time, will the juggernaut have stopped its relentless drive to improve, lower costs and battle to take over the lower-end 8-bit market? I sincerely doubt it. LPC is already well on its way with the 8-pin DIP 810.
If you can't grow your base because you can't afford more rapid innovation, or because you make it a point to not compete, then you will see your market dwindle. To say nothing of the educational markets...
John Abshier - Do the forum sales keep Parallax's Prop development afloat, or do Parallax's large customers keep the Prop alive enough for most everyone here to have a neat toy to play with?
Baggers, those are certainly valid points. However ultimately, Ken/Chip are responsible for not keeping to a sensible plan. Aside from 16 Cores, a lot of what the P2 currently has could have been delivered years ago IIRC.
Currently, the goal posts are in sight and reachable, one way or the other. At this point, I am not sure whether Parallax is feeling pressure to release as soon as possible because of potential customer loss, or because of forum pressure.
Missing a shuttle run is irrelevant to the bigger picture, if Parallax ends up not having the ability to pop out another revision in 1-2 years as Ken has said he would like to do.
There is no fountain of money thats going to appear just because the P2 pops out. The cake is a lie too.
Unless Parallax is truely facing some mass abandonment of the P1, I don't see 3 months making any difference except to those who's patience has worn thin here.
Rod1963, I agree whole-heartedly. I've only worked peripherally around a foundry once (Fujitsu) for a short time, however I follow all the talk on semiaccurate regarding AMD fabbing. I think you are mostly correct with your comments on leadtimes, although Parallax may have a somewhat short timeline simple because this is a uC and not a uP.
Perhaps your Devil's Advocate is right, that was a long post but here we go:
1) There won't be any NEW customers for the P2. Why? Well it's a weird thing from an obscure manufacturer that's impossible to use because it's has no peripherals, only room for tiny little native speed programs, performance sucks for big code etc etc. And well we can get 50 cent ARMs that do everything we need.
2) ARM will sweep the board. There is no safe niche for a PII.
3) Old customers will switch from P1 to PII. No new sales growth there.
4) There is no fountain of money that will magically appear. In fact it's unlikely the PII will ever recoup the investment made in it so far. You need to move a lot of chips to get 4 million back.
5) Perhaps Parallax is truly facing some mass abandonment of the P1. Apart from it's novelty value why would educators persist with using it when there are already a hundred, more mainstream, alternatives?
So, it's all hopeless. No body want this weird little chip, Parallax does not have the resources to make it compete performance wise with the likes of ARM. Architecturally it cannot be made to compete with an ARM anyway.
The PII project should be cancelled.
Ah well. It was fun whilst it lasted. Why are we all still here?
Yes, I know Parallax need to make their money back, but it won't be just from P16X64A it'll be their whole inventory, and I also think it won't need to be recovered over the next 6 months, as long as it can pay for itself, and to keep the next iterations going once a year.
Yes, this is only speculation, as I have absolutely no idea of the inner workings of Parallax, especially on the financial side, but I do know that with a lot of companies, a big investment ( like however much has gone into the last x years of the new chip ) would be a long term investment, they wouldn't have done it if they didn't think it would be recoup-able.
Yes, they are a small company, but they have been going for many many years, and have a good customers around the world, and have a good place in education also, it's not like the new chip will be their only form of income.
Heater, lol at why are we still here it's because we love this place, and Chip's creations, it's' a hive of many creative talented likeminded people, yeah the forum can erupt now and again, but heck that's life, and also the fact that forums have no emotion in the text. but it's ultimately our love of programming the propeller chip that we are here to stay. give me a Prop over an ARM any day! yeah it may not be as powerful, or have a full GPU inside, but I know which chip I'd have more fun programming on! maybe we all back the underdog, because we're all underdogs ourselves, who knows! but I do know it's always more fun with a Prop, no matter what the task!
What I'll write will be harsh. Don't take that as a personal attack. We Germans often are too clear and bold in how we say it. Swallow it for the moment, think about it over night, let it settle. But don't think I want to start a fight!
I haven't posted here since years. I just lurked in every quarter of a year to see how the PII is evolving.
And what I saw was the "malady of committee". A lot of people adding requests that add complexity with the result that the product will ever be finished gets further and further away. Featurism it is called in software-engineering. Add feature by feature and never get done.
I liked the Prop, because it was distinct. It was a new way in thinking and programming that opened doors. But as you start using a product, your needs will outgrow what the current product can do. So you either have to look for alternatives or wait for a better version. But as soon as the product matures slower than your needs, you just can quit it. Or be stuck with what you have.
What I missed with the PI was:
More RAM
A C-compiler that is easy to use. The commercial one that was available didn't want to fix simple bugs.
There was an other thing that I really didn't like:
Lack of streamlined and well documented building blocks that could be used in a simple way. For example serial IO. It didn't support number if bits, number of stop-bits and there was something with the start-bit and stop-bit (somehow the wrong way round, I don't remember). Why didn't Parallax sit down and write some flexible, configurable and readable code? Same for I2C. These are tools that are essential for professional development and a commercial success. Just have a look at the app-notes Parallax offers. Sorry, that is lame, that's of very little help. If you are new to the Prop and look at the tools and all you see is this, you quickly leave. "Nice, but ..."
I also never understood the hype about SPIN. Nice for hobbyists, but why should I learn a new language and migrate my C-coding blocks to a new language? "Nice, but ..."
So, to get back the investments, Parallax should:
* Ship something. And don't promise or make plans of a future product that never hits the market.
* Streamline their app-notes and actually write a bunch of them. Provide building blocks like serial I/O, I2C, program chaining, ... that do work, are well documented and are written in a language most do understand:C or ASM that can be tailored with #ifdefs etc. Forget about lame I2C written in SPIN.
* Provide a serious C99 compiler (maybe that already exists, I didn't check).
* Cast a Prop II with more RAM, more speed, more IO, maybe more cogs.
* Realize that all the zillions of libs written in SPIN are not helpful to attract big volume commercial products. You need to be a real fan to wade through all those -SORRY- hobbyist-level libs written in that odd language called SPIN.
Make smaller steps that actually do result in movement and don't plan huge steps that never get done.
Rant over. And understand that this is my personal opinion!
1. The P16X64A (or whatever it is ultimately called) will legitimise the P8X32A. While some P1 sales may be lost to the P2, there will also be a lot more P2 sales and hopefully (and I think it will be so) P1 will benefit too.
2. I am not going to talk money/profit/return (that's Parallax's private info - they have shared more than I think they should with us).
3. While it is true many of us are here because we love the P1... The way it works, its simplicity, its flexibility, and many more features. But most of us are not going to pay the bills. But we can help newcomers understand the features of P1 (and P2) that makes life so interesting. But more than that, it is precisely those features which are not all that easy to understand, that makes the P1 fit into so many applications using exactly one chip. There are so many benefits that flow from this, such as speed to market, code reuse, no interrupt complexities, common inventories, etc, etc, etc. And this is specifically important for professional users to understand.
The P1 is not likely to be destined for some new phone or other such item likely to sell 100's of millions of chips pa. Not that I don't think it could work in some of these markets, but rather those markets have already standardised on ARM processors. The P1 is far more likely to appear in lower volumes where the product is not so price sensitive (ie not where one has to choose a particular chip from an exhaustive family to save a $1 or 2).
Its more a matter that once a company uses a P1 in a product, the follow-on products are more than likely to use the P1 again. Inventory is a large problem for any company, and being able to utilise the very same chip each time makes a huge impact.
I could go on, but in reality, I am only preaching (mostly) to the choir.
What I missed with the PI was:
More RAM
A C-compiler that is easy to use. The commercial one that was available didn't want to fix simple bugs.
By commercial C compiler do you mean PropGCC? Which bugs did we refuse to fix? PropGCC is still under active development. Please let us know about any problems you have with it so we can fix them.
By commercial C compiler do you mean PropGCC? Which bugs did we refuse to fix?
Sorry, you misunderstood that. At the time I quit, PropGCC was not even thought of. I meant the C-compiler from Imagecraft (their name didn't come to my mind when I wrote the previous posting).
Perhaps your Devil's Advocate is right, that was a long post but here we go:
1) There won't be any NEW customers for the P2. Why? Well it's a weird thing from an obscure manufacturer that's impossible to use because it's has no peripherals, only room for tiny little native speed programs, performance sucks for big code etc etc. And well we can get 50 cent ARMs that do everything we need.
2) ARM will sweep the board. There is no safe niche for a PII.
3) Old customers will switch from P1 to PII. No new sales growth there.
4) There is no fountain of money that will magically appear. In fact it's unlikely the PII will ever recoup the investment made in it so far. You need to move a lot of chips to get 4 million back.
5) Perhaps Parallax is truly facing some mass abandonment of the P1. Apart from it's novelty value why would educators persist with using it when there are already a hundred, more mainstream, alternatives?
So, it's all hopeless. No body want this weird little chip, Parallax does not have the resources to make it compete performance wise with the likes of ARM. Architecturally it cannot be made to compete with an ARM anyway.
The PII project should be cancelled.
Ah well. It was fun whilst it lasted. Why are we all still here?
LOL, is that supposed to be subtle or obvious sarcasm pointed my way, I just can't tell?
I don't necessarily mean to be doom and gloom, sometimes it just happens.
However, from the financial POV, I'm pretty confident I am somewhere in the right ballpark.
As a unique mix-and-match build your own customer I/O device, P2 is probably good for those who want to avoid ARM.
When $3-4 quad-core M0's start appearing with video ( or heck, even a single core M0), and 2x every peripheral under the sun, "large" Parallax customers might start to be more interested in saving $5-8 per unit, since thats potentially $5-8x1000/month gravy.
IF Parallax can afford to release biennial updates to the Prop on revenue similar or slightly more than what they've been getting from the P1 alone, then they probably can remain competitive in their 'space' on either cost (older processes are much cheaper but require more silicon), or through Chip ability to innovate additions that are simply unavailable or not targeted by ARM.
So its not all doom and gloom, however I think Parallax is going to forced to start working harder to keep their market from slipping away.
Heater, lol at why are we still here it's because we love this place, and Chip's creations, it's' a hive of many creative talented likeminded people, yeah the forum can erupt now and again, but heck that's life, and also the fact that forums have no emotion in the text. but it's ultimately our love of programming the propeller chip that we are here to stay. give me a Prop over an ARM any day! yeah it may not be as powerful, or have a full GPU inside, but I know which chip I'd have more fun programming on! maybe we all back the underdog, because we're all underdogs ourselves, who knows! but I do know it's always more fun with a Prop, no matter what the task!
Yes, exactly, I'm totally with all of that.
That is why I was very nervous about the previous reincarnation of the PII, with its 1000 instructions and all kinds of bells, whistles and warts it was not a Propeller any more. Getting to grips with all that is not my idea of fun. That is more like work.
And don't give me all that nonsense about the mythical "professional user" who can handle all that complexity. Professionals hate work as much as anyone else. Perhaps even more, as work is time and time is money. That's why when they have spent months and years mastering the thousands of pages of the ARM manuals they don't feel inclined to do that all again for little obvious benefit.
That is why I question all those who want to add features that are intended to get the Prop to compete with the ARM. It will never compete with an ARM. It's a totally different animal. It's like wanting a 555 timer to compete with an Itanium. That is not what it is for.
No fight there. I agree with pretty much all you observations about the Propeller and it's "ecosystem".
The issue of app notes and "standard" libraries for industry standard interfaces has been broached many times over the years. Sadly with the absence of a C compiler for a long time and the emphasis on Spin that never came to be. There was even a drive for "Gold Standard" objects at one time which seems to have fallen by the wayside and was pointless anyway as the "gold standard" did not include ensuring that any PASM code created could be use from C.
...is that supposed to be subtle or obvious sarcasm pointed my way...
No harm or sarcasm intended. I was just following your observations to a logical conclusion. Which is, can the PII project before it's eats another 4 million dollars with no foreseeable return in investment.
Now, actually I'm not comfortable talking about investment, and costs, and returns and the financial ways of Parallax. None of that is any of my business. Or anyone else's outside the company.
I only make the observation that a Propeller is not an ARM. It can never be an ARM. It cannot compete in the ARM space and there is no point to even try.
ARMs may well start sprouting up with 8 or 16 cores. So what?
No the Propeller is something else. It has to offer the engineers of the world something different that they will find useful like. That does not necessarily mean huge consumer market volumes or a super fast processor.
I agree with pretty much all you observations about the Propeller and it's "ecosystem".
Nice to hear that. Years ago, I had the impression that I was more or less alone.
"Ecosystem" is a really good expression! Yes, in my eyes, it is that ecosystem that drove the P2 to where it is now (or not).
I am fully aware that it isn't nice to hear my opinion. I sometimes thought I'd post something around this a few times. But then I thought I couldn't change anything, I already had given up at that time I was active here. But after seeing that thread and reading what Baggers wrote, I thought "Wow! He nailed it down quite good. And I'm not alone".
But as a positive thing, realize that I still do/did come by here to see how the Prop makes. It means I like(d) the Prop and would like Parallax to be successful.
There was even a drive for "Gold Standard" objects at one time which seems to have fallen by the wayside and was pointless anyway as the "gold standard" did not include ensuring that any PASM code created could be use from C.
"Yes, I know Parallax need to make their money back, but it won't be just from P16X64A it'll be their whole inventory, and I also think it won't need to be recovered over the next 6 months, as long as it can pay for itself, and to keep the next iterations going once a year."
I don't disagree much with what you're saying. Its 0400 here though, so I may be getting loopy....or just more so.
However (of course), after spending/investing $1M, $2M, $4M(?), if the only thing Parallax gets out of the now 2 Prop product lines is roughly the same profit or moderately more than they were getting out of the P1, then what was the point of spending those millions for what is essentially almost zero return? Somewhere in here one of my managers would have said, "This is a teachable moment...".
IF that in fact does happen, then once its taken however long for Parallax to clear another $1M-$1.5M, someone/s at Parallax has to ask themselves, do we really want to spend all this money yet again to probably have the same ROI as before.
Or do they consider keeping the status quo with what they have on the Prop-side generating revenue, and investing in a different product type that has a higher chance of better ROI/generating revenue?
For all we know, Parallax may be easily able to fund this work on an ongoing basis based on webstore, educational, and Scenix SX (still alive?) sales. Or not. Sooner or later I just have to believe the old foggies used to teaching the BS are going to be replaced by Arduino/RPi hipsters.
That $1-1.5M would sure buy a lot of neat equipment for their other revenue generating work, or perhaps allow them to consider something even more radical like a quad-core MIPs/Amber ARM-compatible processor (cheap/free license) which incorporates a P1/P2 on-die as some sort of super I/O controller? Fight the ARM with an ARM (older though) ?
Perhaps "Ecosystem" is not the word I was looking for. It implies the combined system of Parallax, companies that build Propeller hardware solution, people like ImageCraft trying to make a go of a compiler, and of course all the users, contributors and forum members.
What I was getting at was the "Propeller system" as provided out of the box by Parallax.
How cool would it be if I could pull up an IDE. Hit a "chip configuration" screen, check some boxes for "UART", "I2C", "SPI", "PWM", whatever and have the code required for them automagically fetched and included into my project. Hit some more check boxes to configure which pins are used and so on. Such drivers and other code would have to be Parallax created or controlled and rigorously tested and documented with data sheets like any other hardware peripheral on other devices.
Oh, and we are working in C/C++ here not some freaky weird other language.
All of a sudden the Propeller is not "that funny little chip with no peripherals" but that amazingly flexible and simple device where I can configure whatever peripherals I like out of the box.
Perhaps "Ecosystem" is not the word I was looking for. It implies the combined system of Parallax, companies that build Propeller hardware solution, people like ImageCraft trying to make a go of a compiler, and of course all the users, contributors and forum members.
So I misunderstood that. But still I think my misunderstanding is true.
Lack of "Gold Code", Imagecraft failing, featurism driven by the forum. Just a bigger ecosystem.
What I was getting at was the "Propeller system" as provided out of the box by Parallax.
A C-compiler with a well engineered set of libraries does exactly that. And if you need to click on checkboxes to select the platform, templates or a simple front-end that does generate the required #includes will do.
Re Arduino, JavaScript, Python:
I had to do some Arduino-programmin for a friend. It simply was terrible. Libraries that didn't work together, programmed by amateurs. Oh wait, that sounds a bit like Prop (sorry again for the harsh words).
No one doing serious software-development would use JS or Python on a µP. Maybe in 10 years. And that crippled C++ from the Arduino is just disgusting.
Yes, I am a C-evangelist. But you should know that I hated it in the beginning (coming from Pascal and well above 100 KLOC). I still would not like C for tenths of KLOC (kilo lines of code).
That is why I was very nervous about the previous reincarnation of the PII, with its 1000 instructions and all kinds of bells, whistles and warts it was not a Propeller any more. Getting to grips with all that is not my idea of fun. That is more like work.
And don't give me all that nonsense about the mythical "professional user" who can handle all that complexity. Professionals hate work as much as anyone else. Perhaps even more, as work is time and time is money. That's why when they have spent months and years mastering the thousands of pages of the ARM manuals they don't feel inclined to do that all again for little obvious benefit.
That is why I question all those who want to add features that are intended to get the Prop to compete with the ARM. It will never compete with an ARM. It's a totally different animal. It's like wanting a 555 timer to compete with an Itanium. That is not what it is for.
haha yeah, I agree, whilst it was turning into a marvellous beast of a chip, without realising, it was losing that innocence, and simplicity, because we were all craving more and more, instead of letting it run it's lifespan in a slow relaxed simple pace, we were wanting it to grow up and be an all singing all dancing super hero of a chip! one that would require an equally mammoth users manual!
Where it was the simplicity that made it fun to program on in the first place, when all it really needed is what we already had, more speed, more hub-ram, more cogs, more io, not more instructions. no matter how great they are, or beneficial! let it grow up naturally enjoy childhood, and prepubescence. lets face it, if it ended up being the P2 we had, the only it could be updated, is by the multi-million investment to go to 65nm tech, where having a few iterations between now and then, would help bring cash in to parallax to be able to do so.
Even in it's current state, it's an awesome micro controller, and I think that's the issue, everyone is wanting it to compare to ARM's processors, if anything it should be being compared to PIC chips, not ARM, and it already beats PIC chips! ok, it may not have USB integrated, but it can do so much more!
My point was only that years ago Chip put BASIC on a little 8 bit micro and it was a great success. Still loved today. That brought micro-controler use to a huge mass of people who would otherwise never have bothered getting to grips with assembler or even C.
At the time people like me would never have looked at it. "BASIC really, be serious". These JavaScript, Python and such efforts are the modern day equivalent.
Pascal was of course devised by a psychopathic control freak.
What I was getting at was the "Propeller system" as provided out of the box by Parallax.
How cool would it be if I could pull up an IDE. Hit a "chip configuration" screen, check some boxes for "UART", "I2C", "SPI", "PWM", whatever and have the code required for them automagically fetched and included into my project. Hit some more check boxes to configure which pins are used and so on. Such drivers and other code would have to be Parallax created or controlled and rigorously tested and documented with data sheets like any other hardware peripheral on other devices.
Oh, and we are working in C/C++ here not some freaky weird other language.
All of a sudden the Propeller is not "that funny little chip with no peripherals" but that amazingly flexible and simple device where I can configure whatever peripherals I like out of the box.
Sounds great Heater.
That has been my vision for years, but have had little encouragement (or time) to pursue it.
When the subject has come up where it should count, I usually got blank or bewildered stares.
Yes, like that. Knew I must have got the idea from somewhere
It's probably not something I would use myself but one important part of the idea is the high quality, Parallax backed, software libraries that go behind it. Whilst raising awareness that the Propeller(s) can actually do all that, and more, easily.
It seems to be a great idea for all those beginners to C/C++ and the Propeller.
Great, you have enough igniter there to start a serious language war thread. However this is not the place. Do come and join in on the The Official JavaScript Religious War thread
There is no need for a language war. That war is over or maybe never has been fought. And if you or someone else thinks that SPIN is much better than C, try to convince someone who picks processors just by what they offer.
Type B: Not enough I/O. Type C: not fast enough. Type not enough RAM. Type P: no C-compiler.
That is how decisions are made, not by "Wow cool, a new language that no one knows. I'll try it for a new project. And if I fail, I can write off even more hours. My boss will be proud of me. But I got the LEDs blinking".
Do you think that anybody wants Occam back on the transputer ... er ... xmos? XC was the right decision. C for most of the code and add a bit of X where you need it.
It is OK if Parallax calls it PC (for Parallax-C or Propeller-C).
There is never any need for a language war. Only it's a way to vent frustrations from time to time:). Just now my working environment requires I deal with Pascal, C/C++, JavaScript and Python. So far I have managed to keep away from the PHP, thank God.
Occam and the Transputer was great. However David May learned the lesson, people won't use weird stuff, and when it came to XMOS stuck to C and the only slightly weird XC dialect.
Perhaps "Ecosystem" is not the word I was looking for. It implies the combined system of Parallax, companies that build Propeller hardware solution, people like ImageCraft trying to make a go of a compiler, and of course all the users, contributors and forum members.
What I was getting at was the "Propeller system" as provided out of the box by Parallax.
How cool would it be if I could pull up an IDE. Hit a "chip configuration" screen, check some boxes for "UART", "I2C", "SPI", "PWM", whatever and have the code required for them automagically fetched and included into my project. Hit some more check boxes to configure which pins are used and so on. Such drivers and other code would have to be Parallax created or controlled and rigorously tested and documented with data sheets like any other hardware peripheral on other devices.
Oh, and we are working in C/C++ here not some freaky weird other language.
All of a sudden the Propeller is not "that funny little chip with no peripherals" but that amazingly flexible and simple device where I can configure whatever peripherals I like out of the box.
Sorry, you misunderstood that. At the time I quit, PropGCC was not even thought of. I meant the C-compiler from Imagecraft (their name didn't come to my mind when I wrote the previous posting).
Nick
Thanks for explaining. I just wanted to make sure there weren't bugs in PropGCC that we were failing to fix.
No harm or sarcasm intended. I was just following your observations to a logical conclusion. Which is, can the PII project before it's eats another 4 million dollars with no foreseeable return in investment.
Now, actually I'm not comfortable talking about investment, and costs, and returns and the financial ways of Parallax. None of that is any of my business. Or anyone else's outside the company.
I only make the observation that a Propeller is not an ARM. It can never be an ARM. It cannot compete in the ARM space and there is no point to even try.
ARMs may well start sprouting up with 8 or 16 cores. So what?
No the Propeller is something else. It has to offer the engineers of the world something different that they will find useful like. That does not necessarily mean huge consumer market volumes or a super fast processor.
---Re financials, in the main I agree. However part of being 'brutal' is trying to be realistic. This happens on every forum for every manufacturer, so I don't see a need to tippy-toe around it.
If anything, because Parallax has such a limited market, its arguably far more relevant than similar at Atmel/Microchip, etc.
---Per your last, yes I agree as to not necessarily needing a huge market to continue along for x amount of time.
As long is they revision out with minor changes on the same 180nm process, then there is probably a fair amount of life.
---Was going to delete this as I know it won't be well received, however since I wrote it, and find it logical, I have to poke the hornets nest.
However as future revisions/process improvements ARE going to cost up to $1M+, what has radically changed the new P2 from the old P1 that is now going to give Engineers any reason to change their mind ? Or better, how long will it take to recoup whatever part of the past sunk costs, plus make $1M+ in profit, to support a process shift?
Honestly, I think the overall market has spoken on the idea of soft-peripherals.... do not want.
While I like 16 faster Cores, I don't see how it changes the value proposition enough to increase sales, which in turn are required to the tune of $1M (net profit) for any frankly fanciful dream of moving to 65nm.
Parallax can obviously still go ahead with the P2 in some form or the other. Its worth a shot since they've already sunk so much.
There is no evidence that the revenue split between P1/P2 is going to change substantially however.
Lots of supposition and 'I want to Believe" fortune telling, but previous little to back it up.
If they don't see some real uptake within 6 months - 1 year, I think reality needs to be faced.
The Prop idea is never going to fly beyond a limited revenue market, including hobbyists and a fading educational market.
Obviously the need for 5 serial/USARTs/ etc, or Engineer interest in having to do the grunt work of coding their own s/w peripherals, just isn't worth any supposed benefit over the 30 minutes it take to look at some parametric table.
If that is true, then Parallax can continue to make massive investments to keep its niche.
I can't see how that is more profitable than many other possible uses for that amount of investment, but only Parallax knows.
I think some rethinking could blend a LOT of Chips ideas with a more acceptable to the mainstream offering that has a much better chance of wowing people.
Comments
* 16 Cogs 2KB / 512 Longs (less special registers ~16) Dual Port Ram
* 512KB hub (testing can use less)
* 80-100MHz Instructions
* 64 I/O with ADC & DAC (less is fine for testing)
We don't need quad hub, hubexec, hubslots, maths and cordic to start testing. Any of these may be added while we are testing, as can certain other instructions.
Only Chip and Ken can decide what extras go in. All we can do is flesh out proposals - they can be ignored totally if that's what they want.
I'm genuinely interested in what prevents us all from doing more testing with what we already have, while we wait for a major release. We have the proposed P2 instuction set, which is a subset of the current instructions. It'll probably have similar execution speed (effective MIPS) on the DE0. The bonus is we'll get more P16X64A cogs into a DE0
From my point of view, in use the changes were relatively minor, the most significant was probably renaming the AUX memory and associated pointers. Many of the other changes were new additions (serial, or hub execution, for instance) and therefore didn't impact existing code.
I'd be curious what OzPropDev's take on it is since he kept his invaders ticking along between releases.
So what factors prevent people from diving into testing stuff with the DE0?
Unfortunately, I am not sure where I've read from Ken that there are either NEW customers clamoring for a new Prop. I think I've read that many EXISTING customers would greatly like to see a more powerful chip soon rather than later, however not sure Parallax is in any imminent danger of losing them while it sorts out the current SNAFU.
The biggest disagreement I would have with your comments above, is this false belief that somehow Parallax needs to avoid advancing to avoid competing with the likes of ARM, etc.
The real truth is, ARM is a juggernaut that is relentlessly sweeping everyone away on both the high-end, and the low-end that Parallax competes in. Not to say it is unstoppable, however as it relentlessly improves speed, h/w peripheral on-chip, lowers die area, power requirements, and price, it is going to roll through the Parallax space without even noticing or even being aware of the space. Parallax is going to be in a bind in just a couple of short years I believe, once ARM really starts dual/quad Coring even their 'small' chips, and adding simple 720P+ video/lcd as is happening on the $20-30 variants now.
Assume most of Parallax's volume customers quickly migrate to the new P2 simply for future-proofing.
Sales of P2 will go to n-thousand/month, and P1 will drop by a similar margin.
Bottom line, unless Parallax can keep a higher margin for the P2 than the P1 (which is reported to be almost all gravy), chances are high that there isn't going to be ANY great revenue increase for Parallax.
Just a revenue now split between 2 product lines.
Somehow, the majority of people on the forum seem to think that simply by pulling the P2 out sooner, Parallax is somehow going to have some fountain of NEW money pouring in enough to keep the party going ad infinatum.
Must be that New Math I keep hearing about.
So, assume the P2 comes out at Christmas. The 6-12 months after that you'll see demand for the P2, and proportionate drop in P1. Financially, Parallax should break even or come out ahead, though they could also end up with lower total profit as margins may not be as high on the P2 as it is proportionally to the P1.
How easy it will be for Parallax to come up with the additional $100-200K+ to respin the P2 with hubexec, etc is known only to Parallax. Will it take 1 year, 2, 3 ? During that time, will the juggernaut have stopped its relentless drive to improve, lower costs and battle to take over the lower-end 8-bit market? I sincerely doubt it. LPC is already well on its way with the 8-pin DIP 810.
If you can't grow your base because you can't afford more rapid innovation, or because you make it a point to not compete, then you will see your market dwindle. To say nothing of the educational markets...
John Abshier - Do the forum sales keep Parallax's Prop development afloat, or do Parallax's large customers keep the Prop alive enough for most everyone here to have a neat toy to play with?
Baggers, those are certainly valid points. However ultimately, Ken/Chip are responsible for not keeping to a sensible plan. Aside from 16 Cores, a lot of what the P2 currently has could have been delivered years ago IIRC.
Currently, the goal posts are in sight and reachable, one way or the other. At this point, I am not sure whether Parallax is feeling pressure to release as soon as possible because of potential customer loss, or because of forum pressure.
Missing a shuttle run is irrelevant to the bigger picture, if Parallax ends up not having the ability to pop out another revision in 1-2 years as Ken has said he would like to do.
There is no fountain of money thats going to appear just because the P2 pops out. The cake is a lie too.
Unless Parallax is truely facing some mass abandonment of the P1, I don't see 3 months making any difference except to those who's patience has worn thin here.
Rod1963, I agree whole-heartedly. I've only worked peripherally around a foundry once (Fujitsu) for a short time, however I follow all the talk on semiaccurate regarding AMD fabbing. I think you are mostly correct with your comments on leadtimes, although Parallax may have a somewhat short timeline simple because this is a uC and not a uP.
Cluso, I agree.
Perhaps your Devil's Advocate is right, that was a long post but here we go:
1) There won't be any NEW customers for the P2. Why? Well it's a weird thing from an obscure manufacturer that's impossible to use because it's has no peripherals, only room for tiny little native speed programs, performance sucks for big code etc etc. And well we can get 50 cent ARMs that do everything we need.
2) ARM will sweep the board. There is no safe niche for a PII.
3) Old customers will switch from P1 to PII. No new sales growth there.
4) There is no fountain of money that will magically appear. In fact it's unlikely the PII will ever recoup the investment made in it so far. You need to move a lot of chips to get 4 million back.
5) Perhaps Parallax is truly facing some mass abandonment of the P1. Apart from it's novelty value why would educators persist with using it when there are already a hundred, more mainstream, alternatives?
So, it's all hopeless. No body want this weird little chip, Parallax does not have the resources to make it compete performance wise with the likes of ARM. Architecturally it cannot be made to compete with an ARM anyway.
The PII project should be cancelled.
Ah well. It was fun whilst it lasted. Why are we all still here?
Yes, this is only speculation, as I have absolutely no idea of the inner workings of Parallax, especially on the financial side, but I do know that with a lot of companies, a big investment ( like however much has gone into the last x years of the new chip ) would be a long term investment, they wouldn't have done it if they didn't think it would be recoup-able.
Yes, they are a small company, but they have been going for many many years, and have a good customers around the world, and have a good place in education also, it's not like the new chip will be their only form of income.
Heater, lol at why are we still here
I haven't posted here since years. I just lurked in every quarter of a year to see how the PII is evolving.
And what I saw was the "malady of committee". A lot of people adding requests that add complexity with the result that the product will ever be finished gets further and further away. Featurism it is called in software-engineering. Add feature by feature and never get done.
I liked the Prop, because it was distinct. It was a new way in thinking and programming that opened doors. But as you start using a product, your needs will outgrow what the current product can do. So you either have to look for alternatives or wait for a better version. But as soon as the product matures slower than your needs, you just can quit it. Or be stuck with what you have.
What I missed with the PI was:
More RAM
A C-compiler that is easy to use. The commercial one that was available didn't want to fix simple bugs.
There was an other thing that I really didn't like:
Lack of streamlined and well documented building blocks that could be used in a simple way. For example serial IO. It didn't support number if bits, number of stop-bits and there was something with the start-bit and stop-bit (somehow the wrong way round, I don't remember). Why didn't Parallax sit down and write some flexible, configurable and readable code? Same for I2C. These are tools that are essential for professional development and a commercial success. Just have a look at the app-notes Parallax offers. Sorry, that is lame, that's of very little help. If you are new to the Prop and look at the tools and all you see is this, you quickly leave. "Nice, but ..."
I also never understood the hype about SPIN. Nice for hobbyists, but why should I learn a new language and migrate my C-coding blocks to a new language? "Nice, but ..."
So, to get back the investments, Parallax should:
* Ship something. And don't promise or make plans of a future product that never hits the market.
* Streamline their app-notes and actually write a bunch of them. Provide building blocks like serial I/O, I2C, program chaining, ... that do work, are well documented and are written in a language most do understand:C or ASM that can be tailored with #ifdefs etc. Forget about lame I2C written in SPIN.
* Provide a serious C99 compiler (maybe that already exists, I didn't check).
* Cast a Prop II with more RAM, more speed, more IO, maybe more cogs.
* Realize that all the zillions of libs written in SPIN are not helpful to attract big volume commercial products. You need to be a real fan to wade through all those -SORRY- hobbyist-level libs written in that odd language called SPIN.
Make smaller steps that actually do result in movement and don't plan huge steps that never get done.
Rant over. And understand that this is my personal opinion!
Nick
2. I am not going to talk money/profit/return (that's Parallax's private info - they have shared more than I think they should with us).
3. While it is true many of us are here because we love the P1... The way it works, its simplicity, its flexibility, and many more features. But most of us are not going to pay the bills. But we can help newcomers understand the features of P1 (and P2) that makes life so interesting. But more than that, it is precisely those features which are not all that easy to understand, that makes the P1 fit into so many applications using exactly one chip. There are so many benefits that flow from this, such as speed to market, code reuse, no interrupt complexities, common inventories, etc, etc, etc. And this is specifically important for professional users to understand.
The P1 is not likely to be destined for some new phone or other such item likely to sell 100's of millions of chips pa. Not that I don't think it could work in some of these markets, but rather those markets have already standardised on ARM processors. The P1 is far more likely to appear in lower volumes where the product is not so price sensitive (ie not where one has to choose a particular chip from an exhaustive family to save a $1 or 2).
Its more a matter that once a company uses a P1 in a product, the follow-on products are more than likely to use the P1 again. Inventory is a large problem for any company, and being able to utilise the very same chip each time makes a huge impact.
I could go on, but in reality, I am only preaching (mostly) to the choir.
Thanks,
David
Sorry, you misunderstood that. At the time I quit, PropGCC was not even thought of. I meant the C-compiler from Imagecraft (their name didn't come to my mind when I wrote the previous posting).
Nick
LOL, is that supposed to be subtle or obvious sarcasm pointed my way, I just can't tell?
I don't necessarily mean to be doom and gloom, sometimes it just happens.
However, from the financial POV, I'm pretty confident I am somewhere in the right ballpark.
As a unique mix-and-match build your own customer I/O device, P2 is probably good for those who want to avoid ARM.
When $3-4 quad-core M0's start appearing with video ( or heck, even a single core M0), and 2x every peripheral under the sun, "large" Parallax customers might start to be more interested in saving $5-8 per unit, since thats potentially $5-8x1000/month gravy.
IF Parallax can afford to release biennial updates to the Prop on revenue similar or slightly more than what they've been getting from the P1 alone, then they probably can remain competitive in their 'space' on either cost (older processes are much cheaper but require more silicon), or through Chip ability to innovate additions that are simply unavailable or not targeted by ARM.
So its not all doom and gloom, however I think Parallax is going to forced to start working harder to keep their market from slipping away.
That is why I was very nervous about the previous reincarnation of the PII, with its 1000 instructions and all kinds of bells, whistles and warts it was not a Propeller any more. Getting to grips with all that is not my idea of fun. That is more like work.
And don't give me all that nonsense about the mythical "professional user" who can handle all that complexity. Professionals hate work as much as anyone else. Perhaps even more, as work is time and time is money. That's why when they have spent months and years mastering the thousands of pages of the ARM manuals they don't feel inclined to do that all again for little obvious benefit.
That is why I question all those who want to add features that are intended to get the Prop to compete with the ARM. It will never compete with an ARM. It's a totally different animal. It's like wanting a 555 timer to compete with an Itanium. That is not what it is for.
No fight there. I agree with pretty much all you observations about the Propeller and it's "ecosystem".
The issue of app notes and "standard" libraries for industry standard interfaces has been broached many times over the years. Sadly with the absence of a C compiler for a long time and the emphasis on Spin that never came to be. There was even a drive for "Gold Standard" objects at one time which seems to have fallen by the wayside and was pointless anyway as the "gold standard" did not include ensuring that any PASM code created could be use from C.
Now, actually I'm not comfortable talking about investment, and costs, and returns and the financial ways of Parallax. None of that is any of my business. Or anyone else's outside the company.
I only make the observation that a Propeller is not an ARM. It can never be an ARM. It cannot compete in the ARM space and there is no point to even try.
ARMs may well start sprouting up with 8 or 16 cores. So what?
No the Propeller is something else. It has to offer the engineers of the world something different that they will find useful like. That does not necessarily mean huge consumer market volumes or a super fast processor.
Nice to hear that. Years ago, I had the impression that I was more or less alone.
"Ecosystem" is a really good expression! Yes, in my eyes, it is that ecosystem that drove the P2 to where it is now (or not).
I am fully aware that it isn't nice to hear my opinion. I sometimes thought I'd post something around this a few times. But then I thought I couldn't change anything, I already had given up at that time I was active here. But after seeing that thread and reading what Baggers wrote, I thought "Wow! He nailed it down quite good. And I'm not alone".
But as a positive thing, realize that I still do/did come by here to see how the Prop makes. It means I like(d) the Prop and would like Parallax to be successful.
Nick
+1
(Blast! Here I am agreeing with you again!)
Ross.
I don't disagree much with what you're saying. Its 0400 here though, so I may be getting loopy....or just more so.
However (of course), after spending/investing $1M, $2M, $4M(?), if the only thing Parallax gets out of the now 2 Prop product lines is roughly the same profit or moderately more than they were getting out of the P1, then what was the point of spending those millions for what is essentially almost zero return? Somewhere in here one of my managers would have said, "This is a teachable moment...".
IF that in fact does happen, then once its taken however long for Parallax to clear another $1M-$1.5M, someone/s at Parallax has to ask themselves, do we really want to spend all this money yet again to probably have the same ROI as before.
Or do they consider keeping the status quo with what they have on the Prop-side generating revenue, and investing in a different product type that has a higher chance of better ROI/generating revenue?
For all we know, Parallax may be easily able to fund this work on an ongoing basis based on webstore, educational, and Scenix SX (still alive?) sales. Or not. Sooner or later I just have to believe the old foggies used to teaching the BS are going to be replaced by Arduino/RPi hipsters.
That $1-1.5M would sure buy a lot of neat equipment for their other revenue generating work, or perhaps allow them to consider something even more radical like a quad-core MIPs/Amber ARM-compatible processor (cheap/free license) which incorporates a P1/P2 on-die as some sort of super I/O controller? Fight the ARM with an ARM (older though) ?
Perhaps "Ecosystem" is not the word I was looking for. It implies the combined system of Parallax, companies that build Propeller hardware solution, people like ImageCraft trying to make a go of a compiler, and of course all the users, contributors and forum members.
What I was getting at was the "Propeller system" as provided out of the box by Parallax.
How cool would it be if I could pull up an IDE. Hit a "chip configuration" screen, check some boxes for "UART", "I2C", "SPI", "PWM", whatever and have the code required for them automagically fetched and included into my project. Hit some more check boxes to configure which pins are used and so on. Such drivers and other code would have to be Parallax created or controlled and rigorously tested and documented with data sheets like any other hardware peripheral on other devices.
Oh, and we are working in C/C++ here not some freaky weird other language.
All of a sudden the Propeller is not "that funny little chip with no peripherals" but that amazingly flexible and simple device where I can configure whatever peripherals I like out of the box.
I think all that is what you were alluding to.
http://http://www.espruino.com/
https://tessel.io/
Or the Python hipsters:
http://micropython.org/
These guys are doing with JS and Python what Chip did with BASIC all those years ago.
P.S. I have a couple of Espruino's and Espruino code running on an STM32F4 Discovery board. Dead simple, dead cheap. Love it.
So I misunderstood that. But still I think my misunderstanding is true.
Lack of "Gold Code", Imagecraft failing, featurism driven by the forum. Just a bigger ecosystem.
A C-compiler with a well engineered set of libraries does exactly that. And if you need to click on checkboxes to select the platform, templates or a simple front-end that does generate the required #includes will do.
Re Arduino, JavaScript, Python:
I had to do some Arduino-programmin for a friend. It simply was terrible. Libraries that didn't work together, programmed by amateurs. Oh wait, that sounds a bit like Prop (sorry again for the harsh words).
No one doing serious software-development would use JS or Python on a µP. Maybe in 10 years. And that crippled C++ from the Arduino is just disgusting.
Yes, I am a C-evangelist. But you should know that I hated it in the beginning (coming from Pascal and well above 100 KLOC). I still would not like C for tenths of KLOC (kilo lines of code).
Nick
haha yeah, I agree, whilst it was turning into a marvellous beast of a chip, without realising, it was losing that innocence, and simplicity, because we were all craving more and more, instead of letting it run it's lifespan in a slow relaxed simple pace, we were wanting it to grow up and be an all singing all dancing super hero of a chip! one that would require an equally mammoth users manual!
Where it was the simplicity that made it fun to program on in the first place, when all it really needed is what we already had, more speed, more hub-ram, more cogs, more io, not more instructions. no matter how great they are, or beneficial! let it grow up naturally enjoy childhood, and prepubescence. lets face it, if it ended up being the P2 we had, the only it could be updated, is by the multi-million investment to go to 65nm tech, where having a few iterations between now and then, would help bring cash in to parallax to be able to do so.
Even in it's current state, it's an awesome micro controller, and I think that's the issue, everyone is wanting it to compare to ARM's processors, if anything it should be being compared to PIC chips, not ARM, and it already beats PIC chips! ok, it may not have USB integrated, but it can do so much more!
Great, you have enough igniter there to start a serious language war thread. However this is not the place. Do come and join in on the The Official JavaScript Religious War thread http://forums.parallax.com/showthread.php/152201-The-Official-JavaScript-Religious-War-Thread. I'm sure we can throw some gasoline on Pascal and other languages there as well:)
My point was only that years ago Chip put BASIC on a little 8 bit micro and it was a great success. Still loved today. That brought micro-controler use to a huge mass of people who would otherwise never have bothered getting to grips with assembler or even C.
At the time people like me would never have looked at it. "BASIC really, be serious". These JavaScript, Python and such efforts are the modern day equivalent.
Pascal was of course devised by a psychopathic control freak.
That has been my vision for years, but have had little encouragement (or time) to pursue it.
When the subject has come up where it should count, I usually got blank or bewildered stares.
Yes, like that. Knew I must have got the idea from somewhere
It's probably not something I would use myself but one important part of the idea is the high quality, Parallax backed, software libraries that go behind it. Whilst raising awareness that the Propeller(s) can actually do all that, and more, easily.
It seems to be a great idea for all those beginners to C/C++ and the Propeller.
There is no need for a language war. That war is over or maybe never has been fought. And if you or someone else thinks that SPIN is much better than C, try to convince someone who picks processors just by what they offer.
Type B: Not enough I/O. Type C: not fast enough. Type
That is how decisions are made, not by "Wow cool, a new language that no one knows. I'll try it for a new project. And if I fail, I can write off even more hours. My boss will be proud of me. But I got the LEDs blinking".
Do you think that anybody wants Occam back on the transputer ... er ... xmos? XC was the right decision. C for most of the code and add a bit of X where you need it.
It is OK if Parallax calls it PC (for Parallax-C or Propeller-C).
Nick
There is never any need for a language war. Only it's a way to vent frustrations from time to time:). Just now my working environment requires I deal with Pascal, C/C++, JavaScript and Python. So far I have managed to keep away from the PHP, thank God.
Occam and the Transputer was great. However David May learned the lesson, people won't use weird stuff, and when it came to XMOS stuck to C and the only slightly weird XC dialect.
All in all I agree with you.
---Re financials, in the main I agree. However part of being 'brutal' is trying to be realistic. This happens on every forum for every manufacturer, so I don't see a need to tippy-toe around it.
If anything, because Parallax has such a limited market, its arguably far more relevant than similar at Atmel/Microchip, etc.
---Per your last, yes I agree as to not necessarily needing a huge market to continue along for x amount of time.
As long is they revision out with minor changes on the same 180nm process, then there is probably a fair amount of life.
---Was going to delete this as I know it won't be well received, however since I wrote it, and find it logical, I have to poke the hornets nest.
However as future revisions/process improvements ARE going to cost up to $1M+, what has radically changed the new P2 from the old P1 that is now going to give Engineers any reason to change their mind ? Or better, how long will it take to recoup whatever part of the past sunk costs, plus make $1M+ in profit, to support a process shift?
Honestly, I think the overall market has spoken on the idea of soft-peripherals.... do not want.
While I like 16 faster Cores, I don't see how it changes the value proposition enough to increase sales, which in turn are required to the tune of $1M (net profit) for any frankly fanciful dream of moving to 65nm.
Parallax can obviously still go ahead with the P2 in some form or the other. Its worth a shot since they've already sunk so much.
There is no evidence that the revenue split between P1/P2 is going to change substantially however.
Lots of supposition and 'I want to Believe" fortune telling, but previous little to back it up.
If they don't see some real uptake within 6 months - 1 year, I think reality needs to be faced.
The Prop idea is never going to fly beyond a limited revenue market, including hobbyists and a fading educational market.
Obviously the need for 5 serial/USARTs/ etc, or Engineer interest in having to do the grunt work of coding their own s/w peripherals, just isn't worth any supposed benefit over the 30 minutes it take to look at some parametric table.
If that is true, then Parallax can continue to make massive investments to keep its niche.
I can't see how that is more profitable than many other possible uses for that amount of investment, but only Parallax knows.
I think some rethinking could blend a LOT of Chips ideas with a more acceptable to the mainstream offering that has a much better chance of wowing people.