If you're going after commercial design wins, price matters. Period. No amount of wishful thinking changes that.
The bigger point is that the two archetectures aren't the same. There isn't a simple dollar/mips equivalency when comparing the two - at least not for applications in which the P1 shines.
Where I use the P1, an ARM or PIC wouldn't suffice. The hard determinism across multiple channels makes all the difference.
Combining an M4 and an M0 is a step in the right direction. But it's still apples and oranges vis-a-vis the Prop.
Xmos and Parallax aren't the only mulitcore players anymore. Freescale, TI and others have gotten into this arena in a big way. Parallax won't be able to claim uniqueness as a selling point....
And that will no doubt have become even more the case by the time P2 actually arrives.
Cost? Well if you're selling chips to instrumentation makers(industrial, medical, etc) their devices that are used professionally are often quite expensive - especially the medical devices. I don't think a $2-$6 price difference would be major factor.
Agreed. However, that's the area I work in most often, and occasionally I will get called out to explain why I added $5 to the BOM (no matter how irrational it might seem). But yes, the importance of price is tied almost directly to volume. In low-volume (high-end) products a few bucks typically won't matter as much as using something that made it substantially easier to engineer the product. On the other hand, $10 or more extra might still break the deal.
Consumer devices are another category altogether, it seems to be a cut throat arena where price matters and every big player is fighting for a piece of the action because it means millions of chips sold.
Yes, which is why much focus shouldn't be placed on that area.
The most comparable device, at first sight, looks like the XS1-G04B-FB144-C4 : 256K RAM, 88 I/O Pins, 4 cores. 8 h/w threads/core, 400 MIPS per core, Unit price = 20 Dollars.
The bigger point is that the two archetectures aren't the same. There isn't a simple dollar/mips equivalency when comparing the two - at least not for applications in which the P1 shines.
Where I use the P1, an ARM or PIC wouldn't suffice. The hard determinism across multiple channels makes all the difference.
Combining an M4 and an M0 is a step in the right direction. But it's still apples and oranges vis-a-vis the Prop.
Whether or not the architectures are the same is of no consequence in this discussion. From a business standpoint it's *not* apples and oranges. Here's why. There are very very few applications where *only* a Propeller would do the job - in fact, I'll go out on a limb and say exactly *zero.* For each and every unique application you can come up with, there will be a myriad of options with which to implement it. Some applications no doubt will have more options available than others, but there *will* be options.
An engineer with an eye on the bottom line knows this and will investigate and judge accordingly. If the application is higher volume, price will be an important if not crucial consideration.
I think what Bill was getting at, and I what I said, is that it has to be at least competitive. If it's 1.5 or 2x (or more) the price of parts that can do comparable things - that's going to be a problem.
I can only go by what was actually written, which was the post I replies to
IMHO, pricing it higher than LPC4370 would seriously limit market penetration, and be a mistake.
If you are now expanding the price zone to be 1.5 or 2x (or more) the price of parts that can do comparable things, the problem is defining the comparable things.
The closest setup to a P2, is an Arm plus some CPLD/FPGA. Users can buy those as two chips, or they can buy FPGAs with ARMs on them.
I see expanding openings where a design has chosen an ARM, and loaded it with some OS. (RasPi is a classic example).
If that user wants to play in the hard-real-time world, they need something else.
how on earth do you know what the people will pay?
Well, that long post missed the mark. Oh well. I'll attempt it again one day. Maybe the short version is to understand what the product will do for the buyer, and then understand how they value that. A price is in there. Too much, and there isn't any value left for them to profit from. Too little, and it may not be considered seriously.
Market research is all about data. Many of the pricing answers are data driven. The more you have, the more you can infer.
Whether of not the architectures are the same is of no consequence in this discussion.
Agreed, for both the options mentioned, and the other basic parallel is the two are highly differentiated devices, not on the beaten path.
There are very very few applications where *only* a Propeller would do the job - in fact, I'll go out on a limb and say exactly *zero.*
Are we really sure about that? I'm asking because that is actually a very important question. If there are use cases, even specialized ones, it is important to identify them. This is a research question, not something resolved in casual conversation here. Frankly, I would put some money on this one to get a very good answer along with appropriate market segmentation.
Of course! But it's not going to be a commodity ARM. And that is specifically what you were comparing.
Why not? It might be, it might not. I suspect there are a very good many applications for which an LPC4370 would be more than adequate, No? Indeed, for many embedded applications (probably most) that chip would in fact be overkill.
Besides, we've also been talking about other things as well (eg, XMOS as above). In fact, it's open season as that goes. Let's see how P2 stacks up when price is considered (as it must be in many if not most instances).
I can only go by what was actually written, which was the post I replies to
Understood. However, nothing wrong with adding some clarification.
If you are now expanding the price zone to be 1.5 or 2x (or more) the price of parts that can do comparable things, the problem is defining the comparable things.
Yes, so let's do that. Give an example use, with specifics, and we'll see what the options are then compare prices (of course, we need to know the P2 target price to do that truly).
I haven't seen much of that happening here.
The closest setup to a P2, is an Arm plus some CPLD/FPGA. Users can buy those as two chips, or they can buy FPGAs with ARMs on them.
They can also buy XMOS parts with ARMs on them (or simply XMOS multicores). Likewise, there are ARMs with PRUs. ARMs (and xmegas) with hardware event systems. Also, FPGAs alone are an option for the most demanding applications. There are other multicore devices, new ones almost every week. Also, I can pair a $6 CPLD with a $2 Cortex-M0+ and get a lot done.
But again, to what applications are we putting all this horsepower? Probably not many, and in any case, price will be a consideration.
I see expanding openings where a design has chosen an ARM, and loaded it with some OS. (RasPi is a classic example).
If that user wants to play in the hard-real-time world, they need something else.
A BeagleBone with PRUs would be another example, which has that capability built-in. And of course there many other pairings that could be used if the end-user simply wants to play in the real world. In the vast majority of instances a $20(?) P2 probably would be overkill (P1 perhaps a better choice).
It isn't. Not in this instance. I've got tubes and tubes of ARM chips, and I've written tens of thousands of lines of code for them. I've got a pretty fair idea of what they are and aren't. In no way can the NVIC provide the determinacy a Prop is capable of.
There is a smidgen of programmable logic in some ARM Cortex chips but it is pretty rudimentary for now.
It isn't. Not in this instance. I've got tubes and tubes of ARM chips, and I've written tens of thousands of lines of code for them. I've got a pretty fair idea of what they are and aren't. In no way can the NVIC provide the determinacy a Prop is capable of.
There is a smidgen of programmable logic in some ARM Cortex chips but it is pretty rudimentary for now.
Yet less so every day... and P2 is still a ways off. So, give an example application in which an ARM simply isn't up to the task. We'll consider what other options there are (including adding external logic to the ARM) and price things out. I'm game.
I see...so we've gone from commercial non-viability to non-viability in higher volumes. I'll take progress any way I can get it.
Nope, nothing's changed. They still have to sell 100k or more(?) parts a year for it to be viable (as in, Parallax can make a profit and stay in business). It's entirely possible even that can't happen if it's priced too high.
Are we really sure about that? I'm asking because that is actually a very important question. If there are use cases, even specialized ones, it is important to identify them. This is a research question, not something resolved in casual conversation here. Frankly, I would put some money on this one to get a very good answer along with appropriate market segmentation.
Yes, I'm pretty sure, when you consider all the other real options out there: FPGAs, other multi-core parts, and all the different pairings that can be made. That doesn't mean, however, that there can't possibly be applications in which the P2 can do it better at the right price. It just seems highly unlikely to me that there are in fact applications which *only* a P2 is capable of performing.
Say the P2 was a few years off yet. Are they going out right now? I think not.
So then, the development is a sunk cost. Funded by P1 sales to date. It's not money in the bank, which is the risk taken, but it's not risk as debt in the more common sense most of us are used to seeing.
Ken was clear on the fact that they have used little credit in this thing. So that changes the game some.
What they need to do is build a business around the P2 as has been done by P1. Early chip sales need to be as brisk as possible, and frankly that should go and fund the efforts to secure larger volume early adopters and fund the product development needed to incorporate the chip into the established P1 business as well.
Ideally, the P1 business largely remains, with the P2 business being additional business by opening new doors in addition to being an improved value add for existing ones.
Be careful with the word viability. Again, tons of people said Apple was not and is not viable in PC land, due to their small share. However, that small share supports the PC business nicely, and that's viable.
It isn't. Not in this instance. I've got tubes and tubes of ARM chips, and I've written tens of thousands of lines of code for them. I've got a pretty fair idea of what they are and aren't. In no way can the NVIC provide the determinacy a Prop is capable of.
There is a smidgen of programmable logic in some ARM Cortex chips but it is pretty rudimentary for now.
Can someone explain where determinism is important? I understand that it's needed for things like UARTs, USB, PWM, and video but those are usually handled by dedicated hardware on other MCUs. Where is general determinism required? I'm not arguing that it isn't necessary. I'm just trying to understand where it is required that might not be handled by other processors like an ARM.
Can someone explain where determinism is important? I understand that it's needed for things like UARTs, USB, PWM, and video but those are usually handled by dedicated hardware on other MCUs. Where is general determinism required? I'm not arguing that it isn't necessary. I'm just trying to understand where it is required that might not be handled by other processors like an ARM.
Well, I for one will be interested to examine a specific application that someone has in mind when they say this. In terms of both what kind of market we're talking (ie, how many chips sales might this possibly represent?) and what other options there might be in the same price range.
Ideally, the P1 business largely remains, with the P2 business being additional business by opening new doors in addition to being an improved value add for existing ones.
Is interest in P1 as high as it once was? How much effort is being made to generate more? Interest sometimes seems to have fallen off, just going by casual observation.
It is good they aren't using debt to fund anything. We agree there.
Be careful with the word viability. Again, tons of people said Apple was not and is not viable in PC land, due to their small share. However, that small share supports the PC business nicely, and that's viable.
Small shares are all I'm talking about, just enough to keep the business profitable and going. I have no expectations that Parallax will become another Microchip, or even another SiLabs, anytime soon. It's possible of course, but I'm not basing anything on that.
Yes, but there was a ramp up / investment period missing from your statement. Got it other wise though. I think the education is doing fine right now. I don't know about the other areas honestly. Not a forever thing by any means, but it's not exactly a cliff either. Frankly, I think Parallax is in a good place, given the P2 arrives with the best of the features being worked on running good.
I like your current line of questions. And David asked about deterministic behavior in a great context. Good discussion right now. Appreciated.
So, give an example application in which an ARM simply isn't up to the task. We'll consider what other options there are (including adding external logic to the ARM) and price things out. I'm game.
I genuinely appreciate the offer of help, but I believe I have the situation in hand. The flexibility and performance the P1 provides have proven to be perfect for the application. Combine that with the fact that Ken and Chip were uncommonly gracious and helpful (as was their product!!!) to a young engineer 20 years ago, and I'm always going to give Parallax the first nod in any equivalency that may arise.
I genuinely appreciate the offer of help, but I believe I have the situation in hand. The flexibility and performance the P1 provides have proven to be perfect for the application. Combine that with the fact that Ken and Chip were uncommonly gracious and helpful (as was their product!!!) to a young engineer 20 years ago, and I'm always going to give Parallax the first nod in any equivalency that may arise.
I appreciate that. I truly do. I like Parallax myself. I like the P1. I like the people here on the forums. But this conversation has to do with sales and markets more generally. All this business stuff isn't fun, and the questions are tough. But these things have to be considered - eventually.
I genuinely appreciate the offer of help, but I believe I have the situation in hand. The flexibility and performance the P1 provides have proven to be perfect for the application. Combine that with the fact that Ken and Chip were uncommonly gracious and helpful (as was their product!!!) to a young engineer 20 years ago, and I'm always going to give Parallax the first nod in any equivalency that may arise.
I didn't think KC_Rob was implying that you needed help. I think he (and I) were trying to understand cases where the Propeller determinism makes something possible that isn't possible on other MCUs. It sounds like you have such an example and it would be interesting to hear about it.
Comments
The bigger point is that the two archetectures aren't the same. There isn't a simple dollar/mips equivalency when comparing the two - at least not for applications in which the P1 shines.
Where I use the P1, an ARM or PIC wouldn't suffice. The hard determinism across multiple channels makes all the difference.
Combining an M4 and an M0 is a step in the right direction. But it's still apples and oranges vis-a-vis the Prop.
http://www.xmos.com/products/silicon/xa-series
http://www.xmos.com/startkit
Agreed. However, that's the area I work in most often, and occasionally I will get called out to explain why I added $5 to the BOM (no matter how irrational it might seem). But yes, the importance of price is tied almost directly to volume. In low-volume (high-end) products a few bucks typically won't matter as much as using something that made it substantially easier to engineer the product. On the other hand, $10 or more extra might still break the deal.
Yes, which is why much focus shouldn't be placed on that area.
XMOS really has all changed since I last looked. But lets look at the XMOS range on Digikey http://www.digikey.com/product-search/en/integrated-circuits-ics/embedded-microcontrollers/2556109?k=xmos
The most comparable device, at first sight, looks like the XS1-G04B-FB144-C4 :
256K RAM, 88 I/O Pins, 4 cores. 8 h/w threads/core, 400 MIPS per core,
Unit price = 20 Dollars.
An engineer with an eye on the bottom line knows this and will investigate and judge accordingly. If the application is higher volume, price will be an important if not crucial consideration.
Are these really 'new devices' as in new die, or are they dual-die package variants - I see they are only BGA.
I can only go by what was actually written, which was the post I replies to
If you are now expanding the price zone to be 1.5 or 2x (or more) the price of parts that can do comparable things, the problem is defining the comparable things.
The closest setup to a P2, is an Arm plus some CPLD/FPGA. Users can buy those as two chips, or they can buy FPGAs with ARMs on them.
I see expanding openings where a design has chosen an ARM, and loaded it with some OS. (RasPi is a classic example).
If that user wants to play in the hard-real-time world, they need something else.
Of course! But it's not going to be a commodity ARM. And that is specifically what you were comparing.
Well, that long post missed the mark. Oh well. I'll attempt it again one day. Maybe the short version is to understand what the product will do for the buyer, and then understand how they value that. A price is in there. Too much, and there isn't any value left for them to profit from. Too little, and it may not be considered seriously.
Market research is all about data. Many of the pricing answers are data driven. The more you have, the more you can infer.
Agreed, for both the options mentioned, and the other basic parallel is the two are highly differentiated devices, not on the beaten path.
Are we really sure about that? I'm asking because that is actually a very important question. If there are use cases, even specialized ones, it is important to identify them. This is a research question, not something resolved in casual conversation here. Frankly, I would put some money on this one to get a very good answer along with appropriate market segmentation.
Besides, we've also been talking about other things as well (eg, XMOS as above). In fact, it's open season as that goes. Let's see how P2 stacks up when price is considered (as it must be in many if not most instances).
Yes, so let's do that. Give an example use, with specifics, and we'll see what the options are then compare prices (of course, we need to know the P2 target price to do that truly).
I haven't seen much of that happening here.
They can also buy XMOS parts with ARMs on them (or simply XMOS multicores). Likewise, there are ARMs with PRUs. ARMs (and xmegas) with hardware event systems. Also, FPGAs alone are an option for the most demanding applications. There are other multicore devices, new ones almost every week. Also, I can pair a $6 CPLD with a $2 Cortex-M0+ and get a lot done.
But again, to what applications are we putting all this horsepower? Probably not many, and in any case, price will be a consideration.
A BeagleBone with PRUs would be another example, which has that capability built-in. And of course there many other pairings that could be used if the end-user simply wants to play in the real world. In the vast majority of instances a $20(?) P2 probably would be overkill (P1 perhaps a better choice).
It isn't. Not in this instance. I've got tubes and tubes of ARM chips, and I've written tens of thousands of lines of code for them. I've got a pretty fair idea of what they are and aren't. In no way can the NVIC provide the determinacy a Prop is capable of.
There is a smidgen of programmable logic in some ARM Cortex chips but it is pretty rudimentary for now.
I see...so we've gone from commercial non-viability to non-viability in higher volumes. I'll take progress any way I can get it.
Say the P2 was a few years off yet. Are they going out right now? I think not.
So then, the development is a sunk cost. Funded by P1 sales to date. It's not money in the bank, which is the risk taken, but it's not risk as debt in the more common sense most of us are used to seeing.
Ken was clear on the fact that they have used little credit in this thing. So that changes the game some.
What they need to do is build a business around the P2 as has been done by P1. Early chip sales need to be as brisk as possible, and frankly that should go and fund the efforts to secure larger volume early adopters and fund the product development needed to incorporate the chip into the established P1 business as well.
Ideally, the P1 business largely remains, with the P2 business being additional business by opening new doors in addition to being an improved value add for existing ones.
Be careful with the word viability. Again, tons of people said Apple was not and is not viable in PC land, due to their small share. However, that small share supports the PC business nicely, and that's viable.
It is good they aren't using debt to fund anything. We agree there.
I like your current line of questions. And David asked about deterministic behavior in a great context. Good discussion right now. Appreciated.
I genuinely appreciate the offer of help, but I believe I have the situation in hand.
Yep. It can be fun actually. Just need context.