Raspberry Pi Pico

[Nominal Animal]

The Raspberry Pi Foundation is a UK charity, with the aim of:

I see, they are exempt from being a contributor to the software ecosystem they exploit, because one government has decided they are a charity.

Right.  Only human individuals can be "freeloaders" (your term, not mine). 

Like I said, I don't really care.  It is you who brought it up:

The only people who a freeloaders seem to be the semi-professionals, who get the platform but give little to the wider ecosystem, then moan that the can't order in quantity and that broadcom doesn't offer technical support for free...

My point is that that is exactly what the Raspberry Pi Foundation itself is doing.

Perhaps you should examine why you feel so strongly that being a company or a charity is sufficient to deflect any criticism of their behaviour?

I repeat, I am not the one claiming there are freeloaders.  You did that yourself.  I know and understand various licenses, and use a lot of them for my own work (from Creative Commons to GPL to proprietary licensing).  I don't think anyone is being a "freeloader", not even the semi-professional people you mentioned, nor the Pi Foundation.  What I don't like, is when people like you spend a lot of effort in using arbitrary definitions and positions (of being Important Commercial Company, or Good Charity I Swear) as shields against criticism of observed behaviour.

All I have done here, is claim that 1) The Raspberry Pi Foundation is hostile to FOSS developers, including to developers whose work they rely on in the "products" they provide, that 2) The Raspberry Pi Foundation is in real life a PR arm of Broadcom (in legal-speak, supported by Broadcom and University of Cambridge Computer Laboratory), and that 3) because of their history with their hardware products, I would be surprised if the Pico did not have similar hardware faults as previous Pi products.

(I am more than a bit angry, because a lot of Raspberry Pi users believe Linux/FOSS is at the root of the hardware/software problems.  This is incorrect; the Pi hardware is utter crap compared to say Samsung Exynos or Amlogic Meson based Linux SBCs.)

I mock them, and their hardware, because that hostility is unworkable in the long term.  It is damaging, because it leads to actions like RedHat did to counter Oracle's (license-abiding) exploitation, which leads to general losses to the entire community.  It is also completely unreasonable, and seems to stem from Broadcom corporate culture and Eben Upton's personal dislike of everything FOSS.

I really wish they'd picked FreeBSD instead of Linux.  It would have fit their model and culture much better.

[brucehoult]

The only people who a freeloaders seem to be the semi-professionals, who get the platform but give little to the wider ecosystem, then moan that the can't order in quantity and that broadcom doesn't offer technical support for free...

"can't order in quantity" is a huge red flag.

"Profit" is not a dirty word the the education aims of the Raspberry Pi Foundation would be served a lot better by making a $1 profit on every unit sold than by making a $1 loss on it. If they don't care about getting their NRE back because of their charitable aims and that's covered by donors then that it up to them. But they need to make a small profit on every unit sold so that if someone says "We want a thousand Pi Zeros" then they say "Coming right up! And thank you!". Or a million.

Production needs to ramp of course, on new products, but not on a product that is now over five years old.

[Rasz]

I am getting the feeling that those who say "Oh that Pi Pico PIO is new and offers exciting possibilities." actually means "Oh that Pi Pico PIO is new [to me] and offers exciting possibilities [because I haven't been exposed to this stuff much]."

Sure, but how many people are interested in using obscure things like Xmos. This being rpi it will expose a ton of people to the idea of software programmable peripherals.
What a lovely thing to have a modern day Amiga Copper, or Atari ANTIC display list :-)

Would be great to have at least a couple fast(1GHz) serdes blocks to do USB 2.0/MIPI in software etc.

1GHz SERDES bolted to a 48MHz M0+ core... yeah that makes perfect sense.

it would if you actually read the datasheed to learn PIO block can have exclusive access to 500MB/s of DMA bw (maybe even 1GB/s out of 2GB/s available overal). It can be fed internally at least >2x faster than its capability to operate GPIOs. Serdes would complement it nicely.

[hamster_nz]

The only people who a freeloaders seem to be the semi-professionals, who get the platform but give little to the wider ecosystem, then moan that the can't order in quantity and that broadcom doesn't offer technical support for free...

"can't order in quantity" is a huge red flag.

In what context? as a hobbyist tinkering? a student learning? or as somebody trying to design a commercial product around a chip or module produced by a not-for-profit charity aimed at education? If it is the latter, then they are sending the correct signal IMO.

"Profit" is not a dirty word the the education aims of the Raspberry Pi Foundation would be served a lot better by making a $1 profit on every unit sold than by making a $1 loss on it. If they don't care about getting their NRE back because of their charitable aims and that's covered by donors then that it up to them. But they need to make a small profit on every unit sold so that if someone says "We want a thousand Pi Zeros" then they say "Coming right up! And thank you!". Or a million.

Production needs to ramp of course, on new products, but not on a product that is now over five years old.

Of course profit is not a dirty word. They are making physical things and everybody has families to feed and need to make a dollar along the way, and cover NRE. But if somebody rocks up and says "We want a Thousand Pi Zeros" they are not obliged to sell them in bulk - they can say "no, we would rather keep our stock going to hobbyists and students, so we will sell one per person through our established channels". If they have done the math, they end up netting a few percent per piece then its all good by me.

If the foundation believes that getting thousands of boards to kids and hobbyists is more important than having thousands of Pis integrated in the back of some 3D printer or a retro arcade cabinet then I am fine with that, even if they are passing up on a few extra $ along the way.

However, if kids or hobbyists can't get boards because they have sold all their stock to a bulk customers for a few extra $ that would suck bigtime, and not be in line with their aims.

I still remember the great Beagleboard shortage of 2015...

[josip]

Oooh.  32 USB Endpoints!  That's nice.(Compares to 16 endpoints on SAMDxx, 7 endpoints on ATmega32u4)

I've never quite understood why there can't be a buffer bucket for "some other endpoint; deal with it in SW." (like IP "protocol" or TCP "port", or ethernet multicast...)

Kinetis M0+ are with 16 bidirectional EP's. So if this 32 are not bidirectional, than this is the same thing. BTW, this is one of the reason (up to 15 CDC's virtual ports on PC with newer OS's like Win10) why I use Kinetis.

[brucehoult]

You're talking as if there is some fixed supply and when they are sold no one else can get one.

That's just wrong.

There won't be anyone in the supply chain, from chips to boards, to assembly, to distribution, who wouldn't prefer to be doing 10x the current volume if they're making a profit doing it.

If anything, ramping up the production volume by 2x or 10x or 100x because they're integrated in the back of some arcade game or industrial equipment would allow dropping the price for everyone, including the students.

[NorthGuy]

Serdes would complement it nicely.

It is a SERDES. It reads parallel from the FIFO and outputs serially (or vice versa).

I couldn't find in the PIO docs anything about how fast its input clock can be and where it can come from (may be I haven't looked well enough). The input clock may be divided, sort of a built-in NCO, but this only can only make it slower. One instruction requires two NCO clock cycles to execute. Hence, a bit can be shifted in or out every second NCO cycle.

[NorthGuy]

However, if kids or hobbyists can't get boards because they have sold all their stock to a bulk customers for a few extra $ that would suck bigtime, and not be in line with their aims.

You sound just like Canada Post - "we're living through difficult times because people send too many parcels". Customers are good for any business, the more the merrier.

[hans]

Oooh.  32 USB Endpoints!  That's nice.(Compares to 16 endpoints on SAMDxx, 7 endpoints on ATmega32u4)

I've never quite understood why there can't be a buffer bucket for "some other endpoint; deal with it in SW." (like IP "protocol" or TCP "port", or ethernet multicast...)

Because the endpoint field is only up 4 bits, 2^4=16 numbers possible. Then OUT and IN tokens determine data direction, for up to 16 bidir or '32' unidir endpoints: https://beyondlogic.org/usbnutshell/usb3.shtml

Hardware must handle IN/OUT tokens immediately (e.g. real-time paced by the host), which is why it's handled in hardware. Software prepares buffers and signals them to hardware they are ready for the host to be read or written into. Having an 'other' endpoint logic handled by SW must then mean it must handle this IRQ incredibly quickly (much faster than 400k or 1Mbit/s I2C slaves with no clock-stretching possible). Perhaps possible on fast chips.. but really not convenient to the user(=programmer).

[SiliconWizard]

However, if kids or hobbyists can't get boards because they have sold all their stock to a bulk customers for a few extra $ that would suck bigtime, and not be in line with their aims.

You sound just like Canada Post - "we're living through difficult times because people send too many parcels". Customers are good for any business, the more the merrier.

As at least two of us said so far, the RP foundation is a CHARITY. It's not just "any business". Their goal is not just to have customers.

Of course, making money in itself allows them to invest more for future projects, so they have to make money, but not if it actually prevents them from fulfilling their stated goal.

And yes, for most of us westerners, $4 or $10 or $20 may not make a big difference even for kids, but for, say, schools in many parts of Africa, a single dollar makes a difference.

[aheid]

I couldn't find in the PIO docs anything about how fast its input clock can be and where it can come from

From the datasheet "3.2.2. Control Flow":

On every system clock cycle, each state machine fetches, decodes and executes one instruction

And in "2.15.3.1. Instances" clk_sys is listed at nominal 125MHz.

[NorthGuy]

And in "2.15.3.1. Instances" clk_sys is listed at nominal 125MHz.

So, if it runs from the system clock, it's 62.5 MHz highest bit rate it can achive. It's 500 Mb/s for 8-GPIO parallel port. Would be 2Gb/s if it could use 32 GPIO, but only 30 is available, and you would need some GPIOs for clock and control.

[NorthGuy]

As at least two of us said so far, the RP foundation is a CHARITY. It's not just "any business". Their goal is not just to have customers.

So, you're suggesting they selling it at dumping (below cost) prices? Then the more they sell the more they lose, and therefore they don't want commercial entities to buy it. This actually makes sense and explains the low price.

[SiliconWizard]

As at least two of us said so far, the RP foundation is a CHARITY. It's not just "any business". Their goal is not just to have customers.

So, you're suggesting they selling it at dumping (below cost) prices? Then the more they sell the more they lose, and therefore they don't want commercial entities to buy it. This actually makes sense and explains the low price.

No quite, I'm just saying that their goal is to "promote the study of computer science and related topics, especially at school level, and to put the fun back into learning computing.", whereas a regular business's goal is to make profits.

Selling chips to Adafruit, for instance, still fulfills that goal IMHO, since this is also the main 'market' Adafruit is serving. Now selling chips or boards to companies that would make products out of them not related to the above goal whatsoever would not. It's OK if it's a minor part of their activities, as I said, to make some cash - which they can further invest in new projects - but probably not OK if it became more than this.

Then, as a charity, they get significant funding through donations and subsidies, so they can afford to sell their products at what would be a loss for a regular commercial company. They also get very good deals with Broadcom and all their other suppliers. So that explains the low prices.

[NorthGuy]

Then, as a charity, they get significant funding through donations and subsidies, so they can afford to sell their products at what would be a loss for a regular commercial company. They also get very good deals with Broadcom and all their other suppliers. So that explains the low prices.

Makes sense. Fortunately I didn't buy it. I wouldn't feel right if my purchase was subsidized by donations designed to help deprived children to get into computing.

[hamster_nz]

Then, as a charity, they get significant funding through donations and subsidies, so they can afford to sell their products at what would be a loss for a regular commercial company. They also get very good deals with Broadcom and all their other suppliers. So that explains the low prices.

Makes sense. Fortunately I didn't buy it. I wouldn't feel right if my purchase was subsidized by donations designed to help deprived children to get into computing.

A little over 2% of the charities funding come donations,, another 3% from other sources (maybe including sponsorship?) , the rest (approx £40M) is funded from their trading activities. .As they are a charity you can look up their accounts on the UK charities register.

... and it isn't for deprived children, it has always been open for all who want to tinker. It was a large part founded in response to the IT circiulum being focused on using applications, not learning about actual computing.

[westfw]

subsidized by donations

Rarely has a "charitable" organization had such successful products.
Your purchase doesn't have to "subsidize" product give-aways, but then the company probably doesn't need to charge the normal "5x parts cost" MSRP, either.

It's an interestingly different model for this sort of thing (unlike OLPC, for example.)  I wonder if RPi gets commented on as much from charitable organization folk, as it does from technical folk?

[westfw]

I've never quite understood why there can't be a buffer bucket for "some other endpoint; deal with it in SW."

Because the endpoint field is only up 4 bits, 2^4=16 numbers possible. Then OUT and IN tokens determine data direction, for up to 16 bidir or '32' unidir endpoints: https://beyondlogic.org/usbnutshell/usb3.shtml

Oh.  I hadn't realized that 16 (*2) was the limit of the protocol.   I guess that makes my point less of an issue.

Hardware must handle IN/OUT tokens immediately (e.g. real-time paced by the host), which is why it's handled in hardware.

I figured it was something like that.Still, I was imagining something like the way ethernet interfaces handle multicast or promiscuous modes.  Handle the hardware needs immediately as necessary, and count on SW to process the actual data quickly enough...

[Rasz]

And in "2.15.3.1. Instances" clk_sys is listed at nominal 125MHz.

So, if it runs from the system clock, it's 62.5 MHz highest bit rate it can achive. It's 500 Mb/s for 8-GPIO parallel port. Would be 2Gb/s if it could use 32 GPIO, but only 30 is available, and you would need some GPIOs for clock and control.

your keyboard SHIFT key is broken, you kept typing Mb/s instead of MB/s. Makes you only one order of magnitude off.
 62.5 MHz * 30 pins = ~230MB/s, half the available bandwidth

[NorthGuy]

So, if it runs from the system clock, it's 62.5 MHz highest bit rate it can achive. It's 500 Mb/s for 8-GPIO parallel port. Would be 2Gb/s if it could use 32 GPIO, but only 30 is available, and you would need some GPIOs for clock and control.

your keyboard SHIFT key is broken, you kept typing Mb/s instead of MB/s. Makes you only one order of magnitude off.
 62.5 MHz * 30 pins = ~230MB/s, half the available bandwidth

30 pins * 62.5 MHz is roughly 2Gb/s (would be 2 Gb/s with 32 pins), isn't it?

[brucehoult]

So, if it runs from the system clock, it's 62.5 MHz highest bit rate it can achive. It's 500 Mb/s for 8-GPIO parallel port. Would be 2Gb/s if it could use 32 GPIO, but only 30 is available, and you would need some GPIOs for clock and control.

your keyboard SHIFT key is broken, you kept typing Mb/s instead of MB/s. Makes you only one order of magnitude off.
 62.5 MHz * 30 pins = ~230MB/s, half the available bandwidth

30 pins * 62.5 MHz is roughly 2Gb/s (would be 2 Gb/s with 32 pins), isn't it?

1.875 Gb/s, which is 234 MB/s.

Rasz perhaps does not understand that b (bits) vs B (bytes) is a deliberate distinction?

[Rasz]

So, if it runs from the system clock, it's 62.5 MHz highest bit rate it can achive. It's 500 Mb/s for 8-GPIO parallel port. Would be 2Gb/s if it could use 32 GPIO, but only 30 is available, and you would need some GPIOs for clock and control.

your keyboard SHIFT key is broken, you kept typing Mb/s instead of MB/s. Makes you only one order of magnitude off.
 62.5 MHz * 30 pins = ~230MB/s, half the available bandwidth

30 pins * 62.5 MHz is roughly 2Gb/s (would be 2 Gb/s with 32 pins), isn't it?

1.875 Gb/s, which is 234 GB/s.

Rasz perhaps does not understand that b (bits) vs B (bytes) is a deliberate distinction?

I can somewhat understand broken SHIFT key, but m+Shift combination resulting in capital G is really bizarre.
1.875 Gb/s is in fact 234, but MB/s.
http://letmegooglethat.com/?q=1.875+Gb%2Fs+to+Megabytes%2Fs

[DrG]

For those who might be interested....

From a few days ago:


and from yesterday (much ESP32-S2 but also RP2040:

[Sudo_apt-get_install_yum]

I think this is an obvious attempt at entering the manufacturing world similar to the raspberry pi compute-module. It is based on a platform with millions of examples and community support, shortening the development time of a "real" embedded system.
 
It would also function as an entry point for students who will keep it in the back of their head when going into the work force.

The Raspberry pi foundation is a hobby-oriented company so the sales volumes are a fraction of what they could be if they could work there way into the manufacturing/large volume manufacturing world.

I don’t se any other explanation for the PICO and compute modules, the modules kind of flopped and I think that they might believe that it is doe to the cost ~35USD so they released the PICO a more "cost effective" solution.

The module is also designed to be soldered directly to carrier boards.

These are just my thought

[brucehoult]

I think this is an obvious attempt at entering the manufacturing world similar to the raspberry pi compute-module. It is based on a platform with millions of examples and community support, shortening the development time of a "real" embedded system.

I recently got a flightradar24 receiver and was a little surprised it came with a Raspberry Pi USB-C power supply.

It turns out it's actually got a Compute Module inside it doing the UI and internet communications duties.

https://www.flightradar24.com/blog/wp-content/uploads/2020/05/IMG_3936-2.jpg