PossibleRP2040
Posts: 10
Joined: Sat Apr 10, 2021 6:03 am

RP2040 + Future RPXXXX (RP4340?) Naming

Thu Apr 15, 2021 5:00 pm

In the livestream, Upton decoded the RP2040 name.

The number 2 is for two cores because the RP2040's CPU is dual core. The number 0 is for the CPU, an Arm Cortex M0+, the number 4 is for 256KB of RAM 2 ^ 4 x 16KB. The final digit, 0 refers to no flash. Upton then went on to invent another chip to illustrate the convention using a quad core Arm M3 CPU with 256KB of RAM and no flash which he said would be called a 4340, if it existed.

https://www.tomshardware.com/news/raspb ... co-decoded

Image
https://www.cnx-software.com/2021/01/21 ... ex-m0-mcu/
Looks to have been posted back in January.

User avatar
rpdom
Posts: 18493
Joined: Sun May 06, 2012 5:17 am
Location: Chelmsford, Essex, UK

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Thu Apr 15, 2021 5:17 pm

Yes. This is well known. Your point?
Unreadable squiggle

W. H. Heydt
Posts: 14645
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Thu Apr 15, 2021 5:38 pm

The RPT does not pre-announce products. Dr. Upton's "illustrative example" is just that, a contrived example of how the naming convention would work and not a hint that such a chip is under development.

I wouldn't doubt for a minute that the next chip *is* in development, but that example is very likely NOT it.

PossibleRP2040
Posts: 10
Joined: Sat Apr 10, 2021 6:03 am

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Thu Apr 15, 2021 6:04 pm

Agreeing with W. H. Heydt here. That's not the right example but it is at least the right assumption in general of what the next RP2040 derivative is likely to be. Curious if they will still push the cheapest side of things or follow Arduino's (or the opposite of Tesla's for that matter) approach and go with a MEGA instead of just an UNO next time.

hippy
Posts: 9650
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Thu Apr 15, 2021 6:44 pm

I think this is pertinent - https://www.tomshardware.com/uk/news/ra ... co-decoded -
Upton went on to say that, if he can do one more chip with his naming convention, then it will be successful. Later in the stream, Upton revealed that "we are unlikely to do more than one or two products with RP2040 in them" so the next 'Pi Silicon' board could be the last from Raspberry Pi to feature the RP2040.
The way I read that is; there may be another RP2040-based board from RPT, there may be a future board using some RPxxxx variant, and at least one RPxxxx variant is on the cards to call it a success.

Given RPT historically underestimates the potential and interest in its products, the surprise which frequently accompanies how successful they are, I am certain that's only going to push them into producing at least one more chip. RPT appear to have spent $3-$4 million on developing the RP2040, and are probably well on the way to recuperating that from Pico boards alone. To me it's a no-brainer to push onward; industry interests and sales into that sector are just icing on the cake.

I suspect the RP2040 was intended to target industry, but Pico and similar board successes are making it clear the hobbyist and maker interest in RPxxxx microntroller boards is a viable market in its own right, a market RPT can also dominate.

The big question for me is whether the Pico cannibalises Pi Zero and Zero W sales. And that provokes further thoughts.

My prediction; a four core RP4xxx will be delivered with WiFi and Bluetooth on a board, and that will replace the Zero W. I imagine the questions 'how can we make full-HD HDMI viable?', 'how can we have more of everything?' on such a thing are already being researched.

PossibleRP2040
Posts: 10
Joined: Sat Apr 10, 2021 6:03 am

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Thu Apr 15, 2021 7:01 pm

I did post that pertinent URL at the top of my first post here.
Raspberry Pi Pico is in a backlog, with 1.1 million orders being taken and 3 to 400,000 orders being shipped. This leaves a backlog of approximately 800,000 Raspberry Pi Picos, and there are around 60 to 80,000 chips manufactured per week.
Is the Zero even really available in quantity, four plus years later? Seems to still not be but they can make 60,000 to 80,000 Picos per week. It is already being cannibalized but the Zero is also not a microcontroller either.

Really not sure we need EVERY microcontroller board to be Wi-Fi and Bluetooth and so forth by default. Especially not in an industrial environment.

If they have made their money back on just the chip sold alone, they are making a ton more outside of that.

WestfW
Posts: 68
Joined: Tue Nov 01, 2011 9:56 pm

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Thu Apr 15, 2021 10:36 pm

Interesting (unfortunate?) that the naming convention doesn't include any encoding of "number of package pins"

User avatar
Gavinmc42
Posts: 5444
Joined: Wed Aug 28, 2013 3:31 am

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Fri Apr 16, 2021 12:58 am

Code: Select all

Interesting (unfortunate?) that the naming convention doesn't include any encoding of "number of package pins"
Stick a letter on the end for footprint?
8 pin dip package would be nice :D

Probably the next change will be a rerating and qualifying them at a high speed?
They do seen to be able to be overclocked a lot, 400%+

Memory takes up lots of the die area, if they double that then throw in a few more cores?
I would like to see something more interesting added, FPGA or NN/ML block.
Something useful for IoT.

Not sure if adding RF to the die would be easy and testing it would add lots to any release schedule.
Dual die packaging with someone else's known RF tech?

Count down to locking 3, 2 , 1
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

User avatar
scruss
Posts: 4015
Joined: Sat Jun 09, 2012 12:25 pm
Location: Toronto, ON
Contact: Website

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Fri Apr 16, 2021 1:42 am

PossibleRP2040 wrote:
Thu Apr 15, 2021 7:01 pm
Is the Zero even really available in quantity, four plus years later? Seems to still not be ...
You can buy as many Zero WH's as you want. Have been able to for years.
But you can only buy one Zero W. Them's the rules.
‘Remember the Golden Rule of Selling: “Do not resort to violence.”’ — McGlashan.
Pronouns: he/him

W. H. Heydt
Posts: 14645
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Fri Apr 16, 2021 3:21 am

Gavinmc42 wrote:
Fri Apr 16, 2021 12:58 am
Memory takes up lots of the die area, if they double that then throw in a few more cores?
The RP2040 is a 40nm part. If they try for a 28nm part next go around, they could go for 4 cores and more memory, while still keeping the same chip size.

Alternatively, following some fabrication trends...fab the memory on top of the processor silicon.

I don't see the Pico as a general replacement for the Pi0. They're quite different beasts. I do see the Pico stripping off those things that an MCU can do that people are doing with Pi0s, though. That would ease to supply pressure on Pi0s and also get away from the idee' fixe that people have that it's a loss leader.

What is going to be very interesting is: What happens when RP2040 chip packages hit the retail channels? Are people going to start developing their own RP2040 boards? That is certainly within the capability of the more sophisticated amateurs/hobbyists.

User avatar
Gavinmc42
Posts: 5444
Joined: Wed Aug 28, 2013 3:31 am

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Fri Apr 16, 2021 3:45 am

What is going to be very interesting is: What happens when RP2040 chip packages hit the retail channels? Are people going to start developing their own RP2040 boards? That is certainly within the capability of the more sophisticated amateurs/hobbyists.
Opportunity for tiny vapour phase reflow oven or mini hot plate soldering?
Not the easiest things to solder.

Just ordered my 2nd one so I can use it as a programmer.
Hmm, tiny PicoProg programmer debug PCB?

I do like the fact that the tools work on Pi's, other MCU's I use are stuck with Windows tools.
The Cypress PSoC's I use have a snap off programmer/debugger.
Handy little things for i2c/uart testing as well.
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

hippy
Posts: 9650
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Fri Apr 16, 2021 1:29 pm

PossibleRP2040 wrote:
Thu Apr 15, 2021 7:01 pm
Really not sure we need EVERY microcontroller board to be Wi-Fi and Bluetooth and so forth by default. Especially not in an industrial environment.
They don't need it but if it's minimal cost to include it then it's a big USP to have it even if not used.

There's no reason not to produce a board which doesn't have it fitted but, given third-party supply of RPxxxx boards, they could choose not to.
W. H. Heydt wrote:
Fri Apr 16, 2021 3:21 am
I don't see the Pico as a general replacement for the Pi0. They're quite different beasts.
It depends how one looks at it; if an RPxxxx can do what one was going to do with a Zero then any differences have pretty much disappeared.

Superseding the Zero means RPT can stop buying single-core BCM2835 chips, can provide a 64-bit only OS going forward.

As Eben says; "We've always seen ourselves as a PC company". That would give a clear delineation of future product lines; PC and microcontroller.

It won't be an easy decision, there are pro's and con's both ways, but it's a possibility, and feasible when it wasn't before, when there was no cost-effective alternative for the BCM2835.

As usual we won't know until whatever happens, if anything does.

User avatar
HermannSW
Posts: 4008
Joined: Fri Jul 22, 2016 9:09 pm
Location: Eberbach, Germany
Contact: Website Twitter YouTube

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Fri Apr 16, 2021 3:14 pm

Since RPxxxx are microcontrollers, I doubt they will ever run Raspberry Pi OS, and because of that will never replace a Pi0[w].
https://stamm-wilbrandt.de/2wheel_balancing_robot
https://stamm-wilbrandt.de/en#raspcatbot
https://github.com/Hermann-SW/Raspberry_v1_camera_global_external_shutter
https://github.com/Hermann-SW/raspiraw
https://stamm-wilbrandt.de/en/Raspberry_camera.html

W. H. Heydt
Posts: 14645
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Fri Apr 16, 2021 7:17 pm

PossibleRP2040 wrote:
Thu Apr 15, 2021 7:01 pm
I did post that pertinent URL at the top of my first post here.
Raspberry Pi Pico is in a backlog, with 1.1 million orders being taken and 3 to 400,000 orders being shipped. This leaves a backlog of approximately 800,000 Raspberry Pi Picos, and there are around 60 to 80,000 chips manufactured per week.
Is the Zero even really available in quantity, four plus years later? Seems to still not be but they can make 60,000 to 80,000 Picos per week. It is already being cannibalized but the Zero is also not a microcontroller either.
While they may be making 60-80K *Picos* per week, that won't be true of *RP2040s*. It's been stated that they get about 20K RP2040 chips per 300mm wafer. One doesn't make a wafer in a week, and it would be silly (and, possibly, not possible) to order 3 or 4 wafers ever week. Lead time from order-to-chip-packages is around 3 to 4 months (there are queues in addition to the time it takes to fab wafers), and probably longer now due to general demand for chips outstripping manufacturing capacity (you may have seen news about *that*).

Note also that assembling Picos is done at Sony Pencoed, so they compete with other Pis for assembly line time, where Picos are assembled in a Sony facility in Japan and don't have that contention problem. (And, one may hope, there aren't tens of thousands of them sitting in a container on the MV Ever Green waiting for all concerned to sort out who is going to pay who what for tying up the Suez Canal for most of a week. For those not following the news, the Egyptian authorities have impounded the ship until someone coughs up $900 million, and there is no sign of any attempt to transfer the cargo.)

(Corrected that it's Picos being made in Japan, not Pi0s.)
Last edited by W. H. Heydt on Mon Apr 19, 2021 1:00 am, edited 2 times in total.

fivdi
Posts: 388
Joined: Sun Sep 23, 2012 8:09 pm
Contact: Website

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Sat Apr 17, 2021 8:33 pm

If hexadecimal is allowed the top of the range model will be the RPFFFF sporting 15 cores, a Cortex-M15, 512MB of RAM and 512MB of flash. Other interesting variants would be the RPBABE, RPBEEF and RPFACE.

hippy
Posts: 9650
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Sat Apr 17, 2021 8:56 pm

Why limit to four hexadecimal digits ?

They can add "G to Z" on top. To paraphrase what someone allegedly once said; no one will ever need more than an RPZZZZ.

And they could add a "+" suffix, or more than one, if they do :twisted:

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 28411
Joined: Sat Jul 30, 2011 7:41 pm

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Mon Apr 19, 2021 8:12 pm

hippy wrote:
Fri Apr 16, 2021 1:29 pm
It depends how one looks at it; if an RPxxxx can do what one was going to do with a Zero then any differences have pretty much disappeared.
In that case, the Zero was being massively underutilised. Which is fine, if you optimise for cost, because the Zero is cheap. However, its nowhere near as cheap as the Pico. So, yes, I would expect the Pico to take some Zero sales, but not many.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Working in the Application's Team.

hippy
Posts: 9650
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: RP2040 + Future RPXXXX (RP4340?) Naming

Tue Apr 20, 2021 3:35 pm

In the maker arena I would expect many Zeroes and other Pi's are under utilised, are a cheap means to an end with a good supporting eco-system making it a cost effective solution even when over-kill. It might be that most Zeroes are under utilised with people choosing other Pi if they want performance or capabilities, a general purpose computer rather than an application platform.

Last year's big demand on Zeroes appeared to be for ventilator information display which should have rather modest requirements and could have perhaps been Pico-based instead if it had existed.

The Pico, though much more than many traditional micros, is constrained by speed but more so RAM which will probably keep it from replacing Zeroes for now except for those under-utilising applications. As future RPxxxx emerge, carrier boards sport more real world interfaces, the potential for doing so increases. I would expect it to be the amount of RAM which will be the tipping point.

Return to “General”