Keeping secrets and writing about Raspberry silicon
In the latest issue of The MagPi Magazine, Alasdair Allan shares the secrets he had to keep while working behind the scenes to get Raspberry Pi’s RP2040 chip out into the world.
There is a new thing in the world, and I had a ringside seat for its creation.
For me, it started just over a year ago with a phone call from Eben Upton. One week later I was sitting in a meeting room at Raspberry Pi Towers in Cambridge, my head tilted to one side while Eben scribbled on a whiteboard and waved his hands around.
Eben had just told me that Raspberry Pi was designing its own silicon, and he was talking about the chip that would eventually be known as RP2040. Eben started out by drawing the bus fabric, which isn’t where you normally start when you talk about a new chip, but it turned out RP2040 was a rather unusual chip.
I get bored easily. I started my career doing research into the high-energy physics of collision shocks in the accretion discs surrounding white dwarf stars, but I gradually drifted sideways into playing with the toys.
After spending some time working with agent-based systems to solve scheduling problems for robotic telescopes, I became interested in machine learning and what later became known as ‘big data’.
From there, I spent time investigating the ‘data exhaust’ and data living outside the cloud in embedded and distributed devices, and as a consequence did a lot of work on mobile systems. Which led me to do some of the thinking, and work, on what’s now known as the Internet of Things. Which meant I had recently spent a lot of time writing and talking about embedded hardware.
Eben was looking for someone to make sure the documentation around Raspberry Pi Pico, and RP2040 silicon itself, was going to measure up. I took the job.
I had spent the previous six months benchmarking Machine Learning (ML) inferencing on embedded hardware, and a lot of time writing and talking about the trendy new world of Tiny ML.
The rumours of what I was going to be doing for Raspberry Pi started flying on social media almost immediately. The somewhat pervasive idea that I was there to help support putting a Coral Edge TPU onto Raspberry Pi 5 was a particularly good wheeze.
Instead, I was going to spend the next year metaphorically locked in a room building a documentation toolchain around – and of course writing about – a totally secret product.
I couldn’t talk about it in public, and I talk about things in public a lot. Only the fact that almost everyone else spent the next year locked indoors as well kept too many questions from being asked. I didn’t have to tell conference organisers that I couldn’t talk about what I was doing, because there weren’t any conferences to organise.
I’m rather pleased with what I’ve done with my first year at Raspberry Pi, and of course with how my work on RP2040 and Raspberry Pi Pico turned out.
Much like a Raspberry Pi is an accessible computer that gives you everything you need to learn to write a program, RP2040 is an accessible chip with everything you need to learn to build a product. It’s going to bring a big change to the microcontroller market, and I’m really rather pleased I got a ringside seat to its creation.
You did a great job. It is often overlooked that good, human readable documentation is far more important than all those MHz, SRAMs, and buses. Without it, you can blink an LED with the sample code, but you won’t get much further.
The docs for Pico are incredibly thorough and well thought out. Thanks for that.
A fantastic job enabling me to put together a project for learning micro controllers to electronics students and old hands alike.
Good documentation is the key to keep all knowledge flowing.
An interesting thought occurs. Correct me if I am wrong, but wasn’t it said by Eben Upton that the Pi 4 is the last generation to be based on the line of arm CPUs in use in Pis? Is the Pico a step towards creating the next Pi CPU? Probably not, but still an interesting thought.
I’ve not heard that, and I’ve been following things pretty closely. It also doesn’t chime with everything else he, and others at Raspberry Pi, *have* been saying.
So it’s like apple silicon but tastes better that’s all I need to know
When with the RP2040 IC’s be released for users to try and make their own boards? Obviously the big companies have them already, but would enjoy the challenge of designing something for myself.
Raspberry Pi Staff Alasdair Allan — post author
The current timeline has RP2040 available to buy sometime in Q2 this year. In the meantime you can take a look at the “Hardware design with RP2040” book for inspiration.
I was wondering why didn’t they put the effort in Pi_Silicon based on RISC-V ISA instead of ARM ISA ? This way the tinkerers can play all the way into silicon & compilers/optimizers instead of staying till datasheets.
will there be a further version of raspberry pi pico of wifi?
Documentation is excellent (as we have come to expect) but why in the name of all that’s holy was someone allowed to use GREY text in the pdf’s. Please consider releasing a second edition with BLACK text, so that I don’t feel like my eyeballs have been sandpapered after 15mins reading.
Raspberry Pi Staff Alasdair Allan — post author
This has been discussed internally and on Github, https://github.com/raspberrypi/pico-feedback/issues/13#issuecomment-776964889, and the next documentation release — expected this week — will have updated fonts.
You didn’t explain one thing in your article – WHY you had to keep this secret. I mean.. why is foundation so closed about their development? Why do they keep it secret? This is something I really don’t understand.
Because it’s a non-profit that makes a ton of money, that don’t want to lose market share to a company which gets wind of what they’re doing and beats them to market with a competitive product.