Podcast
Join Viktor, a proud nerd and seasoned entrepreneur, whose academic journey at Santa Clara University in Silicon Valley sparked a career marked by innovation and foresight. From his college days, Viktor embarked on an entrepreneurial path, beginning with YippieMove, a groundbreaking email migration service, and continuing with a series of bootstrapped ventures.
Links
Follow Me
Follow Me
Join Viktor, a proud nerd and seasoned entrepreneur, whose academic journey at Santa Clara University in Silicon Valley sparked a career marked by innovation and foresight. From his college days, Viktor embarked on an entrepreneurial path, beginning with YippieMove, a groundbreaking email migration service, and continuing with a series of bootstrapped ventures.
Transforming Tech with Eben Upton: Exploring Raspberry Pi’s Global Impact from Education to Industry
Play On
02 JUN • 2024
1 hour 21 mins
Share:
In this engaging episode of “Nerding Out with Viktor,” the host, Viktor Petersson, sits down with Eben Upton, the visionary founder of the Raspberry Pi Foundation, to dive into the fascinating world of affordable, programmable home computers and their far-reaching impact on education, industry, and technology enthusiasts. The conversation is a treasure trove of insights, anecdotes, and forward-thinking ideas that will captivate listeners with a medium to high level of technical knowledge.
Eben Upton’s passion for computing and his early interest in home computers during the 1980s are expertly woven into the narrative by Viktor Petersson. This sets the stage for Eben’s remarkable journey as he shares the story behind establishing the Raspberry Pi Foundation, with a clear goal to reignite interest in computer science among young people. The duo takes listeners on an epic ride from the 1990s when a noticeable decline in students applying to study computer science at Cambridge University sparked the idea of creating an affordable, programmable home computer.
The launch of Raspberry Pi in 2012 marked a significant milestone, selling an astonishing 100,000 units on its first day. What’s remarkable is how quickly it spread across various sectors, from enthusiasts to educational and industrial applications. Eben highlights Screenly as one such company that utilized Raspberry Pi for digital signage, demonstrating the device’s potential to reduce costs and improve efficiency in many businesses.
As the conversation progresses, Viktor and Eben delve into the challenges faced during the COVID-19 pandemic, particularly the supply chain crisis that affected the availability of semiconductors. The tough decisions made by the Raspberry Pi Foundation to prioritize OEM customers and maintain a steady supply of units are a testament to their commitment to supporting businesses in need. Despite these hurdles, they continued engineering efforts, culminating in the release of Raspberry Pi 5.
Eben also shares the strategic decision to manufacture Raspberry Pi in the UK, which offers benefits such as better communication with the factory and the ability to implement design changes quickly. While most components are sourced globally, the final assembly and some parts, like injection-molded plastics, are produced locally. This approach has allowed the foundation to maintain control over production and ensure high-quality standards.
The conversation then explores the potential for Raspberry Pi to adopt RISC-V, an open-source hardware instruction set architecture. Eben discusses the advantages and challenges of making this shift, including the significant engineering effort required to develop a high-performance, open-source core. While acknowledging the benefits, he hints that Raspberry Pi 6 will likely continue using ARM cores due to their maturity and ecosystem support.
The episode also covers the evolution of the Raspberry Pi software stack, with Eben sharing the shift towards open standards like adopting VDPAU for Linux and moving towards a Wayland-based graphics interface. These changes aim to improve performance and compatibility, particularly for applications involving multimedia and digital signage.
One of the most relatable topics discussed is the reliability issues with SD cards, a common pain point for Raspberry Pi users. Eben explains that while there are many subpar SD cards on the market, high-quality cards like those from SanDisk have proven reliable under rigorous testing. The potential for future Raspberry Pi models to include soldered-down eMMC storage, providing a more robust alternative to SD cards, is an exciting development.
As the conversation wraps up, Viktor and Eben emphasize the importance of mentoring and supporting young people interested in technology. Eben shares a personal story about how mentors played a crucial role in his early development as a programmer, highlighting the value of guidance and encouragement in one’s career.
Throughout this engaging episode, listeners are treated to a comprehensive look at the past, present, and future of Raspberry Pi, showcasing its impact on education, industry, and the broader tech community. With Eben Upton’s expertise and passion shining through, this conversation is a must-listen for anyone interested in technology, innovation, and the inspiring stories behind them.
Transcript
Show/Hide Transcript
Welcome to another episode of nerding out with Victor.
Today, we're thrilled to have a special guest joining us, Eben Upton, the visionary founder of the Raspberry PI Foundation.
Eben and I go way back.
I think we first connected in person at a Raspberry pivant in Cambridge.
After exchanging numerous emails in this episode, we'll dive deep into everything Raspberry PI, so get ready for the fascinating insights and stories from even.
It's fantastic to have you on the show.
Welcome so much.
Well, thank you for having me.
Well, so maybe we can start with just people not familiar with you and the PI foundation just kind of give, I guess, the origin story of how things started.
Yeah, it's an increasingly.
It's a scarily long story now.
And I've been telling.
Yeah, I've been telling.
Obviously, I've been telling the story for most of the history of Raspberry PI.
I was, let's see, I'm a software engineer and I grew up in the 1980s in the UK.
And like a lot of people who grew up in the 1980s in the UK, I had a home computer and I became a software engineer by accident.
And I enjoyed it.
I very much enjoyed it.
I've always been interested in the business.
I think I've been as interested, I suppose, in the business of technology as in technology itself.
I started a games company when I was an undergraduate here in Cambridge.
And so ive always been kind of interested in being a software engineer, being interested in games and graphics, been interested in business.
But I got involved in teaching at the university quite a long time ago now, nearly 20 years ago, and really was struck by the disappearance, I guess, of young over that maybe that ten year period from 1990, 519 94, 95 when I applied to Cambridge in 96 when I came up and then ten years later that sort of declined disappearance of people who had that kind of accidental background.
Yeah, still people who are interested.
We never let anyone into Cambridge who wasn't super bright and super interested in computing.
But I think we got to a point by maybe 2008 where were letting in every single applicant who was bright and interested in computers.
And that's not a great place to, you know, that's not a great place for a university like, I mean, you know, Cambridge is, it's Turing's university, right?
You know, if any university shouldn't be struggling to recruit undergraduates, it should be this one.
Right?
And so I was, I, you know, I was struck by that.
And really Raspberry PI was my response, the response of several of us, including wife Liz, our response to this realization that this thing that we really loved, this thing that had been central to our lives, just wasn't there anymore.
The programmable home computer.
We launched Raspberry PI in 2012.
We sold 100,000 Raspberry PI's on the first day.
We launched it on the 29 February, which has constrained, we always say it constrains our birthday party choices.
We've had three really cracking birthday parties now.
And I sold a lot of raspberry pies very quickly to enthusiasts originally, but they did then make their way into education via enthusiasts.
A lot of those kind of enthusiasts volunteer or teach or simply a parent.
And then I guess, as you discovered, the platform is also useful for industrial applications, for embedded industrial applications.
And that's kind of the journey, I guess, we've been on with raspberry PI, which is increasingly trying to service those embedded industrial applications while staying true, I guess, to the.
To the original mission, which is around enthusiasts and education.
Right.
Yeah, I remember.
So were, I guess this must have been 2012.
Yeah.
When were doing accident, were doing signage accidentally, and we discovered the first raspberry PI and I think it shaved our bomb by a 10th.
Like were down to.
Yeah, and that's the story, you know, the surprising story for raspberry PI in embedded in particular is that there were lots of companies like yours who were, I guess, to the extent they were making hardware, either they weren't able to start because they had no platform or they had started and they were running with some platform they built themselves.
And so you had these kind of reluctant hardware companies.
They were only hardware companies because they needed a venue where they're differentiated with software.
And really, Raspberry PI, I guess, has grown in that space by allowing people either never to start making hardware or to allow people who are making hardware and weren't enjoying it to stop making hardware, at least to stop making the middle bit of the hardware, the compute complex in the hardware.
Yeah.
Because I remember we did look at, I think, was it Nvidia had one of the chipsets you could build, but even if you were like a small company, like the moc's were like, moqs were completely unrealistic unless you were like a multinational, essentially.
It was a complete game changer back in the days, right?
Yeah, that's it.
I mean, if you're building, even today, if you're building, obviously there comes a point, there comes a scale point where a lot of our thinking at the moment, around how we evolve the business is how we develop an offering for people who are at those higher volume points where they start to, where it starts to become realistic to leave the raspberry.
We talk about people graduating and, you know, we've relaxed.
We don't need people.
As long as we have a steady flow in the industrial space, as long as we have a steady flow of new people coming through with new designs, it doesn't necessarily matter to us that some people, yes, it's a source of pride for us, a source of joy, that we bootstrap people to a point where they can depart our ecosystem.
But I think increasingly, you do get these people, once they get out towards maybe 100,000, if you're making 50,000 units a year of something, it's almost never the right thing to do.
To build your own compute subsystem.
You should always outsource.
And to the extent people do outsource, we obviously we think that we're a good candidate for that.
But a lot of the work we do now is about how when we get those people to get into that 50 to 100,000 unit stage where they make that second make versus buy calculation, we win the first make versus buy calculation.
It's how do we keep people with us?
What services?
What products?
What services?
Sometimes customized products we can offer people at that volume scale which give them a reason to stay with us still, not to the million unit range, but out into that four or 500,000 units range.
Yeah.
And let's take a step back to those early days.
Right.
Because I think were, I mean, we have a great amount of success thanks to the PI foundation, largely at screenly.
And I think, if I'm not mistaken, I think were probably one of the first commercial projects ever built out of that.
And it was amazing to see in the early days, the pickup from the forum and the community, that was, I think for the first two or three years, that was our entire growth channel.
Yeah.
And that sort of feeling that there is both a, there's both a source of demand there, you know, a source of, but also a source of supply of knowledge.
But, you know, one of the advantages raspberry PI has lucked into, I guess, you know, is, you know, if you are, if you're selling one or two orders of magnitude more product, damn, the next largest product in your class, all of the, if you have a problem with, if a customer has a problem with the platform, generally somebody else has had the same problem in the past, and either they figured out how to fix it or we figured out how to fix it for them.
And so you've kind of got that kind of virtuous, you've got that virtuous cycle.
Absolutely.
Absolutely.
And I think that was a big part in their ladies.
And I think thats definitely true that a lot of bug fixes were found on the forum and whatnot.
But also I think in the early days, I think a lot of people like myself were on the waitlist and they finally got this device and they were like, now what?
They were looking for use case.
Yeah, there was a lot of those first hundred thousand units.
Theres always been a kind of, theres been this meme, I suppose, knocking around that Raspberry PI's always end up in drawers.
And I kind of have this kind of like, there is this wonderful alternate, we sold 60 million RaspBerry PIs.
Now those draws are going to be pretty full by now.
So while it isn't true that Raspberry PI's end up in drawers, I think it is the case that a lot of people do buy them.
And then, as you say, think, what do I, this is obviously a cool thing.
This is a thing I have to have in my life, but why?
What do I do?
And you see people turn up on Reddit.
So I've got a Raspberry PI.
Will you give me a project?
So that's another thing where you have this huge supply.
Now, obviously, I think even the strap line for the subreddit says something like not just Cody and magic mirrors, because Cody and magic mirrors are the kind of the UR Projects, the first things that people get pointed to when they pie out of the box.
Absolutely.
Absolutely.
What's the strangest or most surprising use case you've seen so far?
Well, I mean, look, to some extent the whole enthusiast, I mean, the whole thing is surprising.
I think that's the first thing to emphasize.
The whole thing is surprising because this is a, remember, we're trying to get, you got 200 applicants to computer science, so you're trying to get 300 applicants to computer science.
So you're thinking, well, 1000 units in the hands of the right thousand kids, you get a ten to one success ratio, a ten to one success rate, you get the extra hundred kids.
Right.
So anything done with a raspberry PI that isn't some very structured, Cambridge centric, possibly open day centric farming of these things out to people still in the back of my mind seems strange and unnatural to me in terms of the sort of things that people do.
I think there's a whole strand of maker pro business, this idea that people buy raspberry PI's to build their own personal passion projects, and then they go on social media and they talk about their passion project, and suddenly they realize that there's a market for this.
So I think that kind of, we talk about ESG, talk about what's the social good that raspberry PI does?
And, of course, the foundation is a huge part of that.
We've returned about $50 million to the thing I run is Raspberry PI Limited, the trading company.
The thing does the engineering and sells the product.
The Raspberry PI foundation is my shareholder, and so I give money to the foundation to do its work.
We've returned $50 million to the foundation.
So there's a huge amount of goods, ESG, social benefit from the work they do.
And of course, we also think that theres a social benefit when we sell a raspberry PI to a child, theres a social benefit there as well.
But I actually really believe that social benefit of helping people start companies, whether theyre obviously helping to do some work to help enable you guys, or if its just somebody who makes a personal passion project, sticks it on Twitter, and then goes and sells a couple hundred units, uses Kickstarter to fund a couple hundred units for people.
You know, those are good.
You know, they are, they're all life changing experiences, right?
They all, they all change.
They can change the vector of somebody's lifestyle.
I do love those.
I mean, there were the individual kind of funny ones.
I mean, recently, obviously, a lot of LLM, I saw a wonderful llama three, the 8 billion coefficient llama three running at about two tokens per second on a PI five.
So I think we are, what's nice is you saw, as soon as the llama stuff leaked last year, you saw people running original llama on a PI four, and there you were getting a token every seven or 8 seconds.
And so what you've seen is you've seen this kind of big jump in performance, say three x or more jump in performance between PI four and PI five.
And that's plugged into this inevitable trend that always happens to models or to classes of model, where people discover that you can make more heavily quantized versions of models, you make more heavily quantized versions of models, you can make more sparser models, smaller models, sparse models, more heavily quantized models, and get really pretty good performance.
So those two things put together to a point where you can now on a Raspberry PI five have a plausible conversation with.
With llama LLM.
So I like that.
That way that Raspberry PI has always tracked what's cool.
You know, there was a huge amount of.
When Nintendo were making those little mini Nes's, there was a huge amount of retro gaming happened on Raspberry PI when it's LLMs, as when it was CNN's, when it was imagepress, when AI.
Remember two years ago, AI meant open.
Source cv was the big thing.
Yeah, right.
That was AI.
It changes every year.
People always talk to me like, you should put an AI accelerator into a raspberry PI.
Like, yeah, but which AI?
I mean, it takes five years to go from chip concept to shipping raspberry PI.
The 20 712, which is in PI 20 712 was in flight when Covid got loose.
It was in development when Covid got loose.
Right.
So, yeah, these are hugely long projects, and that's really only, you think the transformer paper is from the original transformer papers from 2017 or 2018.
So in practice, transformers were maybe a year old at the point where we started that chip.
So there is always that change.
But you say you sort of see.
You know, you see raspberry PI was tracking what's cool, and that's even today.
And that's.
That's.
That is wonderful.
Yeah.
Yeah, that's definitely true.
There's always a pie angle.
And if you write a blog post or you do anything, if you.
If you put in PI in there somewhere, that always drives a bit more traction.
Like a little hacker news.
Yeah, right.
I.
Cool.
So we talked about surprising use cases.
Let's go back to.
Well, you mentioned Covid.
There was obviously a massive supply chain crisis that happened to everybody in around that era.
Talk to me a bit about that, because that was obviously something that affected both of us pretty severely.
Yeah, I think it's a good.
I think it's good to emphasize it happened to everybody.
Occasionally you would meet somebody who seemed to believe that it only happened to raspberry PI.
I remember.
I remember sitting there reading some of these comments online on a day where I had got.
We were re roofing the little barn, large shed at the back of my yard.
And the only way being able to get wood battens for them to build the roof structure, well, to build the tiles would mount onto, was to go down to the roofing yard at 06:00 a.m.
and buy these from a junior employee.
Buy these via a junior employee because he knew that as soon as his supervisor turned up, his supervisor would say, no, you can't buy those roof patterns because it's also a roofing company, and we need those for our roofing stuff.
You literally could not buy wood, and wood grows on trees.
This doesn't grow on trees.
Or does.
It was.
This was an omnicrisis.
And unfortunately, of course, semiconductors were particularly affected.
This world where these pictures of huge car parks at Volkswagen, where they had the one entry management computer, and they put it into a car, you build a car, put it into the car, drive it into the parking lot, take it out of the car, bring it back in, put it in the next car, drive that car into the parking lot.
You have these vast supplies of cars waiting for engine management systems.
So, yeah, it hit us, I guess, for almost exactly 24 months, from March of 21 through to April of 23.
Those are the bad months for chip supply, I guess.
And in terms of the peak impact on end users was sort of in the second half of 2022, early 23, I guess it really was the kind of darkest before dawn.
We did a million, if you think within a good year, we'll do seven or 8 million units, we did a million units in the first four months of last year.
So annualized rate of 3 million units.
So it really was done.
And then suddenly April, May, last year, it's just turned on like a chip supply turned on in April, so finished goods supply turned on in May, and it was like a light switch going on, but it was brutal.
And of course, we had to make difficult allocation decisions.
We have oems, like yourselves, who have invested in the platform and who've made a choice.
You're not a huge company.
When people talk about.
When we talk about having had to prioritize OEM customers, people think we're talking about IBM or someone.
We do have some genuinely huge OEM customers, hundreds of thousands of units a year.
The majority of customers are a little bit smaller than that, don't have infinite agility to.
Because often their solutions, the PI, has a particular set of capabilities, and your solution is very entwined with the capability of our hardware.
You don't have infinite agility to swap our hardware out for some other hardware.
We had to make those difficult decisions, to prioritize oems, to work more closely with oems, to understand demand, because, of course, everyone with the best will in the world, everyone is incentivized to hoard in a difficult situation.
So toilet paper in 2020, right?
Everyone's incentivized to hoard, but most people understand that, you know, that if you can win their trust, when someone comes to you and says, I need 10,000 raspberry pies, and you say, well, how many do you really need today?
Will 50 do today and 50 next week?
What were able to do, we built the sales organization in anticipation of selling a lot of raspberry PI from 2020 onwards.
And actually what we ended up doing was using that organization in a kind of anti sales mode to actually intensively manage customer relationships, win OEM Trust, and then use that to put people on a kind of a drip feed, really.
There was that dream that if we got a chip, because were throttled by upstream silicon supply, if we got a chip, then we wanted to get it onto one of our products, into a customer's hands, an OEM customer's hands, into an OEM customer product, and into their end customers hands as quickly as possible.
So what you're trying to do is you're trying to discover all of the pools.
The way we got through this was went out and we tried to discover where all of the pools were, not just upstream components for us or downstream finished goods, but us as an upstream component for our customers and their downstream finished goods, and said, let's just take all of those buffers, let's just lean them out, because there's one to 2 million units probably in all those buffers.
Let's just lean those buffers out.
And if you think we probably had a deficiency of 3 million units, maybe across the crisis, insofar as were able to find one to 2 million units through that kind of activity, it really did help.
It helped mitigate the impact for everyone.
And I think definitely that the sales organization has improved a lot, at least from my experience, has become a lot better as well.
Our relationship has definitely improved a lot as well.
It was a bit more chaotic before, and now it's more structured.
So that's a good thing on our end.
I mean, they're an amazing team, and generally mostly very young people in that team.
It's probably the youngest average, and we do have a few pretty senior people, but they're mostly very young, incredibly bright people, really.
I think we've, you know, we always think we're pretty good at finding engineering talent here, but I think that what Mike has been able to do, he's really found this very rich seam of very intelligent, very capable, very motivated young people to staff that team.
Yeah, but that decision to prioritize Oems were definitely a controversial, like, the Internet blew up.
Right.
A lot of people were screaming, but that's.
That was.
I mean, from my vantage point, obviously I'm biased, but I think that was.
That was you.
Like, would you like me to have done something different?
No, I think that was the right call from both from a commercial perspective and from like.
Yeah, I think it was the right call from a moral perspective.
I think.
Look, it was horrible.
Yeah.
but as I say, people think you're talking about IBM and you're not.
You know, you guys are big and you guys are very big by the standards of our OEM customer base.
You know, we have a lot of people who just need 50 raspberry PI's a year.
Yeah.
And they put them into some product which sells for a few thousand dollars and that gets them a few hundred thousand dollars of revenue, and that pays to keep the lights on, the mortgage paid.
And if they don't get the 50 units, they go bust.
Right.
Yeah.
And it was really hard for me to say, to turn people like that away.
Yeah.
And I think that is correct in the sense that if you can't deploy your cody at home, nothing much is going to change.
Whereas there are people's livelihoods depending on this.
In a lot of these isvs.
There was an observation that many of.
And this is why I think, yeah, I think the vast majority of our fan base, our enthusiast base, did understand the difficulty of the decision and the inherently short term nature of the problem.
Right.
And so people were prepared to dig into their existing stocks because that's the other, you know, in the enthusiast side, that's the other big pool of inventory.
Right.
We talk about them going in drawers.
They don't go to drawers, but many of them go into projects.
One of the things that drove the surprising sales rate of Raspberry PI, I think, was people putting Raspberry PI so cost effective that you could put it into a project and then when you were finished, you'd leave it in the project to buy another pie rather than scavenging it out of the project.
And that did mean that there was a lot of inventory.
And so there was quite a thing I observed quite a lot on Reddit was people talking about their attrition rate.
So people talking about the, you know, where we got.
We did our best to communicate to people and give them an idea of.
And I think it did go on for six to nine months more than I think when we first communicated to people.
We gave people an idea over in Mid 22, and it was, in the end, over in the second quarter of 23.
But you saw people talk, looking at what were saying and looking at the rate at which they.
The number of Raspberry PI's they owned and the rate at which they were destroying raspberry PI.
Bye.
Sticking 12 volts on them and saying, look, this is the end of the.
This is the end.
This is what the end of the road looks like for me.
You know, I think I will exhaust my stash in 15 months time.
Yeah.
Please be back in stock in 15 months time.
And so, look, it was a horrible situation.
It is a.
It is.
It is not a.
Yeah, it'll never be something that.
It's one of those things where you make the least bad decision, that they're all bad decisions.
And you try.
That's the right lens.
You try to make the least bad decision and you can sleep at night because you think, well, actually, is.
Was there a better thing I could have done today?
But it's a.
It was a once in a.
Once in a lifetime, probably once in an industry.
Yeah.
There has never been anything.
There been cycles.
I remember when you couldn't buy MLCC's 2017, you couldn't buy caps.
The cap vendors were swagger into your building.
Do you want to buy some of my precious 100 n caps?
So they were always cycles.
But I think that this is a super cycle.
This one is as bad as it's ever been since the invention of solid state electronics.
So that's to have come through it at all for us and for our OEM customers and for our enthusiasts and for the educational mission, you know, survival is victory, right?
Yeah, we survived and we got raspberry PI five out at the end of it.
That was the one thing I am proud is not a lot to be.
You know, there's a lot to be to not feel unproud of.
The thing I feel, actually feel proud of is that the team here kept doing engineering all the way through the painful.
In parallel with doing availability engineering where you're.
Oh, it's not just the big silly.
Occasionally a ten cent thing would go out of stock and you have to work around it.
So in parallel with doing the availability engineering, they did the big stuff as well.
So we're six months out of the.
In September last year, we're kind of 6456 months out of the kind of most brutal shortage.
And we launched PI five, which was amazing.
Right.
And it was just the best day.
Yeah, absolutely.
And let's go back to the supply chain a little bit.
So I believe it was the PI four was somewhat reshored and made in the UK, brought back to the UK.
Talk about, I guess the decision and the thinking around that.
Yeah, we re short.
We really unusual.
You have a lot of companies that start in the offshore as they go to volume.
And actually, I think thats modestly insane.
Generally what happens at least with a low to medium touch product like raspberry PI in terms of amount of human fettling required to get it out the door.
Once you reach a certain volume, you can justify massive investments in automation.
Once it's a fully automated process, it doesn't matter where the robots are.
They can be at home or they can as long as the electricity is fairly cheap, which is not always nailed on in Britain, but as long as the electricity is fairly cheap, then it doesn't really matter where the robots are.
We started reshoring in 2012, quite a labor intensive process at that point, but with a little bit of automation to South Wales and then by 2018, 2019, almost entirely re shored.
Obviously hugely valuable for us during COVID because if you lose access to your factory, what are you going to do?
You can't go to your factory for three years.
We have to stay in a quarantine, engineers have to stay in a quarantine hotel for a month in order to go visit the factory, and then a quarantine hotel for a month when they get back.
Yeah.
So that was, that was very valuable to us.
The big benefit is the design for manufacturability feedback that you get by building locally when the factory is 4 hours away in a car and they speak the same language as you.
It's a huge difference versus a twelve hour plane flight in a translator.
So.
And that's one of the reasons why so cost effective.
Right.
You know, it's a rose price, a low priced product.
And that low price is underpinned by low cost, by a low cost structure.
And that low cost structure is underpinned by going into the factory every month and saying, hey guys, what do you hate about our product?
You know, what's, what sucks about making our product, what's taking time, what's leading to rework, what's leading to scrap, what can we change about our product to remove those things?
And we continue to do it.
And there's some real surprises.
We continue to find opportunities to do that even.
We found a packaging, we found a packaging change that we haven't deployed yet.
Roger Thornton, who runs our application engineering team and compliance team found a packaging change that will save us $300,000 a year.
Wow.
Five cents a unit, something like that, on most of our products just in the last few months.
Right?
Yeah.
So, yeah, we continue to find these.
These efficiencies, and being able to have a good conversation with your manufacturing partner is a really important part of that.
But so all the components, how much of the total bomb is made in the UK?
Because I would imagine the assembly is in the UK, but in some of the PCB manufacturing.
But there are still a lot of components that are essentially every.
Every single component is made offshore.
Yeah.
So cardboard box is probably made in the UK.
PI 400.
The plastics are made in the UK.
So the plastic bases are made in.
We do a lot of injection molding here.
It turns out, again, a little bit like electronic light industry is the thing that Britain's really good at.
And so we kind of.
We do our injection molding in the UK.
We do our PCB manufacturer, we do our PCB assembly.
PCB's themselves, the actual substrate is manufactured in China.
Most of the semiconductors are either manufactured in Taiwan for logic, or either Korea, South Korea or North America for DRAM.
Electromagnet is mostly.
So connectors.
PCB mostly.
Mostly China, mostly PRC.
Power supplies of Cambodia, interestingly.
So KTEK build our absolutely massive facility in the middle of nowhere in Cambodia.
So, yeah, it's quite a.
Although final assembly is.
Final assembly is UK.
There isn't the upstream supply chain here to sustain much more than carbon box.
Manufacturing, because I think that's what people don't really understand about electronics.
Well, a lot of people outside the industry, I guess, in electronics manufacturing, you can't make these components in Europe or in the UK particularly.
It's just impossible.
There isn't a possibility to do that.
Yeah, semiconductors are particularly challenging.
We do semiconductors, so we, of course, we make our own chips now.
Right.
So that's the big.
That's the big.
The big change in the last few years is we've gone from being purely a consumer of other people's silicon.
Silicon that we've been deeply involved in, where we've been deeply involved in design in one way or another, to being a silicon manufacturer ourselves.
And that's fun.
So we fab those with TSMC in Hinshu in Taiwan.
We package them either in Taiwan or some of our lower cost products of the microcontroller products.
We actually packaged in EPO in Peninsula, Malaysia, which is great for me, actually, because Liz's family are from very near EPO.
And yeah, that's kind of fun.
So I haven't actually been to the factory, I haven't been to the packaging, the EPO packaging plant.
So that's on my list, I think, for things I would like to go and do fairly soon.
Very nice.
All right, let's talk a bit about the PI five and kind of the future roadmap.
One thing that I very much welcome from the moved, I guess direction of the python as you're moving is more towards open standards.
So adopting video for Linux, for instance, for the graphics interface is a big, I mean, that makes a huge difference in terms of people like ourselves and moving towards open standards.
Maybe speak a bit about the philosophy and the direction of adopting more open standards, I guess.
Has Gordon shown you the Weyland, the latest set of Weyland demos?
No, I don't think so.
The kind of ticker, we have a signage focused Weyland demo, which is kind of video and a ticker and a 2d ticker at 60 FPS, just kind of a little bit.
I mean, you're familiar with the history of this, right?
Yeah, broadly speaking, the history of multimedia in particular on raspberry PI is that we used to use a bunch of broadcom proprietary interfaces for driving the multimedia subsystem.
And so that meant video decode, camera process, video encode and decode camera processing three d and display composition.
And broadly those APIs were open Max.
I mean, okay, some of these are standards based.
So open Max, for example, is standards based.
So open Max or MaML, which is kind of a simplified open max for video codec and camera, I think, or dismantling for display composition and OpenGL.
But a kind of a remoted OpenGL.
The interesting things about OpenGL is that it is very amenable to being effectively serialized across a FIFO.
What we would do is we'd run the OpenGL subsystem state machine inside the VPU.
So the closed source blob, this thing goes called the Blob.
Sometimes on a pie I could talk at length about fear of the blob, and that I think often people who are afraid of the blob imagine that their PC is a lot less blob full than it actually is.
You do have a management engine with the blob on your PC, you know, and you don't know what's the source code of that.
BiOS I had an episode on, I had an episode with coreboot guys, for instance, talk about BIOS tests, just a few episodes ago.
And yeah, I guess that's a good example of blob that you have no idea what to do.
How can, how can I be safe on my x 86 PC when there is this piece of closed source code that I don't understand?
Anyway, sorry, there is a.
Anyway, there is a blob and this is a single star health which is a big VPU program.
A big closed source proprietary VPU program which provides the device side element of all of these APIs which are basically serialized and piped over a subsystem called Vchi or VChiq from Linux host land, from a bunch of Linux host libraries through the kernel across to the.
So everything in a PI one in 2012 is kind of done by remote control.
HDMI mode negotiation.
Yeah, I mean it makes for a very nice platform actually, particularly for people who want to do bare metal programming.
Is that the thing comes up with a frame buffer, right.
You know, by the time the arm is out of reset it's already negotiated a display mode for the television, which is great in a lot of ways.
When you've got a little processor, you've got an arm eleven in there.
It's quite nice not to have to worry about that.
Tv service and DismantX and some of these other APIs provide you a way of renegotiating if you want to renegotiate.
And kind of all is well, but all is very proprietary, non standard.
And so the direction of travel really has been, as you say, towards a standards based approach.
So rather than so probably trying it in the right order, probably the earliest of these is the move to Mesa to drive the 3d.
Emma Anholt joined Broadcom, worked for me at Broadcom for a while and did some of the early work on this Igalia.
Now to maintain the stack, maintain and progress the OpenGL and Vulkan metastack for us, which is kind of fun.
And then you have moving to KMS World for controlling display output, as you say, video for Linux World Force or stateless v for lithe world for controlling codecs.
Even the legacy Codex HG 64 modern codecs.
So from PI four onwards there's a HEVC codec which is a modern codec developed at Raspberry PI that was designed to be driven from the get go.
So you have this kind of gradual depletion of the stock of things that are inside the blob, that are controlled from inside the blob to the point where on the Raspberry PI five.
Really, it's only clocks powered reset and everyone needs all sensible.
I believe a bunch of intel chips actually have Arm cores in them now that do microcontroller calls in that do this stuff.
So everyone has large chips don't tend to come out of reset.
They don't tend to have the big cause be the first thing that comes out of reset.
Almost always some.
I've done a bunch of chip architecture in my career.
Almost always you have some management processor of some sort that comes out of reset first and gets the world ready.
Sort of John the Baptist.
John the Baptist processor that doesn't come to such a sticky end.
But maybe set up the dram.
Maybe they'll set up the dram.
It'll fetch a next stage bootloader.
It probably won't negotiate an HDMI mode for you.
That's probably a bit excessive.
So really, that's what the VPU, the closed processor has been beaten down to almost.
It's still present.
It's a very antique piece of hardware.
It's still present in 20 712 in the Raspberry PI five chipset.
But it's been relegated to this micro control type world where it sets up dram loads, a second stage bootloader, and then sits there providing clocking services, monitoring the temperature of the chip, and reducing clocks if things get too hot.
Those kind of some of the slightly safety critical processes that you don't necessarily want to.
You want those processes to continue to happen even if the OS crashes, even if the big OS goes away.
It's kind of a watchdog world.
Yeah.
It's funny you mentioned the frame buffer, because we are only just now in the process of migrating our entire space to Wayland.
We've been using frame buffer well, since.
Inception, really, I think have used DismantX in the past.
Right.
As the way first class way to get to drive a composition hierarchy until really quite recently, was to use Disman X.
Certainly until 2019 was dismantling the demo I mentioned.
As I say, one thing we do more of now is we have more software resource is to provide reference examples for if you think that your world often consists of some video content, increasingly, of course, I know your world consists, a lot of it consists of HTML five.
So a lot of we do is about making the browser run well.
But to the extent that you have custom players, often you're putting together some 2d content, some 3d content and some video content into a single window.
And where historically we would have made a Dismantx example to show you how to do that.
The example I saw the other day that I need to make sure that you have your hands on is the same thing, but using the modern Wayland composite.
Yeah, I think we even had the PI one running in frame buffer at 60 FPS with transitions and video.
I mean, that was one of the reasons why engagements like ours were the foundation of embedded and industrial use of raspberry PI.
Because raspberry PI one, it's got a 700 MHz arm.
Eleven, right?
It has a very limited amount of cpu.
But the multimedia performance has always been great.
I mean, it's always been as long as you.
As long as you can avoid touching the pixels.
If you can have the pixels just zoom straight through from the codec onto the display, you can get incredible results.
And in fact, one of the things you'll see from us over the next six months or so is a really concerted effort to move desktop composition into Weyland Universe on PI one to a hardware accelerated world.
You know, a distinctive thing about us is that we continue to sell old products, so we continue to sell.
We sold 50,000 Raspberry PI ones last year, which is kind of fun.
And we also continue to support the modern software stack on older devices and to care about performance and to care about delivering performance uplift.
We sold between five and ten.
I don't have the numbers at my fingertips, but we sold between five and 10 million units of PI one.
We sold well over 20 million units of PI three.
Those are still in the field.
They're not in drawers, they're not in landfill.
We continue to care about eking more performance out of those devices over time.
We have thousands of PI three s still in the world.
Right?
It's a workhorse.
Right.
I don't know if you noticed, embedded world.
We pushed the EOL date for PI three to 2030, so it had been set at 2026.
And it's a workhorse.
We still sell a million plus units a year.
Now, when are we?
Nearly five years after we launched PI four.
And we're still selling these huge numbers of the predecessor product.
And it just didn't feel right.
It's something like two x.
So PI three alone, as three and three plus alone have sold more than twice as many computers as the Commodore 64.
And as you say, there are huge numbers in the wild, and it just doesn't feel right to do a kind of a forced transition of people from that world that they're happy with.
Up to PI four and PI five.
Absolutely.
That brings me to a natural transition.
To my biggest pain point with the raspberry PI, which is storage, as we've had multiple conversations about over the years.
Will we ever see a non CM or compute module with an EMMC slot?
Or like the rock by has, for instance, you can slot on an EMC slot.
Is that something on the roadmap or is this still going to stick to SD cards?
Well, I would say that EMMc with a slot is just an SD card.
There isn't a, there isn't a thing that makes Emmcy intrinsically better than an SD card.
I think that's, I think that's happy to be confident.
I would contradict that.
Based on my experience, the biggest by far failure point on the Raspberry PI is Descartes by order of magnitude, right?
And we've tried the most expensive, well within reason, SD cards on the market.
They will break after two, three years after proper usage, whereas you do not see that with some kind of proper flash storage, where you see, in my experience, that's interesting.
I mean, that's genuinely interesting.
I think we've had these conversations a number of times.
I think.
I mean, I would certainly, that point about expensive, I think, is definitely the wrong direction.
I would say the best SD cards are the ones that they sell a lot of, not the ones that cost a lot of, certainly industrial SD cards.
I would not avoid them totally like the plague, but they are often.
What do we look for in SD cards?
Right?
We look for power cycle.
We look for power cycle Volt.
So we have a test rig and a typical.
Yeah, so the challenge with a bad SD card, from our point of view, is one that is not resilient to unplanned power Outages.
And I think it is genuinely true that there are a lot more people who make EMMC are thinking about industrial a lot, and so they are thinking about Power Outage, unplanned power Outage.
And I think it is probably fairly rare for someone to be in the business of shipping EMMC that dies when you cut.
The problem with unplanned parentages, you have your Flash array, you have your nand, and then you have the FTL, the flash translation layer, which turns that fairly rubbish storage flashes.
And Flash is a terrible thing, and it launders that into a Block file system, into a Block device.
And the problem with that.
So there's some state in the system, sometimes stored in other Flash, which manages that mapping and the FTL internal state.
One of the problems with unplanned power transitions is if the FTL is in performing certain operations, when the power goes out, a badly designed FTL can become totally terminally broken.
So we have more broken than.
Bro, I mean, you'll have seen these cards, right?
Yeah.
That they are.
You try and recover them.
It isn't just that they are corrupt block file systems.
Oh, you can't read them.
They're destroyed, right?
They're not, they're no longer a block device anymore.
And what's happened there is that the FTL is, the FTL has failed.
Right.
So it's not that you've run out, it's generally not that you've run out of wear in the device, it's generally the FTL has exploded.
That's interesting.
And if you're in the EMMC business, you are generally not.
It's hard to get any traction in the MMC business unless you've taken that seriously.
You can sell an awful lot of SD cards to people who want SD cards to stick in their camera or their phone.
And unplanned power outages just don't happen in that world, that someone uses them and then they turn them off and they don't suddenly lose power during a flash access write operation.
And so the FTL is not stressed.
So the way that we qualify SD cards.
So we now have a qualification program for SD cards.
And what we do is we, and as I say, it's often industrial cards that struggle the most with this.
We take, I think it's about 32 PI, four s in a rack, and we put the SD cards into all of them.
We set each one building the Linux kernel in a loop, and then we randomly have a little relay on each one and we pull the power.
And a torture test across this array is hundreds of thousands of unplanned power cycle events over the course of about a month.
And if you have a single card that dies as a result, you are, you don't pass the test.
And what cards tend to do overtime.
Cards have got better.
I think there is an element of competition in this dimension, the, to the point now where if you got a random card from a chinese second or third tier vendor five years ago, it would die the first hour of this test.
And now we have equivalent cards from equivalent, the same vendors or equivalent vendors which survive all the way through them.
Cards have always done well.
Sandisk cards have generally done pretty.
Sandiskars, really pretty tough.
Not, not you, not all of them, but generally sound discard is pretty good.
They've been good previously the whole time we've been doing this, and we see these certificates.
So if you go on Pip, if you go on the product information portal, pip dot raspberry, PI.com, comma, there's a whole section of certificates that you see to a pretty broad range of vendors now for SD cards.
So, I mean, that's my general feeling about SD cards is there are more terrible SD cards than there are eMMC.
But a good SD card and price is not the right price, or industrialness is nothing the right metric.
A good SD card can be as good as EMMC because there's no technological reason why it shouldn't be.
It's a decision to be bad.
So that said, particularly, I do have some sympathy in the, I have some sympathy for the fact that it's never a bad idea to help your customer, not have to make a decision providing a service if you help.
So compute module, we provide our customers with a service.
Don't think about flash, we'll think about Flash.
And not requiring people to develop a competence and understanding this stuff is a thing.
Also jungle vibration types and all that.
Also vibration.
You could understand that people are leery of using SD cards which aren't soldered down in a high vibration environment, or there are sort of medium severity industrial environments in which SD cards aren't suitable.
And so you've seen people migrating to compute module under situations which for us feel economically suboptimal.
It is harder to, you know, it's more work to do to integrate a compute module.
Therefore people tend to do it at higher volumes.
But you see people go early.
I wonder if you go early.
So, yes, I think there probably at some point will be, it won't be socketed emMc.
I just think socketed the MC is just a weird thing, but soldered down.
It's worse to both of them, but soldered down EMMC on a big pie.
I think there's a reasonable chance that might well come to one or other of the more modern platforms at some point in the next couple.
Interesting.
Yeah.
Just looping back on the reliability, we've compared notes carefully with Balena, and they are probably one of the other ones that are, I guess, operating PI set high scale.
So they have a lot of data.
We have a lot of data.
And it's funny because we've come to the same consensus on the best sd card that we both use and recommend and using all our builds.
So the screen official build is using the Sandisk extreme, which is the exact same card as Balena is using.
Yeah, yeah.
I mean, they're a great company.
Sandisk.
I mean, we used to have a wonderful relationship with western digital, actually, and it was kind of fun when they combined.
Combined forces.
It is encouraging to see other people coming up on the rails, though.
And it's that power cut torture test is.
Is just, you can have the best cardinal the world with the best random read ops and the best random write ops and great streaming performance, and it will just die after an hour in the rig.
And you have to go back to these people who sent you a card.
They're like, can we get a certificate for this card?
I'm like, this card is so bad.
You know, if we use an industrial context, it won't last a week.
Yeah.
On prime power cycling is.
Is a feature of the industrial world.
Yeah.
It's funny because I.
I spoke to the Sanders guys at MWC a few years ago, and I mentioned our experience, like, hey, there is break.
And they're like, no, that's unheard of.
That doesn't happen.
They're like, well, I have plenty of evidence suggested it does, but you got it.
That's it.
And I think we're still big believers in SD cards, particularly the higher performance modes that are available in modern cards.
I think I would love to be able to ship SD Express, and that hasn't really rocked up yet.
But even the high speed classic modes, the kind of hs 400, you know, the HS 400 modes, the a two class cards.
Yeah.
You're really.
There's always that feeling that the NVMe, the whole raspberry PI system is designed.
The nice thing about Raspberry PI, five, of course, is you can plug a m two drive into it, you know.
Yeah, there are.
There will shortly be a first party accessory to let you do that, but there are currently enough of third party accessories that we love.
They'll let people do that.
But really, the whole Raspberry PI world has been designed to not require NVMe.
Yeah.
The operating system is tuned to be lightweight enough that it runs very acceptable from even a modest SD card.
And then as SD card performance has come up, either with faster bus speeds or with things like command queuing, which are CQ, it's got actually a lot of the benefit of NVMe over classic SD card isn't about bus speed at all.
It's about the ability to queue commands.
Right.
So it's ability to hide latency in the access to the flash array.
And now that's come to SD cards.
That's a big, that goes a long way towards closing that gap into speaking of.
Performance and raspbian, or sorry, Raspberry PI Osdem.
One of the things that we discovered in the early days was that swap was enabled by default in Raspbian.
Has that actually finally been closed now?
I would have to look.
Because that was the absolute killer of SD cards in the early days was that swap was enabled and it just completely wore down and trashed the card.
It depends.
It depends.
I think it's one of these things that has, there are, you know, particularly a modern high end raspberry PI has 8gb of memory.
It has the same amount of memory as this laptop I you on back in the days of 256 meg raspberry PI one.
There is an argument that there is writeable, there is a pool of writable memory that gets written, effectively write only memory.
There are, there are bits of data that get written and never looked at again.
They have to be available because somebody might look at the one day, but nobody does.
And so I think the rationale was that it's helpful to have somewhere to stash that you can eke out some percent of additional functionally available ram by allowing the page cache to push those very low read utilization pages out to storage.
And the problem is, how do you ensure that it only does that and doesn't start opportunistically pushing other things out, which, as you say, does then have that thrashing effect.
I mostly care about it because performance just craters.
But yes, in terms of, in an industrial environment, you need to get that turned off.
So it's less salient than it used to be.
I can see why we did it.
And of course, like everybody, we had that flirtation with Zswap.
That was a real, really sad thing that weren't able to articulate a benefit to Zswap.
I mean, it's the old Ram doubler world.
Anyone who's old enough to remember the very first, oh, we have protected mode now in Windows 3.1.
We can double your ram by compressing it.
It was kind of slightly depressing to discover that actually was really hard.
We didn't not try this, but you know, and on the kind of one gig devices, you know, there's also a pool.
There's a pool of less.
Yeah, that pool.
Perhaps the right thing to do, rather than writing it to a persistent.
To persistent storage, is to compress it and then stash it in ram.
And yeah, I mean, perhaps there's a slightly larger pool of stuff which is slightly more frequently access, which also should be compressed and stashed in.
But were never really able to come up with a convincing configuration of that subsystem, which didn't make some use cases materially worse.
And so we left it turned off.
We know there were users who swore by it and used to get a raspbian out the box and immediately turn Zen swap on, but it was very use case dependent.
That's fair enough.
That's fair enough.
All right, let's take a look at the future.
And obviously one thing that I assume at least is on you guys mind is RiSC V and how that plays into the PI ecosystem.
Will the PI six be RisC V based or is that future out?
Or is it even on the roadmap?
What are you guys thinking?
Would you like it to be?
I would love it to be.
Why?
Well, I guess it's.
The concept of open hardware is very compelling from a security supply chain perspective.
I think the challenge you have though is that it won't be open hardware, right?
I mean no performance risk five.
So the architecture is open.
I mean, what was risk five have going for it, right?
You could implement RISC five and you won't get sued.
Well, probably yet, but you know, it's no convincing demonstration that there is anything suitable, that there is any protected ip in the instruction set itself.
And that's really compelling, right, because it creates a different community.
All of these instructions and architectures create a community of innovation, right?
They create a.
And all of them have a different shape to community of innovation, right?
So X 86 and you know, so intel and AMD together create a community of innovation where, you know, you can get go and buy a chip and you can buy chips at particular performance points.
They're basically just chips.
At least historically.
They didn't have a lot in them other than just the cpu and memory controller.
Well, not even the memory controller for a long time.
But you know, the memory, the cpu and the front, side and backside bus north and south bridge interfaces on them and then later on memory controller and South Bridge interface.
So, but you could go and build a system around that as a hardware integrator and you could go and build a software that ran on it and you'd know what the chip could do.
And it provided you with a kind of a known abstraction as the software and then to a lesser extent, other licensed licensable cpu architect.
Licensable cpu cores and architectures, isas and microarchitectures, provide a different community.
you can now as a software user or a system builder, you interact with that community, you interact with that in that architecture through a level of indirection.
So you buy silicon from licensees, but those licensees have the ability to build much more sophisticated systems.
around the platform.
There's an obligation to pay money, of course.
So there's somebody extracting money, arm extracting money from this system and providing services.
And the primary services they provide are regulation, so they regulate what can be claimed to be arm compatible.
So you have to, you can't just take an Arm core, add a bunch of new instructions to it, kick it.
Well you can a little bit now actually with custom instructions.
But in general, historically, you can't just get ticker and arm core and add a bunch of instructions to it and kick it out the door, or take instructions out of it and kick it out the door.
And then they provide access to a pool of readily licensable.
And some people consume that level, so Apple consume that level.
So they go by what's called an architecture license, which is basically you pay not to get sued, which is nice.
You license the right to make your own core that implements that architecture, or you go and buy.
And this is what raspberry chip's always been built on, the other model, which is where you go and license one of a pool of available synthesizable cores from arm.
And that's the other service.
They provide a pool of good quality, known, verified arm cores that you can integrate.
And so that's a different shaped ecosystem of innovation, but it has many of the same properties, and that's certainly up at the software level.
You can rely on the system presented certain abstraction to you.
And then RISC V creates a new model where there's less value extraction.
There's no value extraction at the architectural level.
There's still generally value extraction at the core level.
So for the first time in this world, what does this world add?
Well, it adds a couple of things.
It adds your ability to write your own core, and therefore the available, quite a large pool of reasonable quality, freely available cores.
So there's not necessarily, so there's no extraction the architectural level, there's potentially no extraction at the does, there's no extraction in the instructions at architecture level.
There's potentially no extraction at the microarchitecture level.
Those are good things.
Plus you can add your own instructions, you can mess around with the design, and that's absolutely fine.
Downsides.
As a software user, there's not the.
You can no longer rely quite on the.
It says risk five on the tin, therefore it's exactly like this.
It may be like this, but it may have some more instructions that you may come to rely on and then discover that some other RISC V Corps doesn't have those instructions.
It's a much less orderly universe than they are.
And also the supply of licensable cores is not great, actually, compared to what you can get in the.
In the art, in the armed world.
And so, and the decent cores are not free, right?
Yeah.
So, and that means that they are closed and they are less.
They are no more or less.
All else being equal, they are no more or less secure than an arm core.
But isn't that where the python actually could make a difference and make an open core for the future?
We could.
Vast engineering effort.
Oh, absolutely.
I mean, a 76, which is what's in PI five, if that's not 200 engineer years of.
I mean, I don't know what the number is.
I have the occasional conversation with people and the sort of number that I think I've been guided towards for a mid a 70 x class core is 200 engine e.
Right.
You're making a courier and you have a few hundred people working on it, and it's not from a standing start.
Right.
Because you have the previous core to build on.
Right.
You're standing on the shoulders of what you did last year.
And so this, these are huge investment and fascinating.
The fascinating thing about the arm world is how cheap it is to participate.
You can get access to that for not huge sums of money and then go do your innovation ready.
Build your silicon, start selling silicon, start building boards.
It's not an expensive place to play.
It's very hard for you to stand up and say, I'm going to spend.
I think you're going to spend 100 years.
You could easily blow 100 from a standing start.
And now you look at the sort of amounts of money that risk five startups are raising, it's clear that you are talking about spending hundreds of millions of dollars from a standing start, and you got to go get that money from somewhere.
So you've either for us, we'd have to go raise it from investors, or we'd have to not give the money to the foundation.
So suppose we made the money ourselves.
You'd have to not give the money to the foundation.
If were to go make an open call and do it as a service to humankind, that would be $100 million service to humankind, which we would make at the expense of a charity that educates.
I guess.
I guess.
Should I be doing that for it?
That's a business question.
I guess the argument for it is.
I think when it's kids who'd be getting the money otherwise, then I think it's not a business.
Fair enough, fair enough.
What I mean is when that has happened, if there's a pool of companies that do come together and do that foundation, the incremental benefit later on for each generation will be insane.
Because now this is back to this idea of a community of innovation is what are the incentives?
You absolutely could imagine some cartel of consumers of coordination is the problem that capitalism solves, right?
So coordination is the distributed attack on problems is what the price mechanism does, right?
And it's been incredible.
I mean, look around, everything behind you there was built by that.
Everything we have, everything we have.
We only don't live in caves anymore, right?
But it can't solve all problems.
And this is one that has historically been extremely challenging, this collective action to move from some.
It's a hill climber, it's a hill climber basically like capitalism's a hill climber.
And what it can't necessarily always do, what you're articulating, it makes a huge amount of sense, right?
That there is a globally more optimal place to be where all of us who use CPU's club together around some architecture, and we all contribute some money and we agree to go make a Cortex a 76, except it's RISC V.
And to put it on and to document that it's IP clean, because that's an important component to our document that this thing was not part to give people.
Because you're going to invest a tape in this, right?
You can put it in a chip, you're going to spend, you even once the core exists, you're going to spend tens of millions of dollars turning it into realizable silicon.
You could argue that one of the things you buy from either from arm or from a commercial risk five core vendor is IP.
Certainty is some evidence that there's somebody who will guarantee you, overwrite you a contract that says this thing is mine to license to you, and somebody you have recourse to in the event that isn't true.
But suppose we all get together and we all do that work and we produce not just a core but some collateral supporting the core that demonstrates that the core is clean.
And we put it on GitHub.
And as you say, thenceforth things will be wonderful.
You are building on.
Then the next year's core becomes much easier to do.
You're building on the shelf of the giants.
You unlock some distributed innovation.
You can see and sometimes something will in the software.
It's much more common in the software world for things to, because you can do things more incrementally.
It doesn't have these tape outs.
And this general Linux was fascinating because where Linus got to was he got a system which was good enough that it could become the nucleus for distributed innovation.
Replicating that in hardware, because of the challenges of making things out of atoms are much larger than challenges of making things out.
Bits, copying bits once they exist.
It's never been the case.
The open source hardware world has never had the vibrant nature of the open source software and the open source chip world.
Is that on steroids?
Is that plus $10 million tape ads on modern process notes?
Right.
It's hugely challenging, but I mean, I'm absolutely in agreement with you.
The question is there a market based, you know, the crazy philanthropist who rocks up one day and says, I'm just going to solve this problem for all mankind is a.
Is a potential.
It does sometimes happen, right?
It happened to.
It appears to be happening to malaria vaccine, right?
Another place where we all agree with the world would better if nobody died of malaria.
But capitalism hasn't provided us with mechanism for collective action other than to have one guy get rich selling operating systems to people and decide he's going to solve the problem.
Right?
Yeah, I guess there's answer this.
I really care about this.
I know Raspberry Six is not going to be risk five based because there are only calls that are any good that you can license and the software isn't mature.
And we're a member of the risk five foundation.
And I hope I've articulated lots of positives about the risk five world.
And that's not to say you might see some deeper involved in risk five from us, but the idea that it's going to be feasible to go and procure an a 76 successor core for Raspberry PI six sometime before the end of the decade, that's risk five.
And be confident that the.
That the core is good, that the core is competitive in all dimensions, technical and financial, versus some Arm core, and that the software ecosystem will be mature.
I don't think it's realistic.
Fair enough.
It would be amazing to see a collaboration with saying the PY foundation, the Linux Foundation.
I don't know.
Meta is doing a lot of things around open hardware.
They are putting the money where the mouth is, doing open server hardware.
That say you can appeal to, you know, that you can, you may be able to find.
That's probably the big hope, I think, for RISC V in the big core space.
Is that one of the hyperscalers?
I would think probably Google the one where I can sort of draw a sketch picture on a piece of paper, might be able to articulate in their minds some business model for some disruptive open documented, ultra high performance risk five core.
That's probably the.
Yeah, if a surprise comes from anywhere, and I think it will be a big surprise if a surprise comes from anywhere, I think you're right.
I think it will be from a hyper scaler.
I'm not sure it's meta, but I think that, you know, some of the other, some of them have a.
Could articulate a strategic interest in driving commoditization in the core, in the microarchitecture and instructions and architecture purchasing space.
Fascinating, right?
Absolutely.
I mean, yeah, I guess for me the anti goal for risk five is that it becomes another proprietary play with just a further layup of abstraction layer.
Right?
Yep, yep.
And that's sad, right?
So, but look, we've been members of the foundation, I don't.
Well, risk five international as it is now, since very near the start.
I wouldn't rule out some involvement, some sort of more deeper engagement with it.
In the Pico maybe, but.
Well, you know, you could imagine, I mean that certainly is a place where you could.
The barriers are shallower there because the cores are simpler and the software stacks are shallower.
And the incumbency advantage of the existing architectures primarily, but not exclusively, you've got to remember microcontrollers are not by any means dominated by.
By arm at the moment, 32 bit ones generally are.
But microcontrollers in general, there's an awful lot of eight and 16 bit stuff out there.
The incumbency advantages are not so substantial.
We have at least one significant player in espressif who've been very to risk five for the time now you can certainly imagine that the microcontroller space might be home for innovation, for that architecture or any other new architecture.
It is a playground.
It's an architectural playground, always has been.
Yeah, absolutely.
No, that makes a lot of sense.
And even if you do solve that, you still have the GPU problem.
Solve by no open source GPU.
Well, you know, I did once build a GPU video core.
Video core.
There was a handful of us in Brooklyn and Cambridge, including James Adams, who's our hardware CTo here now.
So we do have previous.
Although I think building a Vulcan.
Building a Vulcan capable gpu, we built an open GI video core four, which was the one that I was most deeply involved with, was an Openglas two class gpu.
And OpenGL S two is actually very simple.
It's quite a classic view of what three of what 3d hardware.
What GPU hardware is for 3d hardware.
It's for, I think, building a quality Vulcan capable gpu.
Of course, we have a quality Vulcan cable gpu in video core six, video core seven.
I forget what the brand name is now at the video core.
That's in PI five, video core seven.
I think.
Yeah, we have.
Yeah, we have this quality gpu, but a little bit like the arm cores that were talking about earlier.
They are built on the shoulders of giants.
They're built.
This year's core is built on last year's core.
A ground, I think a ground up Vulcan, competitive Vulcan GPU is another 100 engineer program.
Yeah.
So, yeah.
No, I guess we're not gonna see pure open hardware anytime soon than.
It sounds like.
That's it.
Yeah.
I wish.
I mean, look, don't get me wrong.
Yeah.
But it's.
It's.
It's a fickle capitalism is a fickle beast.
It can only access.
It's fascinating.
Was it.
It's Ian M.
Banks, isn't it?
Talks about.
I think it's Ian M.
Banks who talks about how.
Because the culture.
You've read the culture books.
Yeah, because they're the fleebus and things.
And the culture.
I've not actually.
Oh, you should.
I watched Elon Musk sell the culture books on the stage I went to after the AI safety summit in Bletchley earlier this year.
I went to the after party thing, where Rishi Sunak interviewed Elon Musk.
And Musk spends half the thing selling Iain M.
Banks culture novels from the stage.
If you want to know what the benevolent outcome is, the non terminator outcome for AI.
Read the culture novels.
And he's quite right.
Disagree with him about a lot of other things, but he's quite right about that.
But the thing about the culture is that it's a planned economy.
Right.
It's communist.
It's the thing they sometimes call fully automated gay luxury space communism.
When people use that set of words they're generally talking about the culture or something very like it.
And there's a bit, I'm pretty sure it's in there or if it's in some of the notes he wrote where he talks about how the planned economy and it's planned economy, but planned by superhuman AI's right.
And he talks about how the market economy gutters like a candle, like a flicker and candle while the planned economy lazes its wonderful language, right?
And it is true.
Look, I mean we surround it, dont knock it every time somebody, its capitalism like democracy.
Its the worst system, apart from any of the alternatives were surrounded by the products of it.
And we live in unimaginable luxury for a medieval king.
We live in, the poorest of us lives in unimaginable luxury.
But there's a huge fraction of the space of the state space of the world which is inaccessible to market mechanism.
And it's fascinating and you kind of weep for it.
Yeah, that's a good way to wrap up.
Do you have anything else you want to add about things about the power foundation that people should be on the lookout for in the near future or.
I look, I, this is okay, I say.
So I say this every now and then.
I had a BBC micro when I was a kid and I got it school and I got into it at school and I bought one, a very second hand one that I had at home.
And every other Friday night I used to go to a computer club in my town in Ilkley and I used to box up my beeb in its polystyrene box and my black and white 14 inch telly and some stuff used to take it over.
And at about 07:15 every other Friday night a chap called Alan Drew would rock up in his people carrier at my parents house and take me off to a computer club and bring me back at 930.
Or later on when I got a bit older we would go to the pub and he's not, sadly not with us anymore.
But I wouldn't have got where I got if it hadn't been for Alan and a few other people like him, often teachers.
But Pete volunteers as well, who paid attention to me, right?
Who gave me a place to get a physical place to go and gave me an audience for the things I did and I was their audience for the things they did and gave me as a ten year old kid, 1112 year old kid, a way in to this community that we're all part of very early, like much earlier than you get it into it.
If you're, if you're rooting to computing, is to study maths at school and computer science university and then go get a graduate job.
And I was into this world when I was still tiny little.
I've got seven year old daughter, right?
You know, when I was not much older than her, I was surrounded by adults.
People should go and be Alan Drew, right?
Go be Alan Drew to some little eleven year old squid who's probably as annoying as I'm sure, lost to these people and just put up with them and answer their questions and talk to them like they are worth something even before they are.
And it's good to do good at scale.
It's good to give money to charity, and it's good to try and build things like we've built with raspberry PI, which is a business thing that Liz and I dreamt of all those years ago.
But just try and do something at the individual scale as well because it makes a difference and it, because it'll make your life better and somebody else's life better.
There you are.
So that's the thing you should watch out for.
Don't watch out for things from the Raspberry foundation.
The Ross we are foundation is doing amazing stuff, Ross replied.
Limited is doing amazing stuff.
We're making great products and selling.
We're changing the world, right?
And we'll keep trying to do that, but we can all do as well.
I think that's a great way to wrap it.
I'm a huge fan of mentoring myself.
So I can definitely sign that off.
And I can relate to the fact that I'm not where I would be.
I'm not, I have not, would not have been where I am today without mentors either.
So I can definitely agree with that.
So very wise words.
Thank you so much for coming on, Eben.
Much appreciated.
Good catching up with you and thank you so much.
Have a good day.
Awesome.
Thank you very much indeed.
Cheers.
Bye.
Found an error or typo? File PR against this file or the transcript.