The new Raspberry Pi 4 Model B is out today, and sees the low-cost computer transformed into a far more capable machine.

Based around a new hardware platform, the new board packs in up to 4GB memory, USB 3.0, dual 4K display support, and a faster and newer CPU and GPU.

If you’re interested in what the new board can do and how it compares to earlier Raspberry Pi computers, check out our review and benchmarks or our guided tour of the new board.

The Pi 4 has been three years in the making and arrives one year earlier than anyone expected.

We spoke to Eben Upton, co-creator of the Raspberry Pi and Raspberry Pi Trading CEO, about the decisions that shaped the Raspberry Pi 4 and what new uses it opens up for the Pi.

How long has the Raspberry Pi 4 been in development?
It’s been a long-running program. We’ve done a lot of prototypes and this is the final result. If you include the silicon development, it’s been three years, essentially what’s happened is the silicon got good faster than we were expecting [the Pi 4’s new BCM2711 system-on-a-chip], so we’ve been able to pull the schedule in.

“It’s a realization of the vision that this thing is a PC,” says Pi co-creator Eben Upton.
Image: Raspberry Pi Foundation

Does it feel good to be finally be releasing the Pi 4?
Yeah. This has been three years of my life. Three years of a lot of our lives. We were talking about this in the tail end of 2015, so we hadn’t even launched Pi 3 when we started thinking about it.

It’s been in active engineering since the tail end of 2016. But these things take a long time, and the expectation was that this was going to come out in 2020.

Why are you selling a Raspberry Pi for more than $35 for the first time?
For the first time we’ll have SKUs [different options]. There should be a 1GB product for $35, a 2GB product at $45, a 4GB product at $55.
I think that’s pretty keen pricing.

We want to do larger memory SKUs, and today you can’t do larger memory SKUs at $35. We love the $35 price point, but we also want to have more memory, the only way to do it today is to have SKUs.

Why did you decide to add support for dual displays to the Pi 4?
We’re a PC and PCs have dual display outputs. If you don’t have dual display output, you’re not a PC right? In our office, a lot of people have two displays, a docked laptop and then an external display. For power users, it’s a thing. I think people are going to find that really useful.

What new uses will the Raspberry Pi 4 make possible?
Well it’s PC-like use cases. This is the one I’m going to give my parents to replace their 10 – 15-year-old [AMD] Duron [based PC] that I built for them a long time ago. This is a PC.

It’s a realization of the vision that this thing is a PC. You could surf the web on it, run office applications, open a lot of tabs in Chromium, because, as you know, the ‘open a lot of tabs’ is probably where the existing one falls down. This feels subjectively more PC like, in terms of the performance.

SEE: More Raspberry Pi coverage (TechRepublic Flipboard magazine)

What were the priorities for the Raspberry Pi 4?
I’m going to be fascinated to hear what’s wrong with this board. We’ve fixed almost everything I think. People’s objections were non-multimedia I/O all going through a USB 2.0 link, lack of 4K support, a lack of SKUs with more memory.

Some of them are legitimate complaints, so we’ve tried to address everything we could. The ability to put USB 3.0 on there was a surprise. We’d initially felt that we might be prepared to put up with USB 2.0 for this design, but we’ve got a very good bridge chip that works nicely.

Will the Raspberry Pi 4 be able to run Windows?
I don’t know. That’s a Microsoft decision. I continue to be enthusiastic. On some level, Windows is less interesting than it might have been five years ago.

I think we’ve got a Windows-equivalent environment on there. Increasingly an operating system is something that runs on my browser and we run Chromium very well.

On the other hand it would be great, it would be lovely to run Windows. But it’s an issue, it’s a question for Microsoft really. I’d love to see it, but I think it’s less critical than it used to be.

Could the Pi 4 run Windows on Arm?
Well, the power is definitely there. It’s got a lot of performance. I think it’s a multimedia question. It’s how do you get DirectX on there? Because there isn’t a DirectX driver for the GPU. We’re not in a position to develop a DirectX driver for the GPU, so where do you get one from? That’s probably the blocker.

How many Raspberry Pi 4 boards will be available at launch?
We’ll try to build 100,000 units to 200,000 units before launch.

What work is going on to make the Raspberry Pi and Raspbian more open-source friendly?
There’s ongoing transition on the multimedia side from stuff being in the BLOB [opaque binary data] to being open, to taking 3D out of the BLOB and then taking display management out of the BLOB.

You can run Pi 3+ now, where the BLOB does nothing on the display side, so your graphics is all done in Mesa [the open-source 3D graphics library] and your display output is all done using KMS.

We never shipped an operating system release with that mode enabled by default. The hope with this one [the Raspberry Pi 4] is that we can ship with 3D under the control of Mesa, and probably the display still under the control of the firmware for now, but with an active program to get KMS done for this platform.

SEE: Inside the Raspberry Pi: The story of the $35 computer that changed the world (cover story PDF) (TechRepublic)

Does better open-source support and a more powerful Pi open the door to new operating systems?
That makes things like Android a little bit easier because there’s less special magic in the platform, display and graphics are all in open-source land, so an enterprising hacker can put something together. You’ve got a lot more CPU performance.

So I think it’s feasible for your BSDs and your Androids to get further on this platform, with less help from us than before. It’ll be nice. It’s nice to have a variety of options.

Will there be a Raspberry Pi 4 Model A+ at a later date?
I’m not sure we have an obvious A+ to release. The economics of A+ work by getting rid of the Ethernet and the USB hub, cutting some board area off, taking support components out, losing some connectors, halving the memory.

I’m not sure that we have an obvious option for a 512MB LPDDR4 [RAM], which is significantly cheaper than a 1GB LPDDR4.

Therefore I’m not sure we have a way to make an A+ product which has a sale price of less than $30 and that’s only a $5 difference.

For me, I’d have to be able to get to $25, and I don’t think I have a route to $25. I suspect this generation doesn’t get an A+.

What are the advantages of using USB-C for power on the Raspberry Pi 4?
That gives us another half amp, so that’s three amps. That’s not really for power on the board, that’s to increase our margin that can be pushed to power downstream.

That’s a host port as well, so it’s the device port. It’s got the old legacy USB 2.0 OTG controller that everyone loves so much, which is now connected to that port, which might allow us to do fun things like booting it over USB from a PC.

How the Raspberry Pi 3 Model 3 B+ paved the way for the Raspberry Pi 4
3B+ was an interesting product because it was quite incremental in a lot of ways. It added really useful stuff on the connectivity side and a little bit of a bonus in terms of performance, partly because we had more clock speed and partly because we had better thermals, so you’re able to sustain peak performance for longer.

But it was also interesting because it retired a lot of technical risk from this project, so the 802.11ac dual-band Wi-Fi [on the Pi 4] is exactly the same as 3B+. The PMIC [power management integrated circuit], which was new for 3B+, is exactly the same. It has the same compact power-over-ethernet MagJack, which was an innovation for 3B+.

There’s quite a few places where we benefited. There’s a lot of new stuff obviously on this platform [the Pi 4], and I think if we packed all of the new stuff into one product release, it would have been too much development.

Will the previous boards remain on sale?
Yeah, we’re not EOLing [End of Life] anything. Raspberry Pi 1 is still available.

What we tend to do is five years [of producing boards] from launch, because we can be sure of that. Then we try and do better than that.

Raspberry Pi 1 B+ is coming up on its fifth anniversary in July, but we can still get the chips, so we’ll still keep making it.

How many Raspberry Pi boards have you sold?
25 million, about six million a year. We’ve generated £28.5 million of dividend to the foundation. Which is pretty good. We’ve given a lot of money to charity. It’s nice. It’s about a pound a unit.

Do you see Raspberry Pi sales continuing at the present rate?
I hope so. These things are industrial single-board computers. They’re more robust and heavily tested than most things that people put in the wild and say it’s industrial.

When it comes to controlling factories and signage and all of these things, they’re great machines. We’ve sold three million computers into the industrial sector last year, so half our volume, I think what this [the Raspberry Pi 4] will do is this opens up a new chunk of market for us, that we’ve then got to expand into.

What new markets do you see for the Raspberry Pi?
I think there are a lot of interesting new niches in industrial and commercial sectors. I’m hopeful for this in the thin client market too, obviously dual display is useful for that, we think that this will allow us to address a larger fraction of that market.

So, each thin client is probably $300 – $400. If we can bring the sub-$100 product into that space then a): I think we can, over time, take a good bit of that market, and b): grow the market, because we’ll change the cost calculation of a PC on every desk versus a thin client on every desk and a PC in a data center.

Read more about single-board computers