"I can't go out today and license a RISC-V core,"
Now there is a project I wish I had time to get involved in - and he wouldnt need to license the bugger either!
Today marks 10 years since the Raspberry Pi was made available to purchase. We spoke to Pi supremo Eben Upton about the last decade and what the future might hold. The Raspberry Pi team has elected to mark the milestone on 28 February, it being the last day of the month this year – for many 29 February 2012 was when units were …
The Pico could've been RISC-V, but they went with ARM for their own silicon. You have to draw the conclusion that they're really not interested in open source, for all the money they've made out of it.
I think they're making a real mistake in not embracing RISC-V : their core market is still open source techie enthusiasts, who are eager for open hardware. Upton et al might one day soon wonder where all their customers went.
they're really not interested in open source
The Pi originally shipped with a huge binary blob containing Broadcom proprietary code. The size of that blob and the amount of closed-source code diminished over time.
The Pi team will have to speak for their own motives, but it seems to me a crucial part of the Pi's success has been in getting the price right and if that meant some foot-dragging on fully-open licensing that seems to me to be a perfectly reasonable trade-off.
I agree with you, and with others here who applaud what they've achieved (and for using chips that the industry had thought were waste):
- the first of its kind
- documentation and, as you say,
- price.
Remarkable.
But I do wonder if they've also missed some tricks. Look at how Pine64 riffs on their boards, with the SBC, the laptop, the phone, the watch, the tablet, the e-ink... Meanwhile RPi have left it to 3rd parties to do things like the CutiePi.
And I do kind of resent them for going after Arduino with their Pico.
Competition is normally a good thing, if both sides have to raise their game to get the traffic.
Minor complaint with both ecosystems. In my line of work I need folks that can deal with the down and dirty of TTL logic circuits and the like.
Still, Pi is a hell of a useful tool for getting "easy-mode" electronics out into a wider audience.
I forget the name of them exactly, but in the early 90's I recall my school having a ton of logic boards showing and/or/not gates etc that you could plug bits of stuff into. Teacher I remember set me away building a 4-bit safe lock (not very safe!!!) - which took me about 5 mins to prove (through brute force naturally). I built a half-adder out of the boards later that day. I suspect scared the bejesus out more than a few of the staff.
Of course, having used the C64 since 1985 it was all pretty trivial. The C64's manual covered logic gates well, and a copy of "the way things work" had me building a half-adder later the same day out of a stack of these boards.
One downside, they didn't have a bloody clue what to do next, being "that far ahead". Which wasn't really that far ahead, but principle of the thing stands...
I do wish education generally would leap on the strengths of individuals and develop them. Yes, yes, some fundamentals needed by everyone. But not to the detriment of actually getting good at what interests you!
The problem is having staff skills and time to support it. I was lucky to have staff who could & would go the extra mile when I went off-piste with the curriculum (eg the biology teacher who just happened to have her 2m handheld in school for me to borrow the day I told her my ham radio licence came through - first QSO at lunch time from outside the science block :-) but that was a while a go and funding has got tighter since then (and curriculum much more prescribed). There are some schemes out there which will stretch the brightest without too much staff input, eg UK Mathematics Trust offer mentoring for their challenges, but you're still dependent on the school's awareness of such things - and then of course the volunteer mentors.
And I do kind of resent them for going after Arduino with their Pico.
So do you also resent the ESP-based modules and the ARM-based modules by the likes of Adafruit that are "Arduino-like" but not actually Arduino?
I think the opposite. The RP2040 (that is, the chip at the heart of the Pico) is just another option for people building Arduino-alike boards, yes they built it into the Pico, but that was essentially just a demonstrator. The really interesting boards are by Pimoroni, Sparkfun, Adafruit etc. Adafruit have "Feather" boards featuring ATMega chips, ESP32 chips, AT SAMD (ARM M4 and M0 chips) and, of course the RP2040 which is two M0+ cores and "a bit extra". All their Feathers can be programmed in an Arduino-alike way, though they are also pushing Circuit Python on capable boards. I'm not sure that's a bad thing.
Each chip has its niche, and if you can see a gap in the market - which the Raspberry Pi people obviously did - why not take it and see what people do with it?
M.
I'd be inclined to believe Upton that there isn't a RISC-V SOC that is as cost effective as the ARM options. Its probably beyond expecting the Raspberry gang to be able to rapidly obtain the chip design skills to contemplate evaluating a no-yet-gone-to-silicon design. I'm fairly sure in the next couple of years there will be available parts if not a complete silicon SOC available and then perhaps they can have a look at cross-compiling Rasbian and all the apps to it and launching another revolution in credit-card sized PCs. For now they are best to stick with what works. Pi5 may be a RISK-V design but its too much to ask Upton and co to make that leap now.
It's true if Sipeed have gone with the unratified AllWinner D1 (on their Nezha SBC) which is nowhere near as powerful as the ARM on a Pi4, and can't sell it for less than £87 direct from China (or £140 all taxes paid on Amazon UK) then obviously the economics aren't there yet.
Pi5 will be ARM just like all the other back-compatible Pis.
There are already RISC-V SBC offerings from SiFive, Sipeed and others.
Beagleboard are promising two RISC-V SBCs this year, and it will be really interesting to see what they produce. If a 64bit OS on RISC-V gets close to the speed of a Pi 4, then that's sufficient for general computing - at least, that's what the RaspberryPi marketing says!
Then it'll come down to apps, and price.
Wrong conclusion. We ARE interested in open source, which is why so much of the latest software has replaced the proprietary firmware blob with open source alternatives. KMS, DRM, libcamera, V4L2 etc. It just takes time to make this transition, and make that transition easy for users - we like backwards compatibility.
With regard to RISC-V there clearly is not a core available that is better than that in the Pi4, never mind the more recent Arm architectures.
Well... just so that you know : my next hardware purchase won't be ARM architecture, it will be (open source) RISC-V. And a number of people I know are saying the same thing.
Your marketing was right - sufficiency is enough. So long as it's capable of doing general grunt, there's no need for the hyperbole in tech any more, though I understand the headache that that causes tech companies.
At twice the price? Ok, there may well be a hobby market for that.
But why are you waiting for Raspberry to do this for you? Just fund it on GoFundMe or Patreom or something with other hobbyists. Remember, the source is available, the silicon will work first time and it absolutely is as easy as the open-source crowd think it is.
JFDI.
For selfish reasons I'd be happy to keep the video blob on the Pi - even if it's as an installable option. That thing (using Omxplayer) is insanely good - I have had three videos playing simultaneously, scaled, with transparency on a 256MB Pi1, for heaven's sake (four on a 512MB board). I've not had a chance to try anything similar with VLC or Mplayer on the 64-bit OS on a Pi3B+ or 4 yet, but even though it is said to be "accelerated" I'm told it uses a lot more CPU. At least for now I can stick with the 32-bit OS and Omxplayer...
M.
It's a trade off. A processor with fatter I/O would obviously solve that, but half of the point of the Pi is stack em high and sell em cheap using readily available parts that were intended for Phones or similar.
I'm not sure Broadcom (or others) would be desperately interested in making a SOC purely for the Pi Market, though it is not impossible given the volume demand. Fat I/O isn't "really" a requirement for IOT devices either.
Small form factor computers with rather more I/O are available - whether you would want one is another matter of course!
You don't NEED cooling of any type. You can add it if your particular workload causes thermal throttling but the latest DVFS firmware does a great job of keeping things cool.
Passive cases work fine, we sell a case fan if you want one. But it's not essential. I never use any extra cooling.
Not really. It can help if you have it in a tight space, or you are heavily loading it all the time. My main Pi 4 is in a tight space and I do load it quite often and the PWM seems to fire up about once a week for a moment. Give them a decent PSU and attach an SSD (I use a geekworm x820 board) and they are very happy.
If you need more power, pay more money. Easy.
I'm not sure Broadcom (or others) would be desperately interested in making a SOC purely for the Pi Market, though it is not impossible given the volume demand. Fat I/O isn't "really" a requirement for IOT devices either.
Didn't I hear that the SoC used in the Pi4 is essentially custom silicon, though nothing terribly out of the ordinary?
People have long called for SATA, but SATA isn't a terribly useful interface unless you want to connect an SSD or HDD, and I'd contend that very few Pi projects actually need that kind of local storage anyway. Eben once said (and I can't find the reference now) that they do a cost - users analysis of everything, so if it costs 20¢ to add a specific feature that will be used by 20% of people, it has to compete with a $1 feature which is used by everyone.
I predicted, because it seemed obvious to me, that rather than adding SATA, the best thing to do would be to swap the slow USB2OTG port of the original Pis for a proper USB3 port. Much more flexible (you can do a heck of a lot more with a USB port than a SATA port), and the bandwidth can be used wherever you need it - it has allowed a (near) gigabit network interface, and a USB3 to SATA adapter is practically native SATA speed anyway.
By updating the USB interface in the Pi4 for presumably not very much money, they have created a multi-purpose feature that will be used to a greater or lesser extent by just about every Pi user. Even if you are only connecting a keyboard, it's a more stable interface than the OTG one, which needed a "special" driver to work properly, but it opens up faster networking, faster external discs, better external cameras, audio interfaces, sensor sets and much more. Good choice I reckon.
M.
I've said it before, but I'll keep saying it... include an eMMC or similar.
The original idea of using SD cards was great at the time, but SD cards aren't as ubiquitous now - we have to buy them specifically for the Pi. Add (selectable?) USB mass storage mode to write to it and you cover the 'unbrickable' aspect.
(yes, net booting helps, but I would prefer something that can operate independently of the network/server)
SD cards aren't as ubiquitous now
Really?
Methinks the (micro-)SD card is more ubiquitous than it's ever been -- the competing standards (Compact Flash, Sony Memory Stick, etc) have all vanished and SD is what there is. It's also cheaper than ever and available in larger capacities.
It's true that SATA and NVMe SSDs have better controllers, better (re-)allocation algorithms, and consequently longer lives ... but they'e also an order of magnitude or two more expensive.
I/O is always likely to be awkward on a device that size, so I'm used to ignoring it now... and I suppose there are a few boards available now that use the computer-model-4 (CM4) that may provide alternative I/O layouts.
However, I will just mention that having swapped the side where the ethernet jack is on the Pi4 is surprisingly more annoying that I originally expected. I think the (my) problem is that some ports are static, e.g. power, HDMI, ethernet, and things like USB are more likely to be hot-plugged. But with the re-positioned ethernet, the Pi4 puts the USB ports between those static ports.
Might just be me though; I thought I'd get used to this, but, nope!
Otherwise loving my Pi's!!! :D
Best £25 I've spent was on a PiZero. My wife wanted to test out WordPress, PiZero with Apache2 + PHP7.4, MariaDB and a copy of WordPress from source. A little sluggish but hell that's a full blogging site LAMP stack in a form factor the size of a matchbox all running off the 5v supply off a USB port!!
Now compare that to the multi-GB sized dev stacks we all code in that need 16GB+ memory and multi-core processors just run a few command line utils we've coded.
I keep hearing about the new Pi models. Then I go to online sources and they're never in stock. Sometimes obsolete models are available, or higher end previous-gen models in expensive bundles. Or I could buy one on Amazon at a huge markup, from profiteers trading on scarcity. It seems like this "disruptive" computer has fallen to the same old greedy vendors and supply chain problems.
I have three. Two aren't doing anything, but only because they've been superceded by my current Pi 4. Mine is used as a media centre machine with Kodi on it. It was a doddle to set up, streams HD video over my network with no issues and can handle 1080p H265 video perfectly. What's so difficult?
Seriously?!
My Dad ( now in his 80s! ) is building projects with Arduinos and RaspPi's to make sure he keeps his noggin active. We get together every so often to try out a few silly projects. He's built remote controlled cars using bluetooth, security door locks, all sorts of silly things, needs help with the coding as that's not his strong point.
My Dad is the first admit is not the sharpest knife in the box any more at his age but he can find ideas to keep him going then I'm sure you can!
I have three. One is a music streamer (Squeezeplayer) with an amp HAT, driving a pair of Gale speakers - the smallest little stereo I've ever owned ! One more is the Logitech Media Server for the system. The third is currently idle - it was a PiHole, but I had some problems and haven't had the time to solve them.
And the fourth I owned was sold to a collector as it was a very early one.
I've got a couple of PIs, and I'm trying to buy a 3rd. A Raspberry Pi 4 8 Gig. Trouble is CPC were supposed to deliver the week before last. Then last week, I got an email saying my delivery had been delayed by 2 weeks. I got another email yesterday saying the Pi had been delayed a further 2 weeks. I think at this rate, it's going to be a while before I get it..
I think the Pi is great. It's a great way of getting kids into electronics and coding. It's powerful enough for general coding and general use. It's cheap enough that if you are wiring it up to something, even if you damage the PI, you haven't lost much. It's certainly easier to justify wiring up random circuits to a £60 Pi than it is an £800 PC or £1000 Mac. Having GPIO on board makes it a *lot* easier to add extra sensors and circuits than using USB, even if you are wiring in a button, light sensor and LED, and just programming the PI to switch the LED on if it receives an input on one of the other two.
I also love that projects like the Pi Tube and Pi Storm are enabling the user to use the Pi to emulate the CPUs (and other parts) of old computers, such as the BBC Micro/Master and the Amiga. Well, in the case of the Pi Tube, it doesn't emulate the Beeb's CPU, it emulates any of the co processors Acorn made available via their Tube interface
PiHole helps save my sanity.
PiVPN.
An old model A in the garage runs a bunch of weather sensors and publishes MQTT to a broker. Another Pi runs a Smalltalk MQTT client I wrote and displays that data along with a fleet of other sensors around the estate.
Another runs a Smalltalk code repository.
Two Pi4s serve as main development machines, one for 32bit the other for 64bit, though the 32bit setup is really only for Scratch legacy support work for RPT. Several more live in PiTop ceed units for teaching.
Some are setup as MotionEye camera systems.
Another is an ISS backup of sorts.
I do have one or two in the cupboard, including an NIB Pi 2, but the others are damaged for one reason or another and are just in honourable retirement.
If I ever find time I want to play with Asterisk/IncrediblePBX, some home automation stuff to replace my current systems, a laser cutter, a CNC router, 3D printer etc. Lots of work left for Pis.
I bought a Pi Zero last Summer, stunning piece of kit. Just today I installed MariaDB, Apache2 with PHP and a full version of WordPress, functioning on 512MB memory and a tiny processor, all running off the USB out the back of my laptop in a form factor that's barely larger than two fingers on my hand, all for under £25!!
Now that's engineering to the proud of!
Can't fault the PI. They are brilliant.
Currently have three in continuous use.
1. Retropie connected to a home made arcade joystick table which resides in our living room and uses the TV for the screen. It's common to have an Asteroids party when visitors are allowed.
2. The ever present PI-Hole that every house should have.
3. As an Amateur Radio Hotspot for Digital Radio Transmission. I use a ZumSpot hat on a Pi-Zero and it's used almost daily to talk to radio repeaters around the world. Mostly Florida and Okinawa at the moment but that changes depending on how busy these repeaters and digital rooms can get. Using PI-Star as the front end software.