Great overview of the situation; I think one good outcome of the whole thing is that efficiency is something that's actually cared about these days, as moore's law keeps getting bumped out. Performance per watt is a more important metric than ever, and we are seeing some pretty powerful hardware using less than 5 or 10 watts!
Greetings Jeff. Loved your interview with Eben -- and I found it particulary interesting how his take on RISC-V seemed to have softened a bit. I'm looking forward to your factory tour!
Without arm chips you would still be using house heaters as computers. It forced all computer manufacturers to make more energy efficient computers. The running of a very large data centre with climate controls and air con to keep them cool was becoming exorbitant and if they can make servers run cooler then air con is not needed.
Yep. I use an 8GB Pi 4 as a server for a whole variety of tasks (SMTP/IMAP, Asterisk, Pi-Hole, backup server, IRC server, etc.) Even though the system as a whole was not very cheap, the fact that it uses so little power makes me feel better about leaving it on 24/7. The fact that it's completely silent is a bonus.
Thats why i always ask myself for what those ARM SBC with old or really slow SoCs are intended for, specially if i look at the sub $80 market, in the last 3 years of my life, both at home and at work, ive been using microcontrollers for anything i needed GPIO I/O or even a basic usb host (thanks Pico), for a home server ive been using a J4105 ITX board because it is really low power and have all the I/O and performance something like the RPI4 do not have. And for a more general use pc the RPI4 is just not enoght and all SBC below $80 are petty much at the same place or even worse. The RK3588/S is a good breaktrought in ARM SBC in performance and I/O but the price has been going up non-stop. These days i have more hope in the Intel N100 landing in a good price/power/performance area for SBC and minipcs, because even the RK3588 that is old ARM arch by today standarts keeps increasing in price.
The pi was successful in it's goal. It created a huge market of low-priced SBCs. Maybe they are no longer made by them, but the foundation created the market, and manufacturers now create an endless selection of cheap SBCs
Feature creep and bloat are the banes of computing. For instance, years ago, we had perfectly functional word processing, database, and spreadsheet programs. They were easy to use and did anything that 99% of the users wanted. And, because computer systems didn't have many resources, these programs didn't use many. But then companies started adding features -- often features no one wanted -- to their software. The new programs were more complicated to use and used many more computer resources. Great news, if your business is selling new pieces of computer kit. Bad news if you are a consumer. And just as soon as the user managed to learn how to use the new software, the software company came out with a "new and improved" version. This version had MORE features no one wanted, used more computer resources, and had a new and more complicated UI. This cycle continues unabated today.
@@ExplainingComputers I wish I could get a version of WordPerfect from the early days. It had a feature that was really useful -- "reveal codes." With that feature, you could see what characters the program put into your document and change or remove them when you wanted to. Word has never had such a feature, and if it randomly formats your document to suit itself instead of to suit you (as happens all too often), you've got the choice of living with it or spending hours trying to figure out how to fix it. But no one ever said that Microsoft created GOOD software...
And wastes lots of electricity worldwide and makes people buy new updated CPU's even when only 10% of the old one was needed to run stuff that the user clicks on to use
I bought an Orange Pi Zero Plus and an Orange Pi PC back when they were both under 15 €. They are very versatile and useful devices (one is still in everyday use) & exceeded the expectations set by the low price, but nowadays you pay closer to 35 € or 40 € for the same exact models, probably plus shipping and taxes. At that price (for which you can sometimes get a used laptop), their limitations make them of dubious value-might as well pay twice as much for an x86 device with eight times the performance. The upper end of the 2020 SBC market (e.g. RK3399 devices) has been underwhelming in terms of user-ready software development. Three years later, something like the Rock Pi 4 still requires arcane Frankenstein-patching to have working video decoding-in only some circumstances, with some programs, under some desktop environments. It hasn't come down in price either, while X86 mini-PCs have closed the gap. The RK3588 may be powerful enough to bruteforce a lot of these problems with software decoding, but that somewhat defeats the purpose of an efficient and optimised ARM-SoC.
From $35 to $200 it has no longer been an option for my University. We’ve completely dropped the use of RPI and switch to alternative SBCs that are more available and affordable. The support of other SBCs has greatly increased over the years compared to what it used to be. Given that the price and availability made RPI attractive both of those positives are now a thing of the past. When they meet both of those criteria again we may switch back but until then we are done with the RPI Foundation.
That's perfectly understandable. "When they meet both of these criteria again" - that'd be a few weeks ago for the Pi Zero, over the next few weeks for the Pi 3, and a few weeks after that for the Pi 4.
@@senseisecurityschool9337 - You keep posting this. Upton has been saying this since the summer of 2022. It's now the summer of 2023. Tomorrow, tomorrow I love ya tomorrow You're always, a day away
@@originalbluebuddha Since December 2022, he's been saying EXACTLY this - Pi Zero at end of Q1, Pi 3 in Q2, Pi 4 in Q3. Which is exactly what's happened. If you go back & read his post from December 2022, you'll see it exactly matches what has happened.
PS end of Q3 you should see Pi3 & Pi4 stock sitting on the shelf for a while, rather than selling out quickly - that's what he said in 2022 and what he still said last week, because he KNOWS when his shipments from the foundry are due, how long Sony takes to SMD package them, how long it takes to put them in boards and put those boards in boxes ...
@@originalbluebuddha th-cam.com/video/P9vna9jao9I/w-d-xo.html That's 2022. That's what he said in 2022 - and it's exactly what's been happening, exactly when he said it would. So please kindly stop going around lying about people. It's not a good look.
I was really into the Raspberry Pi a few years ago, simply for the challenge of seeing if I could create usable, low cost desktop replacement. Thanks primarily to this channel I succeeded but unfortunately, after a while the shine wore off and I moved on to other challenges. I'm still a fan of SFF computers.
This is perfectly normal, in my opinion. Just experiencing the SBC world is fun. If you don't have an advanced task in mind or don't like messing with boards forever, you can just... move on.
There are two additional categories of SBC consumer: Small business/ISV back office user and the home automation user. The SBC is perfect for the back office for hosting network shares and running scheduled processes (e.g. nightly cron jobs, offsite backups). It's an always on, low wattage device with enough compute to handle the basic tasks demanded of it. A SBC is a very good, fairly inexpensive option for small business owners instead of having to invest in expensive server hardware. They are also good machines for a centralized home automation host. One machine on the network that IoT devices can connect to and are controlled by. People are slowly realizing that cloud managed home automation isn't all that great.
Raspberry Pi was the leader in SBC a couple of years ago. My opinion. Their inability to provide hardware to the maker community offered an opportunity for others to fill the vacuum. I was looking at several projects but the lack of hardware shut them down. While other manufactures are out there I am hesitant to change hardware and generally the cost is greater. I am not sure Rasp Pi can recover the base they once had. I think they are still struggling and are now in the catch up mode against other suppliers. The state of the art has moved on as you so aptly demonstrated. Great presentation.
Their problem was to put all their eggs in one basket. This meant when one of their suppliers was unable to supply them with components they couldn't make the boards. I can only assume it was the Broadcom SOC they couldn't get.
The trading side of RPI hasn't shut down, it is prioritising its industrial customers so THEY.dont have to shut down (or go elsewhere).. I expect that Pi's will become more available and hopefully eliminate the need for people looking for JUST one board ( of anyversion) to have to pay over the odds to a scalper who gad somehow managed to acquire some.
@@6581punk I think it's more of a case of having to make contractual commitments to investors who put up the money for the Pi Foundation in the first place. They probably had no choice but to send what stocks they had to industrial partners first. I guess you'd have to say "that's just Capitalism" and leave it at that.
They have not had a supply chain problem. They have instead focused on industrial applications and staying with the current tech for way too long. I love their ecosystem but the need to start moving on the consumer side or just give up an focus on the industrial side.
I have to say you are definitely the kind of lecturer I wish I had at university. Always engaging, always interesting, always concise... well worth the many hours I've binged so far looking at your previous content!
@@ExplainingComputers Great video, thanks. The first thing I check is free RAM vs. SWAP file loading. As soon as OS starts to swapping onto the ultra-slow SD card the usability goes to zero (and SD lifetime too). So the slow performance is not from bloated SW directly but from swapping instead. Funny thing is that once OS swaps all unimportant data from RAM then it runs fine. So clearly there is a ton of libraries which are almost never used (but Linux wants them in RAM because that's reasonable thing for Linux servers with >100 GB RAM). I totally agree with you about SBC becoming full capable desktop machines. I bought the best ARM CPU SBC available today with 16GB RAM for non-stop home server (Orange Pi 5 with legendary RK3588). Surprisingly it runs fine on 4K LCD and TH-cam fullHD videos are flawless. My old Rapsberry Pi 4 struggled on 4K LCD (FullHD was fine) and overall desktop user experience was very poor (usable but delays everywhere including mouse pointer). This RK3588 run multiple times faster, wow. And it draw 1.7 Watt only at idle (LCD turned off and running Armbian as file server via ETH), and about 8 Watt full loaded 8-cores by benchmark (less than my old RPi4). On the other hand I do not agree about x86 SBC. They will never run file server at 1.7 Watts as my RK3588. x86 is CISC ISA which is outdated and inefficient for modern computing by design. Even CISC x86 runs inside as RISC computer so all this unnecessary CISC-to-RISC translation comes at power and transistor cost (hence inefficient in many areas). The biggest issue for x86 is variable instruction length encoding (instruction can be 1 byte up to 15 byte long). Modern CPU execute 7-10 instructions per clock so CPU needs to load and decode 10 instructions in parallel. ARM and RISC-V as 32-bit encoded fixed length RISC (every instruction is 4 byte long so CPU know exactly where 10th instruction is) runs flawlessly. CISC x86 suffer from polynomial explosion (2nd instruction can start anywhere between 2nd byte and 16th byte, 3rd between 3rd and 31th, 4th ... 4th and 46th, and so on). x86 need sophisticated predictors with millions of transistors for task where RISC needs few thousands transistors and minimal power draw. Am I the only one who thinks it's ridiculous resources wasting on outdated closed ISA? Have you seen the new KING of CPU world? The most powerful CPU today is ...... ARM Cortex X4. Absolutely insane CPU monster and broke several world's records: - Cortex X4 .... is 1st CPU on the world with 3x branch units - Cortex X4 .... is 1st CPU on the world with 8x Integer execution units in scalar core (previous champ was Apple A11 from 2017) - interesting is the fast evolution of Cortex cores each year (X2 had 4xINT+2xBrach, then X3 came with 6xINT+2xBranch, and now X4 with 8xINT+3Branch). - Cortex X4 .... is 1st CPU on the world with decoder capable of 10 instruction per clock (last champ was Apple A14/M1 with 8 instructions/clock) - Cortex X4 .... 13% IPC increase - Cortex X4 .... 10% increase in area only (performance per transistor increase, this you never see in x86 world because all those old ISA workarounds) - Cortex X4 .... 40% less power consumption than X3 at same performance. - Cortex X4 .... is 3rd CPU which is 64-bit only (ARM32-bit decoder removed for the first time in X2, Apple removed 32-bit in 2017 on A11 Monsoon core) - Fun fact: .... AMD Zen 4 .............. has 4x INT + 1xBranch ..... decoder 4 instr/clock (plus huge mOp cache with expensively decoded instr to workaround, but it works only for loops). - Fun fact: .... Intel Raptor Lake .... has 5x INT (shared branch) ..... decoder 6 instr/clock (plus huge mOp cache with expensively decoded instr to workaround, but it works only for loops). Cortex X4 is absolutely state-of-art CPU. X4 server version (Neoverse V2 ) is used by Nvidia in Super Grace 144-core server CPU. I wonder how 60-core Intel Sapphire Rapids and 96-core AMD Epyc will fight X4/V2 core (which has higher IPC, more cores, much better efficiency). Also I hope Jim Keller's Tenstorrent RISC-V CPUs will enter SBC and server market soon.
This chap is friendly too. One of the professors I had at university was a curmudgeon. He failed the whole class and curved the grades. My 17% became a 3.2.
In the context of bridging the digital divide, I hope SBCs stay in an affordable price range. It would be fantastic if more schools and libraries could use Raspberry Pis to teach digital literacy skills.
I would love to see something like a Zero 2 W but with a RK3588S at $30-40 even if it only has 2GB of ram it would be good to bring the desktop computer experience to the masses.
We tried this at my school. Massive failure. Everyone was frustrated because they couldn’t browse the web at a reasonable speed, let alone open Google Docs or LibreOffice in less than 3 minutes. (They were Pi3Bs with 32GB Samsung EVO+ SD cards
The term affordable is entirely relative to inflation. The reality is that in todays economy the various SBC makers cannot afford to produce them for the same prices that they once did.
Today's 32-bit microcontrollers can do that job. You aren't going to run a web browser on them. You can plug in a bunch of sensors and run a web *server* though. You're not going to play streaming HD video through them, but you *can* do serious image processing on them.
Thanks for your assessment and summary of the SBC situation. I must admit, I'm one of the customers who owns several Pi 3 machines and a Pi 4, all of which I used for writing, internet and the Pi 4 is streaming something else right now while I type this comment in your TH-cam channel. I never seem to get around to programming with them, which is sad. Of course, it's easy to wander off when I'm looking for BASIC, FORTRAN, MODULA and various flavors of C to play in . . . sigh. With Pi 5 over the horizon still, I purchased and setup a $79.00 UBUNTU tiny PC last week. By the time I added a mouse, keyboard and monitor, I'd purchased it twice, and I will soon spend a bit more when I put a micro SD into it. Hope it works out well as a writing machine, and maybe I'll have some luck programming with it. Ha.
I'm a RPi enthusiast. I've been collecting them since the RPi 2. Whenever they released a new Pi I would buy one immediately. I'm so glad I did over the years. I have a small collection of them now. I mainly use them for retro gaming and media playback.
Well, I started in 2014 buying one, then in 2015 and following years I bought 3 of the 2015 model, I did not check when a new one came out, just that I hate those mini-mini ports that they think are convenient. The Pi 4 model I bought had USB-C power, nice step, but not everything. The Pi-zero I had setup with Pi-hole and one 2014 model I had Music box running, very unstable by the way and not updated anymore. The burden of open source! All in all, they are fun to toy with, but really stable and reliable? I would say no!, those micro sd cards, if a factory default shows up and by taking one out of the slot and having one piece in your hand and the other is still in the slot? Man, what a crap is that. Instead of making things smaller, they should scale up!
IMO on-board flash storage would be a huge mistake. Right now, you can repurpose a Raspberry Pi simply by changing out the micro-SD card. If you want to experiment with something that could leave your OS unusable, just copy your main card first. It's not just the option to boot from a different drive - it's the ability to remove your normal drive and know that whatever damage your experiment does, even if you manage to erase or corrupt all connected media, your disconnected main OS card is absolutely immune. There's other use cases where that's not an issue, of course, but there's also other SBCs that already provide for those use cases.
Disagree, it needs to have an onboard storage as an option (note the word OPTION). Raspberry PI/whoever could sell a version with M.2 or onboard flash for those who want that option. MicroSD cards are not that sturdy or vibration resistant, short of soldering the card direct to the PCB. I just want an SBC with a secondary drive for data/files (not the OS), without resorting to a janky USB thumb drive.
my odroid sbc has both, sdcard and emmc (removable), rockpi has emmc to usb and emmc to sdcard adaptors , so i can have 3 emmc on the odroid, onboard, via sdcard and via usb, sdcard adaptor opens up options for all sdcard readers, usb adaptor can be expanded a lot (raid?)
I just go the route of a cheap SSD. Since they're so cheap now it's not a bad option. Just create multiple partitions to try other things with. To produce other installs, I create loop devices onto files on my main PC Linux. Just dd a file of the size you expect the partition to use, create a loop device for it, format it, mount and use an install script. Then just send it over to the SBC and use dd to write it to its own partition. Update the bootloader and you're GTG.
A lot of people jumping into the hardware space forget that Raspi or any other ARM solution is the ARM tax that you pay ... I hope we see a lot more RISC-V architectures being manufactured as they'll keep the cost down and also speed up the development of the RISC-V platform as a whole
Thank you for the survey. I use SBCs exclusively as headless controllers for tools (e.g. Octoprint or LinuxCNC) or to run home network services like a media server. I wish SBC manufacturers would make boards that focus on that instead of trying to be desktop computers.
SBCs are kind of dead. If you want something that feels like the SBCs of the old days, you should look at something like an ESP32-WROOM or STM32 development board. Most SBC makers are pushing towards making a microPC instead of staying focused on a computer suited for embedded systems. OTOH, if you like having all this modern power in a tiny package, you're much better off using one of those SFF workstations because they're getting thrown out left and right by schools and businesses, unlike RPis which are expensive and unavailable these days. No real reason to use a 32-bit ARM processor when you could have an x86-64 processor with integrated graphics for the same price.
@@OhHeyTrevorFlowers those esp32-wroom boards I mentioned come with wifi/bluetooth and they're smaller than a credit card if you get a headerless version. might he worth checking out in your use case, the boards I use are called ESP32 Feathers.
My view is that the Pi5 may change things but a lot of the market has switched to either microcontrollers or low end x86 hardware. Maybe the Custard Pi 6B, whatever happened to it? 😅 It certainly doesn't help that a lot of non-raspberry SBCs tend to have poor software support, whereas any old distro can work on x86 hardware. An SBC costing more than $100 has a limited niche audience.
I think what Nvidia are doing and Orange PI are valuable (whatever happened to the Asus Tinkerboard?) but you can't beat the PI for support. For learning and experimentation you'd use a PI or a Pico. For a product I think you should be prepared to solder the chips to your own board.
As someone who's just gotten into this stuff and is looking to get in I'm looking at mostly orange Pi and odroid or just an esp32 for things I can get away with that.
The main desktop computer on my boat is a Pi 4B. My navigation computer is a 3B. A 4B with a couple of SSD or HD drives makes en excellent cheap file server. A Pi Zero W makes a great board for building a wireless security camera. These plain vanilla ARM boards will always be useful and punch well outside their weight class. Keeping the OS mean and lean is key to maintaining useful performance levels. I just watched a video on the new Sigma SBC that retails for $648 with a 500GB SSD. I can get a fairly capable laptop for that! The whole thing is getting out of control. If you make a SBC with true desktop power, it costs as much as a real desktop, and the unit is too big for a lot of maker projects. No real advantage. Now that the supply problems are sorted out, the Raspberry Pi is a great option for getting a lot of work done for very little money, with a device that is smaller than a pack of cigarettes. But an OS that is too busy and too resource hungry, cripples it.
My only gripe with the Pi is that it is not fast enough. Totally agree with you about software bloating, way back I squeezed my code into 2k EPROM's !!!
First of all thanks for the benefit of your brilliant insight to make a very interesting and informative video. I was particularly drawn to the reference to “Has the Pi bubble burst?” in the title. Well I hate to say it but to be honest I think it’s about time that it did. As a Raspberry Pi fan from day one I’ve purchased probably every variant produced to this day (on the odd occasions that they were actually in stock of course). I’ve always been frustrated by the virtually always “out of stock” and even in some cases the ridiculous rationing of the main products. Imagine any other business model that says we’ll only allow you to buy our products if you’ve been a very good boy and only ordered one. And of course pay the full three quid postage cost for each and every tiny item. I’ve sometimes wondered why I even bother to carry on trying to buy them when you look at much more capable and cheaper competition that are always available and easy to buy but it always comes back to “well yeah but the Pi has such a great “ecosystem” and is much better supported by the “community” etc. etc.” Indeed this is true but we should all be very aware that this “ecosystem” and the support it provides has been built not by the RP Foundation themselves but by the users - the so called “makers”. It’s this group of people that has made the Pi the great success that it is despite the Raspberry Pi foundation failing in many respects to support them not least in being able to supply the damn things. But despite this bizarre business model we still rave abut the products and turn a blind eye to superior competitive products. I’ve been guilty of it myself saying stuff like “oh well, they may be impossible to get hold of and the SD card socket is awful and it still only has a 32bit OS and no on/off switch etc. etc. etc. but they are the best supported and we still love them” What finished me however is that when things got really bad and COVID followed by chip shortages and a plethora of other supply problems came along who did they support? The educational and maker communities, for whom the products were supposed to be for in the first place, the people that created their wonderful “ecosystem” and made them what they are? NO. The lovely Eben Upton threw them all under the proverbial bus and instead supported his “corporate” customers, i.e. lazy cheapskate companies profiteering by charging for perceived high value goods incorporating low-cost off-the-shelf hobby SBCs meant for educational use inside them rather than doing their own R&D. And leaving his loyal core customers having to sit through lockdown either with nothing or at the mercy of eBay scalpers cleaning up by selling the surplus stock at insanely inflated prices. Good call Eben! Now nearly three years later when they are finally beginning to have some stock to spare they come back to us and say “hey guys you’re really so lucky now because we’ve decided that we can finally let you have some of our stuff again”. Well I say you can shove it up your proverbial Eben. I’ve had no choice but to move on and now I have no reason to come back again. What goes around comes around. How about if we have a go at actually bursting that bubble and divert our efforts into creating an Orange Pi “ecosystem” instead?
Since my interest in SBCs was basically inspired by this channel, I've done many projects and expanded my modern computing knowledge in directions I'd never imagined it heading. I'm appreciative for the gains I've made thanks to the community of enthusiasts. I was lucky to get some cheap pis when the market was good and they've been wonderful companions in my ventures. Sorry we couldn't all enjoy the bounty. Still dislike TH-cam comments like the "pi is dead" "we've moved on" etc. A web search shows me a University of Cambridge stat from 1 year ago - 40 million pis sold. Seems like many of us have working boards and will continue to bask in our good fortune. "Long live the (inexpensive, functional) SBC!"
I think lots of people started with an RPi and while learning to make things with those they also learned that it was overkill to use an SBC for those small projects and they moved to microcontrollers (STM, ESP32, ... etc). Today we also see people buying a stronger SBC to replace their desktop. When you see what an SBC like the recent LattePanda Sigma is capable ... If those "high-end" SBCs continue evolving like that it might be the end for the desktop computer as we known it. Looks like the SBC market is splitting and the desktop users will get their own range of "high-end" SBCs.
This is exactly the path I've taken. As much as I would love to buy something like a Raspberry Pi 400, it makes more sense to buy a refurb Mini PC and one or more Pi Pico boards. The Mini PC is more versatile (and in some cases expandable), and the Pi Pico can go off and do its own thing once programmed!
This was a great insight into the consumer SBC market! Nobody could do it better than you, really. SBCs have certainly changed in the past decade, where we went from a low-cost SBC to learn coding and do projects on to SBCs that have about as much power as a typical desktop and exciting new things like a RISC-V processor. Keep that Pi bubble floating! (well, at least for as long as possible...)
I felt it in my bones that this would be a possible future, in the short term rather than the long term. It was fun while it lasted. *slides in an i5 Mini PC*. Thank you as ever Chris.
Thank you for another fine video. I've tried using Raspberry Pi SBCs for 3 different purposes. 1 Temperature logging to the internet. However, I replaced that with an Arduino type board a few years ago and the Arduino is far superior for that application. The Arduino was more reliable, with lower power consumption and better software options. 2 Motion Eye Pi Security cameras. The Pi is very good for this. 3 A lightweight mini PC for using when travelling. The idea was to use a small size bluetooth keyboard and hook it up to LCD TVs in hotels. With the advantage of having a package small enough to fit in any hotel room safe. However, it wasn't really fit for such a purpose. (i) The wifi was weak, so where my Android devices worked, in some cases the Pi could not get a signal (ii) The bluetooth keyboard was laggy and kept dropping out.
One of the main reasons other than price that the Raspberry PI did so well was the community and forums around it that supported makers etc. This is just not available to any great extent with the other Arm based SBC's. Maybe it will come, but getting stuck without that knowlagable support can be a killer. I think we will see more diversity in the computer resources being deployed with microcontrollers at the edge and sbc and micro pc's as servers and data processing nodes
Yup, it was the community and the support that made it such a smashing success. Similar with the Arduino. We had microcontrollers before the Arduino, e.g. the PIC series was relatively popular. But the Arduino really exploded the market, in great part because they made microcontrollers easier to use through great documentation, and fostering a community around the platform.
We used RPi’s as a mix of all three categories. Regardless of all the naysayers, it is and was a solid piece of gear for its intended purpose. At first, my son used it to learn about the computing world, coding, and maker projects. I also used it for some of the some purposes but migrated toward some home automation and low-cost video production. Due to the current price point, I’ll use what my money can buy. Thank you, Chris! Cheers
Greetings Chris on this beautiful (at least here) Sunday. I put myself firmly in the enthusiast camp. Whether I need all 50-60 SBCs that I own remains a mystery, especially to the missus. The SBC market has changed drastically changed since the Pi made its debut and truly launched the form factor into the public eye. The original idea of getting more people, especially chldren into STEM has seemingly been mostly lost. Supply and demand has had a huge effect on the cost of SBCs along with the demand for more power, demand from users as well as demand from the heavier software requirements aka bloat has raised the need and therefore cost of some of these boards. The Jetsons have that higher cost but you do seem to get more with added GPU capability. LattePanda has also been a bit of a standout with their x86 based boards which comes along with the higher price tags. Their current Sigma release has finally put a line in the sand for me because I cannot justify that cost for my use. RISC-V, while still in its infancy, should eventually change the landscape of SBCs in my opinion. At least, being open, I hope so. Well that's it for another ramble. Please stay well and keep the excellent content coming.
I thought of you when I wrote that label! :) I'm very interested to hear that your reaction to the Sigmas was the same as mine. And I also agree that RISC-V will offer new enthusiast opportunities . . .
"I put myself firmly in the enthusiast camp. Whether I need all 50-60 SBCs that I own remains a mystery, especially to the missus." I can relate to that. I can relate to that, yes.
I switched some projects to repurposed min-PCs away from Raspberry Pis as they became harder to get, eg Octoprint for 3D Printing and agree with your analysis. As a maker I am switching to the industrial pi hardware using the CM4 mezzanine cards and IO boards systems for experiments and prototyping with the intention of getting a ULX4S FPGA card when they become available. But I am still benefiting from the vast Raspberry Pi ecosystem of products, software and information which I think will maintain it's advantage over the competition that essentially relies solely on offering a linux capability without putting in the effort to offer further value to the community.
I expect the SBC Market to become more fragmented as different manufacturers specialize in different areas, in response to enchrochement from Mini-PCs (especially fanless) capturing the more general Users and the MCUs capturing more of the Maker market. The day may come when every SBC comes with a DIN Rail connector.
FWIW: I think you're probably right, but it's also unfortunate. If one considers the reason for the development of the Raspberry Pi, it's a cheap way to get people in to 'coding' and 'making' with a supportive community. In many ways we are sadly moving toward the situation that existed before - a fragmented space of options, none of which can boast the RPi's focus on leaning. The result will surely be that once again the learning curve for those interested will get ever steeper and many will be put off. As an aside I'd love to see the Pi foundation do a deal with a big mobile phone manufacturer. Not to make Pi's, but to make *good* phones designed to be gutted for the motherboard and other parts at the end of it's life on to which the Raspberry Pi's OS's could be run. Consider: -What happens to all those old mobile phones today? Landfill for the most part. -Most older mobile phones still have more power and capabilities built in than most SBC's, primarily because of economies of scale. -Recycling your old phone and possibly the screen to use as an SBC is 'free' - selling dismantled phones for 'SBC' use would become a thing. -An 'SBC' that came out of a phone is an 'SBC' that would likely be very easy to integrate in to other projects.
You Sir, are right on the money ! A phone designed from the outset to be re used as a SBC is a wonderful idea and just the kind of green approach that should be adopted ASAP. A hat tip from the northern UK to you ! 😃
My first computer was h89 from heath kit. I ran hdos and later cp/m. It was a slow move up to x86 machine from 8080 machine. But I have enjoyed the ride. I get a big kick out of my RaspberryPi machines 4b and 3b. I use them with my trains and DCC++EX. But just recently my old 86 computer died. And I was directed to a mini pc by my mentor Chris and I have been a very happy person. Enjoy your channel your the wizard of acknowledge. Thanks Ed
The SBC complexity and power has creeped up over the past several years while mini PC's including USB Stick versions have shrunk in size and cost. The decision to select one or the other has become more difficult. The SBC is still a great tool for teaching computer technology even in its simplest forms.
As for SBCs, I transitioned to the Orange Pi, and Banana Pi platforms. I turn them into remote control cars, drones, NAS servers, retro gaming machines, and as microcontrollers. I did successfully turn a Orange Pi 4 into a mini PC, it could stream videos, browse the internet, online shopping, email, word processing, retro gaming, streaming video, etc. As for emulation, I use a SFF computer hooked up to my television. SFF computers has more power than any SBC, so it is the best choice for retro gaming up to a Nintendo GameCube.
Raspberry Pi have made it very clear what they think of their maker and enthusiast market in recent times. Lets hope that same market lets Raspberry Pi know what it thinks of them!
You totally nailed it classifying the type of people buying an SBC. Pi isn't going anywhere. There are plenty of options that are technically better. The Nvidia Jetson is a good example, however. Economics is the one driving force for everything you'll ever want to do. The Pi might not be the toughest dog on the blog, but when the Pi foundation sneezes, everyone else catches a cold. I think the Pi Pico with wifi is going to surpass all the SBC as far as being the most popular to use in projects. One of the biggest weaknesses in using a SBC in a control application is the additional layer of an OS. This adds another layer of complexity to the design and increases the chance of having a fault. Running a single or multiple application that directly address the hardware, and using a technique called supper looping will probably surpass applications requiring a multilayered programing environment. The pi pico can give you the best of both worlds within its limitations. I think the regular raspberry pi is going to be good option for at least the next 10 years, if the company remains stable. As Windows can run on different hardware, Raspbian should be Multiplatform between clone manufacturers. That probably won't happen because once a better platform is out there people will dump the Pi hardware and move to new hardware. Imagine what it would happen if the Orange Pi, or banana Pi foundation built a much better SBC that was cheaper, but had the ability to run the Raspberry Pi OS.
A good overview of the situation. I own a Pi4 that I managed to buy just before they became unavailable. I like the ease of swapping SD cards to try different retro gaming images, and Pimiga3, but I will admit to not using it anywhere near as much as I should. I have not considered using it as a desktop replacement, and I haven’t used it to control external devices through the GPIO. I also own an original Pi that I used to suck the EEPROM contents from an Original Xbox. As for the future, I think that if the Pi situation gets resolved as promised then the product will continue to be a popular choice for a lot of people due to the community support.
Picked up an AMD SBC, with 8gb RAM, from Amazon for $110. It works great and I use it daily. Picked up another, upgraded to 16gb RAM and use it for the controller of my Fiber laser. Works great! Finally picked up a third, 4gb version for about $90 and it also works perfectly for what I need it. Streaming music from my server to the sound system in my shop. About to buy another to put a workstation near my milling machine so I can alter code at the machine instead of going all the way back to my desk. Started programming micro-controllers back in the mid 90j's with the Microchip 16F84, now i'm using the ESP32-S3 and that thing rocks. About to release my first commercial product based on the ESP32. Wish me luck. While I have my share of Raspberry Pi, I don't use any at the moment other than a Pi Zero for running PiHole.with 8gb RAM
You've dissected the grave situation in detail. A lot of people are resorting to buying second-hand computers, rejected by government offices after 4 years of use. Consequently, the cost of building a refurbished PC has increased astronomically in recent months. I think people will use a computer in a maker project in extreme cases, where there are no other options like an MCU to keep the cost low. Some projects require a regular computer, such as smart agriculture, traffic monitoring on private roads etc. On top of that everything needs some sort of AI which requires decent hardware. The supply chain problem has created a tremendous amount of turmoil in the consumer PC market here. Things are going to get worse! Computers will no longer be as affordable as they were 5 years ago. You got it right. Anyways, Thanks for the insight!
@gr-os4gd those second-hand PCs are not as cheap as they were in 2021. In fact, these are not cheap anymore. Everybody is looking for some old computer. Yes, recycling old electronic equipment is an excellent way of saving money and it's good for the environment.
Great video, as always. Even today the Raspberry Pi is still the first thing that comes to most peoples minds when anybody mentions SBCs, despite there not having been any new offerings from the Raspberry Pi foundation for years and there existing many more powerful alternatives on the market. I was quite surprised that the latest Orange Pi boards haven't become the present day rage given how much more powerful and affordable they are. But then again I guess that the community support that Raspberry Pi has built up over the years is an asset that competitors have no equivalent of.
Another excellent overview; really enjoyed this. Though you touched on it, one branch of SBC users is: the low-power, portable 'hardcore'. For amateur radio, quite a big market for SBCs as we also tend to be into electronics in general, portability and 5V USB, low current-drain are two critical aspects. As the distros become more demanding, so the tendency towards more power-hungry design tends to follow. I suppose the use of 12V for transceivers means we can accommodate greater current draw, though it is very undesirable, given power is always in short supply in the field.
With regards to the categories of "Consumer SBC Buyers" (starting around 1:33 in the Price Changes segment on the video), I suppose I'd consider myself to be in the 2nd category, it'd be nice to have a small and very affordable computing device that can be used for typical office work. As time goes by, Raspberry Pi might have its competitors, and while it is somewhat unfortunate that 32-bit support is being phased out (especially when it might increase the cost of newer SBCs) but we have to appreciate what the Raspberry Pi has done, in terms of laying the foundations for Single Board Computers in this day of age.
I am getting old but love everything about computers and have been a subscriber for a long time exclent quality video and perfect English easy to understand.. Saying that I i have been bying second hand hardware i waa in London last week bough a Intel cor i5 1151 for under 20 pounds will be trying some Linux software.. Wish you well
I made my own SBC (RK3566) and received it a month ago, it has been very fun. I developed the camera driver (Sony IMX586), and recompiled the Kernel. It was fabricated and assembled by PCBWay. This level of fun and satisfaction couldn't possibly be replicated by buying a Pi As a matter of fact I wouldn't waste one second of my time developing a project for a Pi, if I can't make a custom board with it's SoC, it's worth nothing to me
Raspberry Pi did something amazing, by literally building the market for low cost SBC's. Now the market has grown, changed and evolved - including the entrance of RISC-V SBCs. R-pi is now a victim of its own success - especially with supplying boards to industry in bulk. It could have a party to play going forward, but the market has evolved, so R-pi must recreate itself as well - much like another tech company achieving their initial vision/goals.
Good and even handed discussion as always but I’d add a few points: -Pre chip shortage the SBC particularly the Pi4 and Pi0v2 were very good at doing one given thing exceedingly well, be it a print server, SDR rig,Octopi server, KODI box, signage, emulation station. In such configurations they could be easily paired down to a minimal linux configuration and work well. -The Pi4 and Pi0v2 in particular were excellent for having an excellent support and knowledge base. The base units were impulse buy pricing, and you could easily bring up many of the basic use cases in half an hour or so if you had an SD image. Since every Pi was nearly identical you could replicate and disseminate implementations easily. You could just as easily repurpose a Pi4 for any other app if one didn’t pan out. -The low cost of base units avoided the “sunk cost fallacy” in that you could fail quickly and write off the cost of a Pi4 and more so a Pi0. Also this was a boon for STEM in that there was much less anxiety that you would break an expensive computer. -The path forward may well be the compute module. If you can pick and choose hardware options on a carrier board as well as paired down linux a "CM5" could be a world beater.
I played with Pi's back when you could get one for 30 to 50 bucks, but now the prices have made me look at other sources for projects. If I've got to spend 100 bucks or more then why not look at and all in one PC and better if I can get a renewed one for maybe a little less.
My reason for staying with Raspberry Pi is the software and software support that comes with the Pi. The forum community also keeps me up to date with Pi related things.
Microcontrollers will certainly be useful for projects requiring minimal computing power. Mini PCs can take over the expensive SBCs because any computer can certainly be equipped with GPIO as we have seen in your video. SBCs need to have a reasonable price for surviving in the consumer market as the alternate methods are cheap and still work okay.
You make a great point about heavy distros and prices. With the scalpers having driven the prices up, it's getting as cheap or cheaper to get a mini-PC, or repurpose a five-year-old Dell from eBay.
Sometimes I wonder if we're going to ultimately find ourselves in a cycle of making minimalist devices like this, the software ecosystem gets bloated, until people get sick of it and go start over with some other minimalist platform. All other issues aside, I feel like it'd be less messy if the SBC world basically fractures in two, one for people who want to use it for a mini PC and another for people who want a minimalist tinkering environment. I don't really like NUCs and whatnot but I wonder if they'll eventually end up taking over that niche.
There will always be a less expensive, better performance per watt ARM chip than x86 can deliver simply because of the instruction set difference. I see x86 taking over the mid to high end of the market, with ARM continuing strong where having the most lightweight hardware possible is a valuable advantage.
@@editingsecrets I think a more traditional microcontroller is a better option at the low end. An SBC is wasted when it sits all day doing nothing but blink lights and read switches, which is what most Arduino/RPi projects amount to. Even an ESP32 is overkill for most of the stuff that people use their SBCs for, excluding the people who want to use them as little desktop PCs.
I do think that once they have the stock of parts again, RPi will take over the maker community SBC space again. It's not about having the most impressive hardware. Other companies have made much more powerful SBCs in the same form factor. It's the software support that RPi offers that's really the winner. I feel like there's been a three year window of opportunity for a different SBC maker to dominate the market, and all they had to do was offer a stable, well supported OS and drivers like RPi does. It's a big task, for sure, but absolutely achievable. None of them really did that though, and I can't understand why. A company like Orange or Radxa could have pulled the rug out from under RPi. Not that I wish a bad outcome for the RPi Foundation, I think they've done amazing work for the maker community. I just am surprised that this didn't happen.
To be fair, I think it didn't happen for the same reason that Pi costs stayed up. Pi costs staid high because people were already in the Pi ecosystem, thus demand for Pis remained high even though they were exorbitantly over priced. It's the same reason the Microsoft window's environment isn't going any place regardless of how many people grouse about it. Or the fact that it outright spies on its users. This wasn't helped by the fact that the Pi has been seeing a lot of use as a ready made 'brain' in commercial/industrial products. Just toss into a hardened case and plug it into your product with proprietary software as a ready made solution. For a lot of industrial products, the added price was peanuts in exchange for a well supported code base. It's a chicken or egg situation. Yes, someone could have benefited from switching to the cheaper alternative, but the first person to defect would suffer from poor software support and have to start from scratch. And the savings often just wasn't worth that.
Yeah. Turns out everyone can slap together a working Single Board computer (just a matter of slapping some reference designs together on a single PCB) and start selling them. But when it comes to software? most cop out and expect the users to sort it out for them. proclaiming it as "Community driven".
I have my doubts Raspberry Pi will ever get supply back up to demand again. They've gotten in bed with the wrong suppliers and they're making too much bank from their industry customers (who are probably scalping a lot of the supply they're buying up).
I can confirm for our market in Poland. Recent refurbished offers included: - Pentium J5005 8/16GB RAM/eMMC Dell Wyse 5070 for 70$ - 10W, passively cooled, with M.2 SATA for storage - Ryzen 2300U 8/256GB RAM/SSD Thinkpad ultrabook for 100$ (with poor HD screen, but hey, that's 100$ for full fairly modern laptop) You really have to want ARM SBCs design wins (even lower power consumption, though that Wyse is pretty great; GPIO, form factor) to pick them over x86 HW like that.
Hello, Chris my first Raspberry Pi was Raspberry Pi 2 Model B V1.1 2014, and as I had mentioned before I now have the Raspberry Pi 4 4GB model and there is a definite difference between the two. Thanks for another great video and the future of SBCs. I also have a Dell Optiplex 790 SFF Desktop. I upgraded to 1TB HD and 16GB of RAM. as my main computer. Also, have a Lenovo ThinkCentre M92P purchased from e Bay for $85.00 barebone, had to purchase the power supply, WiFi Dongle, and wireless keyboard and mouse. 240GB SSD. 4GB RAM Windows 10 Home. On another note, I was using my Raspberry Pi to learn Python to learn to code.
Excellent general synopsis, as always, Chris. I'd perhaps consider a "spanner in the works", at least in terms of current suppliers, as being that soon many needs won't need their products. In that, I expect that in the next few years, printing one's own SBC will become an additional Maker's choice in some sense. I believe we're currently at the "not quite yet" period (for the _average_ maker's home setup) of being able to download a model of _some open source_ board, like the Blueberry Pi, or suchlike, and have consumer 3d printers which are tuned and capable of printing copper traces on a previously printed plastic base, with a design suitable for home flow soldering ovens. I say we're not quite there yet, as I'm personally no where near as experienced an electronics engineer or maker as the majority of those who might categorise themselves as such, as my history is mostly in software. But I still think that I've _probably just about_ got the skills to do this, given some investment in current consumer equipment, and I think that feeling improves each year that goes by, simply from new, and usually open source, community advancements in 3d printing methods. Firstly, often as some print-it-yourself printer mod to enable some bleeding edge feature, and then the best of these hacks and mods tend to get brought into new printer models. If you couple RISC-V, with some consumer printer capable of multi-sided board fabrication, a software package to build the CAD/CAM using a set of board-requirement options, and reflow oven post-processing to make a board with "just the connections I need" (tm) - then I think there's a branch of the future where entry level SBCs will be the "3d printed door latch because my local hardware store doesn't sell one" of the coming years. Or at least I hope so :)
With all the opened source hardware out there, if you were to produce something you’d like based on those open source sbc designs, I don’t think you’d ever have to bother worrying about future revenue streams , ever again.
First, last, and always, I'm all about embedded controller header boards with some flavor of AVR, ARM Cortex M, Propeller, PIC, RISC V, LX6, etc, onboard. Thanks to the open source community, one can get a great C cross-compiler and IDE for most of them. The RPi was miraculous to behold, but was something completely different from what I was used to. It was never big or fast enough to be a real desktop computer nor did it provide the full I/O access and throughput I wanted, needed, and expected from an embedded controller. For me, the RPi was primarily a great opportunity to use Octave and Mathematica, and that was worth a lot! 🙂
How cool to see Chris sitting at a desk rather than just his hands! All tis is just the evolution of a system - I've been in this business since 4-bit (yes FOUR) and it's amazing how it swings around and changes. Evolution in nature takes millions of years, we see it here in a matter of months.
👍 Very interesting. I ended up in your 3rd category with my last purchase. I noticed SBCs were becoming more powerful, more expensive and paradoxically less capable with recent software updates--very puzzling. I think the software bloat issue is good topic by itself. Many simple task apps have become massive; you wonder what these apps are ACTUALLY doing with most of their cpu cycles.
Very insightful summary of the situation. In my opinion, an interesting option that few people know about is the possibility of using an Android TV box as an alternative to an SBC. For example, TV set-top boxes based on the Amlogic S905X3 can be easily converted into a general-purpose SBC running the Armbian Linux distribution.
Now that is an idea that i have been mulling for a year or two now, as i have a couple of S905's with 2Gb/8Gb i got for $15 and thought i could reformat them as SBC's with Linux instead of Android. However, i just recently read that they may not be able to be wiped completely clean and could still continue to house the original malware that they came with. I am curious.....what do you think?
We are thinking of using a compute module on our CubeSat, be-spoke board obviously. I was at our parent company last week, their labs were full of PI’s. The thinking is, the PI platform and software has such a large user base it’s reliable and stable. Just need to consider shielding and rad hardening issues, but some chunky aluminium cases may do the job.
I think your predictions are spot on. Even with the current supply situation, Raspberry Pis still outsell the other SBCs that are more powerful but much more expensive. I think for SBCs to compete with Mini PCs, ease of use needs to improve. I was recently asked to recommend a low cost computer for someone, and I recommended an Intel N95 based Mini PC. If an RK3588 SBC was easier to use, I probably would have recommended that. But for a novice user that doesn't want to climb a steep learning curve, any of the RK3588 boards would be way too difficult, and would probably turn into tech support time for someone like me. So for that market segment, a Mini PC that is preloaded with WIndows is still the most practical solution. If Arm based boards get much easier to use they might start taking a larger (but still minority) portion of the desktop space.
Thanks for your valuable insights Chris. I have a range of SBCs and mini PCs which I use mainly for computational tasks like file hashing and audio and video encoding, which keeps my desktop free for daily use. I very rarely use a GUI so use BASH instead. I'm more interested in computational power comparisons between SBCs and Mini PCs e.g. your $100 Mini PC vs the Orange Pi 5. I'd very much appreciate if you could cover this at some point.
I also want to see that comparison. I think if you need gpio or spi/i2c bus then go to SBC or choose MCU. If not then Mini PC or put it in container or just run in server. In my case I use RPi4 for ruuvi tags but now also my main computer has BT so I can also run listening scripts there but I still like to run those outside my main rig. Other stuff is using AVR, ESP8266, ESP32 or MSP430.
Yes i completely agree. I have moatly been involved in the SBC space for its SFF and low power consumption hardware. I don't think I'm alone in that. That said I just bought a very powerful (yet relatively low power consumption) Intel I9 NUC with 64GB of ram and 2TB SSD for around $1000 which fits in my back pocket. That said i really support the Pine64 and Raspberry Pi, I think there should be a place in the workd for them so i will continue to support projects like these because i believe in them.
Amazing video with a great insight into the SBC market. Raspberry Pi dominated the SBC marketplace and was originally meant for hobbyists and low-cost gateway into coding. Seems that isn't simply the case nowadays, after the pandemic Raspberry Pi seems more inclined towards manufacturers and not hobbyists. Not sure whether the Pi will be in the same position as it did 3-4 years ago given the current circumstances. The worst thing after all is that other SBCs don't have a good software support or community and will take sometime to come near Raspberry Pi. Coming to software, like we've always seen before, bulky software absolutely destroys a perfectly capable hardware.
Your last sentence reminded me of a decades-old saying "What Grove giveth, Gates taketh away." Other players are now in those positions, but the game remains the same.
I think you are spot on with your assessment. I had wondered why my Pi 3 units were preforming so dismal compared to how they used to. The pi 1 models are a hidden gem, without modification (LED removeal etc.) it idles at 80ma ! I too am moving to microcontrollers simply for the ability to just cut power, with zero data loss, that fact alone makes me choose a microcontroller over a raspberry pi zero in which bloat is making it darn near useless in most situations. Many Pi zero wireless and bluetooth capabilities can be replaced by a simple esp32. Bloat, cost, the long duration of unavailability have pretty much killed the Raspberry Pi ecosystem for me. These SBC are widely available now and stock is up (July 2023) but I fear they won't sell very many. The combinations of Raspberry Pi burnout, the thrill of finding one being available being quenched (the hunt is over), the extremely unlikely but possible release of a raspberry pi 5, and people moving on to less bloated, lower power, and more stable product unavailability will put the raspberry pi foundation in a situation where they spent a lot of money to get stock, only to find that no one even really wants them.
I had once used a Pi4 as a desktop, but after 2 years of trying to get as much performance as possible out of it, the limitations with TH-cam playback, etc, were just too much to make it workable any more. Now I'm using a X86 SFF desktop, but I do still use my trusty old Pi. It's now being used as my digital desk clock, a task that it performs perfectly.
You can make it workable but you have to invest time and effort learning to build a more efficient Linux on it - I do it that way myself using Gentoo Linux. Raspberry Pi OS is a great all-rounder OS but on a board of that size and limited power, every CPU cycle and byte of RAM that you can claw back is important - and therefore customising it more can achieve a lot. And that's just a statement of fact, by the way - it's not my opinion of what I think you or anyone else should be doing to follow my example. I am simple giving you a solution to the problem you state that there is with the Pi.
Firefox wants to implement V4L2 M2M for hardware video decoding for the Pi (and possibly other SBCs?). But I have moved on to the Rockchip RK3588 and 1080p is no problem, even just with software decoding.
I have to say that your last video on buying refurb business systems pushed me over the edge to buy a system just like I use at work. It's a wonderful system with twice the RAM as my work system and it runs Windows 10 just fine. It's my son's gaming system now. All for $110.
One other thing that surely contributes to make SBCs less interesting is that ESPx microcontrollers are getting more powerful and easier to use, and their low power needs make them a better fit for IoT home projects. The other side is that the new cool thing for tinker with seem to be Large Language Models. So yes, the space for SBCs has been squeezed a lot.
I think the biggest shift we’re going to continue seeing is away from the rpi as an appliance where the user is expected to just image an sd card, pop it in, and go, and back to projects creating installers/mgmt utilities that just target Linux with the expectation that users install at that level on top of whatever the base distro for whatever SBC they have is. Which to me is a positive both because Linux knowledge becomes more prevalent and because the supply constraints on a single type of SBC will be less impactful in the future.
The chip shortage has come to an end, and Raspberry Pi will soon have enough boards again. Several resellers already have sufficient stock, leading to a gradual decrease in prices. It's important to note that we've experienced significant inflation of 5-10% in the past years, resulting in slightly higher prices across the board. I anticipate that the new prices for the Raspberry Pi 4 will settle around $50 in the coming weeks or months.
And on what basis or authority are you making these claims? I am not saying I don't believe you, I'd just like to understand where your information comes from.
@@terrydaktyllus1320 it's all available out there ... just a simple Google will help you. Inflation has been steadily rising over the past 3yrs due to the effects of the Rona lockdowns all around the world. The fed has been increasing rates to curb it and that has not been quite great for the economy and that doesn't affect only the US, it has ripple effects all throughout out.
@@weiSane You clearly misunderstand me here - I don't do "homework" to make you or anyone else look right. If you have evidence to back up your point then you present it when asked - because I do not have to take on trust anything you tell me. And the opposite is true also. You'll forgive me for ignoring most of your "economics lecture" given that it is entirely unnecessary as something that adds to the knowledge that I currently have on the topic. Did the OP appoint you as their official spokesperson, by the way? I do want to believe what they are telling me but only they can provide the evidence to back up their own claims.
It’s important not to lose the biggest picture here - the OG RPi created this whole space of $5-100 SBCs and there is now a perfect choice at all price points for all applications. That’ll never change so Ebun’s mission is complete. Just because the average price of this market is now eating into the low-cost x86 PC has no effect on the affordability of access at the bottom - quite the opposite as Wi-Fi is now on the $10 SBCs. Love this channel, keep it up
Thanks for a great video. I've always thought SBCs were a really nice idea, but I've also always been able to talk myself out of buying one because I had no particular use for one. I think most people don't realize that for the most common uses - browsing, email, video, music, etc. - a used office computer will do the trick just fine at a much lower cost.
I do everything with scrapped machines- but there's always the increased risk of failure and relatively low power efficiency of older gear you have to plan around. (backups and standbys, etc) I'd have no qualms about jumping back into text-only office apps and doing everything on an SFF again when this current sub-$100 laptop craps out the rest of the way- I can run a separate STB on the TV for any video I need (I don't do ANY editing anymore) and use less power AND BE MORE PRODUCTIVE with multiple devices. That probably has a larger carbon footprint for manufacturing and operation of all this junk I much prefer to call "my toys", except the only actually new COTS are the SFF and STB...I think I'll have to do a deeper dive on real total costs, but I'm pretty good at guessing after 60 years in tech, and my best wild guess right now is I'm a little less burdensome to the planet than with a spiffy new laptop I can't afford.
For me, it is not so much about whether the hardware is mounted upon a single board or not; or whether it is a microcontroller or computer; it is about the technology being accessible. When TI's MSP430 demo boards first came out, I remember struggling with 900 page PDF files of the smallest readable print, written for embedded engineers. At most I managed to do is blink an LED after a month of effort. The 2 great revolutions came with Arduino and Raspberry Pi in the form of super supportive communities geared to amateurs. Being able to 'creatively play' with the tech massively amplified my learning. Tutorials written for laypeople. Your channel, the 'Swiss-Guy' and Adafruit are massive contributors to my pursuit of technical growth. I hope there will always be a market for inexpensive electronic learning systems. Thanks!
To think what was achieved on the systems in the 1980's and 1990's.... say the Amiga and the Atari ST... with 512k or 1MB/2MB Ram (if you were posh) - and the levels at which the systems were pushed. Every last bit was squeezed out of those machines - no bloat then, I miss that. I don't know, I loved the video Chris, but crikey, it left me feeling rather depressed at the state of things... is it a case of 'get off my lawn'? possibly....
Even if it is a misnomer , people use "SBC" _(perhaps they mean small-board-computer instead of soley single-board-computer)_ for those that are upgradeable such as the compute modules. Upgradeable RAM makes sense because you can replace the defective memory but also increase RAM size. That saves money since it prolongs life. Pi prices did fluctuate and anyway, a HBA or other PCI-e expansion could add cost and yet people do it. The 100Mbps mode for ethernet is worth keeping (even if a port for Gigabit rj45 exists too) because the 100Mbps driver is tiny. There are a great may use cases for that including being able to boot from various sources. This explainingcomputers is a well done channel and the enthusiasm is a factor that makes it even better. My comment has no hate in it and I do no harm. I am not appalled or afraid, boasting or envying or complaining... Just saying. Psalms23: Giving thanks and praise to the Lord and peace and love. Also, I'd say Matthew6.
I think you nailed it. I started to use esp32 microcontrollers for sensors and led strips/strings as they are cheap and easy to acquire. I still have some pis running some network services. I guess i am mostly maker and part enthusiast. I also agree that if an sbc is going to cost about 200$ i might as well buy a small form factor pc. With regards to a raspberry pi 5, maybe a cpu update, drop 1 and 2GB just focus on 4 & 8GB units. Maybe a single full size hdmi connector and a focus to keep the cist down. The pi wont compete well against small form factor PCs. I would rather resolve the pi 4 supply issue than get a pi 5.
Great summary and opinio. I've had a number of RPi's over the years and have now settled on using an HP Elite Mini 800 UFF machine running Proxmox. This has given me a significant performance boost, allowed me to run multiple Virtual Machines, Linux Containers, Docker Containers all in a single computer. Busy setting this up as a cluster for failover, etc. These SFF/UFF machines can be purchased for very reasonable prices and fill the needs of multiple RPi's. The power consumption tradeoff is not that big, considering performance per watt as Jeff Geerling points to as a consideration. In addition, these are available in large numbers and not subject to the current RPi supply chain 'tax' or scalping. There are cases for the more industrial focussed use of RPi's as you mention and I can see a use in specific Maker projects where a small board makes more sense (e.g. MagicMirror, or DIN Rail mounted such as Kunbus RevPi)
4:04 there are also a large number of used x86 thin clients available, most starting at just about $25. They are usually much faster than ARM SBCs and a lot more flexible (choice of OS, upgradeable, PCIe and SATA, ect.)
A good option is the Fujitsu Futro S920 that comes with a dual core AMD SOC (1.5.GHz, I think) and 4GB RAM by default - it will take 8 GB DDR3 SODIMM RAM. It also has an additional PCI-E slot that could take a low profile GPU, though I put an Ethernet card in mine to build a pfSense router with it.
Had been trying to get one of the these long long time. Pointless to bother looking, or it was back when I cared enough to bother looking. The whole Pi situation is far more trouble than a Pi is worth or the solution it could ever hope to provide. I'll spend my money on something avail and fight thru the lack of support. I can learn whatever other SBCs with better hardware specs lack in support LONG before Pi can get their **** together.
Glad you took this subject, as this could save many from the frustration of adapting these limited devices to aspirational projects, looking for help in the Pi forums only to be ridiculed for 'not knowing' what to do. The acronym SBC makes one think that its a computer, that way, even smart watch is a computer, but with serious limitations one ought to know before investing in them. Most mini PCs and refurbished SFF computers better suited for a versatile computing environment.
I would like to see the development of an entirely new region/zone/layer of the internet dedicated to legacy, vintage, and low bandwidth computers. The vintage and low spec system world has been growing by leaps and bounds. I think it is time for dedicated internet spaces and or layers that work for anything from a text only 8-bit CP/M or DOS vintage computer on up through the computers of the 90's, 2000's, on up to about 2010. I am sure there are people out there already working on things like this, but it needs promotion and development.
@@editingsecretsThat is true, but far too many people don't really see it as an option yet. We need more promotion of interesting internet attractions and content for every tier of bandwidth. The more on ramp opportunities, and attractive options, people have to entertain and amuse themselves with low spec hardware, the more pressure it puts on high spec makers to stop making poor products.
@@markhatch1267 It would seem appropriate for RPi foundation to be leaders in the promotion of lower tech but still very useful free protocols, which could be a good fit for small hobbyist and embedded single board systems. Three minutes to load a full Linux GUI seems like overkill for a lot of great uses for a Pi.
Frankly, my dear, I don't know what to think! Thirty years ago I managed a Web site for a major software company (still in business!), and today I walk around with a computer (Pico-W) literally on a keychain. What I really wonder about the future is when "smartphones" will eventually merge with SBC-type applications --- and sort of blow up the whole thing. On the other hand, I have a friend who does not have a smartphone, and when I gave her a small radio/MP3 player, she couldn't handle the idea of inserting a micro-SD card, eventually buggering it up to be unusable. Admirably, Rasp-Pi has done a great service in aid of education, but more is needed. You're helping in that regard, EC. Thanks.
My use cases tend to fall into the "I need a computer that can run software without driving up the electricity bill" space. Stuff in the vein of OctoPrint, Home Assistant, or custom software like Slack and Discord bots to solve certain problems me and my teams have. For some cases more compute power like you'd find in older SFF PCs is incredibly useful, but a Pi at MSRP just nails the value proposition for so many of my needs. My gut feeling is that once Pi availability is back to pre-pandemic levels, you're gonna see a shift back to the status quo in some circles at least. There will be some cannibalization for more performance oriented tasks from the mini PC space, but I think the recovery in the other demographics may be stronger than you're expecting.
Until I discovered this new SBC, I would agree with you. IMO the main issue holding back consumer SBCs is drivers and crippled hardware. Fortunately Orange Pi seems to have corrected their ways and just released the Pi 5 *PLUS* with the uncrippled RK3588 and has many supported O/S options including *Armbian*. 8GB model reportedly ~110$US not including power brick or case. Looking forward to detailed reviews ....
I think SBC’s will go a similar route than computers did from the 70th into the 2000th. First the market was very fragmented. Many architectures, many operating systems, and a lot of incompatible hardware and software. This was causing high prices. Then there was the PC and things changed. I hope we will see similar things in the SBC space where we have software that just works and hardware that can be used on many systems.
@@uis246 Does that matter? Linux is entirely customisable anyway and we have RISC-V potentially entering the market. I am looking forward to messing about with RISC-V to see what it can do but I don't think it's going to dramatically change what SBCs do, it will just bring more onto the market with a more open architecture.
@@uis246 How's this for starters? Are Arch Linux and FreeBSD the same operating system or different operating systems? And what about Ubuntu Linux and Debian? Gentoo Linux and Chrome OS? AOSP and Google Android? That's a few to be getting on with, I welcome your responses to those questions - take your time...
100% agree with your analysis. I was an early adopter of the Raspberry Pi when they originally came out - I think I even have the original version with only 256K RAM. I was also lucky enough to get one of the first Raspberry Pi Zero 2 W when they first came out, and I've been trying to get another one, however they have been out of stock ever since. I have therefore given up completely on Raspberry Pi and moved to Arduino micro-controllers for all my Maker projects ... absolutely no problems with supply, ridiculously cheap, loads of clones & a vast amount of resources and support from the maker community. I can fully understand people moving to mini PCs if they want to run Linux & desktop apps etc. but unless Raspberry Pi can fix their supply chain issues very soon I think their days amongst the maker community will be numbered.
In my country, the price tag of Raspberry Pi 4B 4GiB is around 100 USD or sometimes goes up to almost 200 USD and that just a board itself which makes it unreasonable to buy, same as other sbc.😢
Nice overview, thanks. I think the biggest misconception they made for Raspberry Pi was to position this SBC as a desktop replacement. Who really needs LibreOffice on a Raspi for example? A full SW stack on the Raspi - crazy. I absolutely agree that these days a MC can often replace a Pi - ESP32 has two!!! cores, boots extremely fast, needs low power comes with BT and Wifi and so forth.
Great overview of the situation; I think one good outcome of the whole thing is that efficiency is something that's actually cared about these days, as moore's law keeps getting bumped out. Performance per watt is a more important metric than ever, and we are seeing some pretty powerful hardware using less than 5 or 10 watts!
Greetings Jeff. Loved your interview with Eben -- and I found it particulary interesting how his take on RISC-V seemed to have softened a bit. I'm looking forward to your factory tour!
Without arm chips you would still be using house heaters as computers. It forced all computer manufacturers to make more energy efficient computers. The running of a very large data centre with climate controls and air con to keep them cool was becoming exorbitant and if they can make servers run cooler then air con is not needed.
Tech TH-camr crossover!
Yep. I use an 8GB Pi 4 as a server for a whole variety of tasks (SMTP/IMAP, Asterisk, Pi-Hole, backup server, IRC server, etc.) Even though the system as a whole was not very cheap, the fact that it uses so little power makes me feel better about leaving it on 24/7. The fact that it's completely silent is a bonus.
Thats why i always ask myself for what those ARM SBC with old or really slow SoCs are intended for, specially if i look at the sub $80 market, in the last 3 years of my life, both at home and at work, ive been using microcontrollers for anything i needed GPIO I/O or even a basic usb host (thanks Pico), for a home server ive been using a J4105 ITX board because it is really low power and have all the I/O and performance something like the RPI4 do not have. And for a more general use pc the RPI4 is just not enoght and all SBC below $80 are petty much at the same place or even worse. The RK3588/S is a good breaktrought in ARM SBC in performance and I/O but the price has been going up non-stop. These days i have more hope in the Intel N100 landing in a good price/power/performance area for SBC and minipcs, because even the RK3588 that is old ARM arch by today standarts keeps increasing in price.
The raspberry pi is useless if people can't afford it.
To be fair, that could be said about any product. The RPI range covers a lot of bases, all well supported. Still good value IMO.
The cost and performance of a Beelink U59 out shines a Raspberry Pi 4B after adding in a case and power supply.
The pi was successful in it's goal. It created a huge market of low-priced SBCs. Maybe they are no longer made by them, but the foundation created the market, and manufacturers now create an endless selection of cheap SBCs
@@kelargo thanks for mentioning this. This 12v model is actually exactly what I was looking for for amateur radio use I think.
@@StepDub That is true and the Rasperry Pi has been put on the list.
You've been a constant source of knowledge in the Tech Space, keep doing what you do.
Absolutely 👍👍
This is a TEST COMMENT, as I believe my comments are being SHADOW BANNED by TH-cam!
EDIT: Apparently, this comment WASN'T Shadow Banned!
Feature creep and bloat are the banes of computing. For instance, years ago, we had perfectly functional word processing, database, and spreadsheet programs. They were easy to use and did anything that 99% of the users wanted. And, because computer systems didn't have many resources, these programs didn't use many. But then companies started adding features -- often features no one wanted -- to their software. The new programs were more complicated to use and used many more computer resources. Great news, if your business is selling new pieces of computer kit. Bad news if you are a consumer. And just as soon as the user managed to learn how to use the new software, the software company came out with a "new and improved" version. This version had MORE features no one wanted, used more computer resources, and had a new and more complicated UI. This cycle continues unabated today.
I totally agree. And this is why I still run Word and Excel 97! :)
@@ExplainingComputers I wish I could get a version of WordPerfect from the early days. It had a feature that was really useful -- "reveal codes." With that feature, you could see what characters the program put into your document and change or remove them when you wanted to. Word has never had such a feature, and if it randomly formats your document to suit itself instead of to suit you (as happens all too often), you've got the choice of living with it or spending hours trying to figure out how to fix it. But no one ever said that Microsoft created GOOD software...
I used reveal codes all the time! :) And missed it in Word as you say. Happy days.
And wastes lots of electricity worldwide and makes people buy new updated CPU's even when only 10% of the old one was needed to run stuff that the user clicks on to use
I bought an Orange Pi Zero Plus and an Orange Pi PC back when they were both under 15 €. They are very versatile and useful devices (one is still in everyday use) & exceeded the expectations set by the low price, but nowadays you pay closer to 35 € or 40 € for the same exact models, probably plus shipping and taxes. At that price (for which you can sometimes get a used laptop), their limitations make them of dubious value-might as well pay twice as much for an x86 device with eight times the performance.
The upper end of the 2020 SBC market (e.g. RK3399 devices) has been underwhelming in terms of user-ready software development. Three years later, something like the Rock Pi 4 still requires arcane Frankenstein-patching to have working video decoding-in only some circumstances, with some programs, under some desktop environments. It hasn't come down in price either, while X86 mini-PCs have closed the gap. The RK3588 may be powerful enough to bruteforce a lot of these problems with software decoding, but that somewhat defeats the purpose of an efficient and optimised ARM-SoC.
From $35 to $200 it has no longer been an option for my University. We’ve completely dropped the use of RPI and switch to alternative SBCs that are more available and affordable. The support of other SBCs has greatly increased over the years compared to what it used to be. Given that the price and availability made RPI attractive both of those positives are now a thing of the past. When they meet both of those criteria again we may switch back but until then we are done with the RPI Foundation.
That's perfectly understandable.
"When they meet both of these criteria again" - that'd be a few weeks ago for the Pi Zero, over the next few weeks for the Pi 3, and a few weeks after that for the Pi 4.
@@senseisecurityschool9337 - You keep posting this. Upton has been saying this since the summer of 2022. It's now the summer of 2023.
Tomorrow, tomorrow
I love ya tomorrow
You're always, a day away
@@originalbluebuddha Since December 2022, he's been saying EXACTLY this - Pi Zero at end of Q1, Pi 3 in Q2, Pi 4 in Q3.
Which is exactly what's happened. If you go back & read his post from December 2022, you'll see it exactly matches what has happened.
PS end of Q3 you should see Pi3 & Pi4 stock sitting on the shelf for a while, rather than selling out quickly - that's what he said in 2022 and what he still said last week, because he KNOWS when his shipments from the foundry are due, how long Sony takes to SMD package them, how long it takes to put them in boards and put those boards in boxes ...
@@originalbluebuddha th-cam.com/video/P9vna9jao9I/w-d-xo.html
That's 2022. That's what he said in 2022 - and it's exactly what's been happening, exactly when he said it would. So please kindly stop going around lying about people. It's not a good look.
I was really into the Raspberry Pi a few years ago, simply for the challenge of seeing if I could create usable, low cost desktop replacement. Thanks primarily to this channel I succeeded but unfortunately, after a while the shine wore off and I moved on to other challenges. I'm still a fan of SFF computers.
I'm using my Raspberry Pi 400 as a low-cost Linux PC, and for ARM development reasons.
This is perfectly normal, in my opinion. Just experiencing the SBC world is fun. If you don't have an advanced task in mind or don't like messing with boards forever, you can just... move on.
Strangely it seems all the manufacturers keep putting out those SBC with GPUs with no drivers for them.
@@knerduno5942 I was so sad, when my H616 SBC didn't have any GPU drivers for it.
@@ZILtoid1991 I have a RPi400 also and really like it . I use it to play with GPIO experimentation, etc.
There are two additional categories of SBC consumer: Small business/ISV back office user and the home automation user. The SBC is perfect for the back office for hosting network shares and running scheduled processes (e.g. nightly cron jobs, offsite backups). It's an always on, low wattage device with enough compute to handle the basic tasks demanded of it. A SBC is a very good, fairly inexpensive option for small business owners instead of having to invest in expensive server hardware. They are also good machines for a centralized home automation host. One machine on the network that IoT devices can connect to and are controlled by. People are slowly realizing that cloud managed home automation isn't all that great.
I agree with these application areas, but I included home automation in my first maker category, and excluded business applications.
There's also another one, the people who buy SBCs, don't use them and leave them in drawer. The solution for them is to simply stop buying SBCs.
@@Talkathon408 no, the solution for them is for them to give those drawer SBC's to me.
Did you comment before watching the video?
Raspberry Pi was the leader in SBC a couple of years ago. My opinion. Their inability to provide hardware to the maker community offered an opportunity for others to fill the vacuum. I was looking at several projects but the lack of hardware shut them down. While other manufactures are out there I am hesitant to change hardware and generally the cost is greater. I am not sure Rasp Pi can recover the base they once had. I think they are still struggling and are now in the catch up mode against other suppliers. The state of the art has moved on as you so aptly demonstrated. Great presentation.
Their problem was to put all their eggs in one basket. This meant when one of their suppliers was unable to supply them with components they couldn't make the boards. I can only assume it was the Broadcom SOC they couldn't get.
The trading side of RPI hasn't shut down, it is prioritising its industrial customers so THEY.dont have to shut down (or go elsewhere).. I expect that Pi's will become more available and hopefully eliminate the need for people looking for JUST one board ( of anyversion) to have to pay over the odds to a scalper who gad somehow managed to acquire some.
@@6581punk I think it's more of a case of having to make contractual commitments to investors who put up the money for the Pi Foundation in the first place. They probably had no choice but to send what stocks they had to industrial partners first. I guess you'd have to say "that's just Capitalism" and leave it at that.
It's ultimately for the best for us that one company does not occupy the entire market..
They have not had a supply chain problem. They have instead focused on industrial applications and staying with the current tech for way too long. I love their ecosystem but the need to start moving on the consumer side or just give up an focus on the industrial side.
I have to say you are definitely the kind of lecturer I wish I had at university.
Always engaging, always interesting, always concise... well worth the many hours I've binged so far looking at your previous content!
Thanks for watching. :)
@@ExplainingComputers Great video, thanks.
The first thing I check is free RAM vs. SWAP file loading. As soon as OS starts to swapping onto the ultra-slow SD card the usability goes to zero (and SD lifetime too). So the slow performance is not from bloated SW directly but from swapping instead.
Funny thing is that once OS swaps all unimportant data from RAM then it runs fine. So clearly there is a ton of libraries which are almost never used (but Linux wants them in RAM because that's reasonable thing for Linux servers with >100 GB RAM).
I totally agree with you about SBC becoming full capable desktop machines. I bought the best ARM CPU SBC available today with 16GB RAM for non-stop home server (Orange Pi 5 with legendary RK3588). Surprisingly it runs fine on 4K LCD and TH-cam fullHD videos are flawless. My old Rapsberry Pi 4 struggled on 4K LCD (FullHD was fine) and overall desktop user experience was very poor (usable but delays everywhere including mouse pointer). This RK3588 run multiple times faster, wow. And it draw 1.7 Watt only at idle (LCD turned off and running Armbian as file server via ETH), and about 8 Watt full loaded 8-cores by benchmark (less than my old RPi4).
On the other hand I do not agree about x86 SBC. They will never run file server at 1.7 Watts as my RK3588. x86 is CISC ISA which is outdated and inefficient for modern computing by design. Even CISC x86 runs inside as RISC computer so all this unnecessary CISC-to-RISC translation comes at power and transistor cost (hence inefficient in many areas). The biggest issue for x86 is variable instruction length encoding (instruction can be 1 byte up to 15 byte long). Modern CPU execute 7-10 instructions per clock so CPU needs to load and decode 10 instructions in parallel. ARM and RISC-V as 32-bit encoded fixed length RISC (every instruction is 4 byte long so CPU know exactly where 10th instruction is) runs flawlessly. CISC x86 suffer from polynomial explosion (2nd instruction can start anywhere between 2nd byte and 16th byte, 3rd between 3rd and 31th, 4th ... 4th and 46th, and so on). x86 need sophisticated predictors with millions of transistors for task where RISC needs few thousands transistors and minimal power draw. Am I the only one who thinks it's ridiculous resources wasting on outdated closed ISA?
Have you seen the new KING of CPU world? The most powerful CPU today is ...... ARM Cortex X4. Absolutely insane CPU monster and broke several world's records:
- Cortex X4 .... is 1st CPU on the world with 3x branch units
- Cortex X4 .... is 1st CPU on the world with 8x Integer execution units in scalar core (previous champ was Apple A11 from 2017)
- interesting is the fast evolution of Cortex cores each year (X2 had 4xINT+2xBrach, then X3 came with 6xINT+2xBranch, and now X4 with 8xINT+3Branch).
- Cortex X4 .... is 1st CPU on the world with decoder capable of 10 instruction per clock (last champ was Apple A14/M1 with 8 instructions/clock)
- Cortex X4 .... 13% IPC increase
- Cortex X4 .... 10% increase in area only (performance per transistor increase, this you never see in x86 world because all those old ISA workarounds)
- Cortex X4 .... 40% less power consumption than X3 at same performance.
- Cortex X4 .... is 3rd CPU which is 64-bit only (ARM32-bit decoder removed for the first time in X2, Apple removed 32-bit in 2017 on A11 Monsoon core)
- Fun fact: .... AMD Zen 4 .............. has 4x INT + 1xBranch ..... decoder 4 instr/clock (plus huge mOp cache with expensively decoded instr to workaround, but it works only for loops).
- Fun fact: .... Intel Raptor Lake .... has 5x INT (shared branch) ..... decoder 6 instr/clock (plus huge mOp cache with expensively decoded instr to workaround, but it works only for loops).
Cortex X4 is absolutely state-of-art CPU.
X4 server version (Neoverse V2 ) is used by Nvidia in Super Grace 144-core server CPU. I wonder how 60-core Intel Sapphire Rapids and 96-core AMD Epyc will fight X4/V2 core (which has higher IPC, more cores, much better efficiency). Also I hope Jim Keller's Tenstorrent RISC-V CPUs will enter SBC and server market soon.
I agree.
This chap is friendly too. One of the professors I had at university was a curmudgeon. He failed the whole class and curved the grades. My 17% became a 3.2.
In the context of bridging the digital divide, I hope SBCs stay in an affordable price range. It would be fantastic if more schools and libraries could use Raspberry Pis to teach digital literacy skills.
I would love to see something like a Zero 2 W but with a RK3588S at $30-40 even if it only has 2GB of ram it would be good to bring the desktop computer experience to the masses.
We tried this at my school. Massive failure. Everyone was frustrated because they couldn’t browse the web at a reasonable speed, let alone open Google Docs or LibreOffice in less than 3 minutes.
(They were Pi3Bs with 32GB Samsung EVO+ SD cards
@@tejasraman6913 That's not a massive failure, that's just organizers not knowing how to lesson plan/teach SBC projects lmao
The term affordable is entirely relative to inflation.
The reality is that in todays economy the various SBC makers cannot afford to produce them for the same prices that they once did.
Today's 32-bit microcontrollers can do that job. You aren't going to run a web browser on them. You can plug in a bunch of sensors and run a web *server* though. You're not going to play streaming HD video through them, but you *can* do serious image processing on them.
Thanks for your assessment and summary of the SBC situation. I must admit, I'm one of the customers who owns several Pi 3 machines and a Pi 4, all of which I used for writing, internet and the Pi 4 is streaming something else right now while I type this comment in your TH-cam channel. I never seem to get around to programming with them, which is sad. Of course, it's easy to wander off when I'm looking for BASIC, FORTRAN, MODULA and various flavors of C to play in . . . sigh.
With Pi 5 over the horizon still, I purchased and setup a $79.00 UBUNTU tiny PC last week. By the time I added a mouse, keyboard and monitor, I'd purchased it twice, and I will soon spend a bit more when I put a micro SD into it. Hope it works out well as a writing machine, and maybe I'll have some luck programming with it. Ha.
I'm a RPi enthusiast. I've been collecting them since the RPi 2. Whenever they released a new Pi I would buy one immediately. I'm so glad I did over the years. I have a small collection of them now. I mainly use them for retro gaming and media playback.
I have about 10
Well, I started in 2014 buying one, then in 2015 and following years I bought 3 of the 2015 model, I did not check when a new one came out, just that I hate those mini-mini ports that they think are convenient. The Pi 4 model I bought had USB-C power, nice step, but not everything. The Pi-zero I had setup with Pi-hole and one 2014 model I had Music box running, very unstable by the way and not updated anymore. The burden of open source! All in all, they are fun to toy with, but really stable and reliable? I would say no!, those micro sd cards, if a factory default shows up and by taking one out of the slot and having one piece in your hand and the other is still in the slot? Man, what a crap is that. Instead of making things smaller, they should scale up!
IMO on-board flash storage would be a huge mistake. Right now, you can repurpose a Raspberry Pi simply by changing out the micro-SD card. If you want to experiment with something that could leave your OS unusable, just copy your main card first. It's not just the option to boot from a different drive - it's the ability to remove your normal drive and know that whatever damage your experiment does, even if you manage to erase or corrupt all connected media, your disconnected main OS card is absolutely immune.
There's other use cases where that's not an issue, of course, but there's also other SBCs that already provide for those use cases.
You could have both.
Disagree, it needs to have an onboard storage as an option (note the word OPTION). Raspberry PI/whoever could sell a version with M.2 or onboard flash for those who want that option.
MicroSD cards are not that sturdy or vibration resistant, short of soldering the card direct to the PCB.
I just want an SBC with a secondary drive for data/files (not the OS), without resorting to a janky USB thumb drive.
my odroid sbc has both, sdcard and emmc (removable), rockpi has emmc to usb and emmc to sdcard adaptors ,
so i can have 3 emmc on the odroid, onboard, via sdcard and via usb,
sdcard adaptor opens up options for all sdcard readers, usb adaptor can be expanded a lot (raid?)
You could argue that some SBC should come with an M.2 slot instead of an microSD card reader.
I could see Raspberry Pi offer that as an option.
I just go the route of a cheap SSD. Since they're so cheap now it's not a bad option. Just create multiple partitions to try other things with.
To produce other installs, I create loop devices onto files on my main PC Linux. Just dd a file of the size you expect the partition to use, create a loop device for it, format it, mount and use an install script. Then just send it over to the SBC and use dd to write it to its own partition. Update the bootloader and you're GTG.
A lot of people jumping into the hardware space forget that Raspi or any other ARM solution is the ARM tax that you pay ... I hope we see a lot more RISC-V architectures being manufactured as they'll keep the cost down and also speed up the development of the RISC-V platform as a whole
Thank you for the survey. I use SBCs exclusively as headless controllers for tools (e.g. Octoprint or LinuxCNC) or to run home network services like a media server. I wish SBC manufacturers would make boards that focus on that instead of trying to be desktop computers.
Agreed, desktop computer is a stupid target market, especially for ARM.
SBCs are kind of dead. If you want something that feels like the SBCs of the old days, you should look at something like an ESP32-WROOM or STM32 development board. Most SBC makers are pushing towards making a microPC instead of staying focused on a computer suited for embedded systems. OTOH, if you like having all this modern power in a tiny package, you're much better off using one of those SFF workstations because they're getting thrown out left and right by schools and businesses, unlike RPis which are expensive and unavailable these days. No real reason to use a 32-bit ARM processor when you could have an x86-64 processor with integrated graphics for the same price.
@@tissuepaper9962 Yeah, I agree. I just like the small boards as I can tuck them away and just use them over the network.
@@OhHeyTrevorFlowers those esp32-wroom boards I mentioned come with wifi/bluetooth and they're smaller than a credit card if you get a headerless version. might he worth checking out in your use case, the boards I use are called ESP32 Feathers.
My view is that the Pi5 may change things but a lot of the market has switched to either microcontrollers or low end x86 hardware. Maybe the Custard Pi 6B, whatever happened to it? 😅
It certainly doesn't help that a lot of non-raspberry SBCs tend to have poor software support, whereas any old distro can work on x86 hardware.
An SBC costing more than $100 has a limited niche audience.
I think what Nvidia are doing and Orange PI are valuable (whatever happened to the Asus Tinkerboard?) but you can't beat the PI for support. For learning and experimentation you'd use a PI or a Pico. For a product I think you should be prepared to solder the chips to your own board.
As someone who's just gotten into this stuff and is looking to get in I'm looking at mostly orange Pi and odroid or just an esp32 for things I can get away with that.
The main desktop computer on my boat is a Pi 4B. My navigation computer is a 3B. A 4B with a couple of SSD or HD drives makes en excellent cheap file server. A Pi Zero W makes a great board for building a wireless security camera. These plain vanilla ARM boards will always be useful and punch well outside their weight class. Keeping the OS mean and lean is key to maintaining useful performance levels.
I just watched a video on the new Sigma SBC that retails for $648 with a 500GB SSD. I can get a fairly capable laptop for that! The whole thing is getting out of control. If you make a SBC with true desktop power, it costs as much as a real desktop, and the unit is too big for a lot of maker projects. No real advantage. Now that the supply problems are sorted out, the Raspberry Pi is a great option for getting a lot of work done for very little money, with a device that is smaller than a pack of cigarettes. But an OS that is too busy and too resource hungry, cripples it.
My only gripe with the Pi is that it is not fast enough.
Totally agree with you about software bloating, way back I squeezed my code into 2k EPROM's !!!
First of all thanks for the benefit of your brilliant insight to make a very interesting and informative video.
I was particularly drawn to the reference to “Has the Pi bubble burst?” in the title. Well I hate to say it but to be honest I think it’s about time that it did.
As a Raspberry Pi fan from day one I’ve purchased probably every variant produced to this day (on the odd occasions that they were actually in stock of course).
I’ve always been frustrated by the virtually always “out of stock” and even in some cases the ridiculous rationing of the main products. Imagine any other business model that says we’ll only allow you to buy our products if you’ve been a very good boy and only ordered one. And of course pay the full three quid postage cost for each and every tiny item.
I’ve sometimes wondered why I even bother to carry on trying to buy them when you look at much more capable and cheaper competition that are always available and easy to buy but it always comes back to “well yeah but the Pi has such a great “ecosystem” and is much better supported by the “community” etc. etc.”
Indeed this is true but we should all be very aware that this “ecosystem” and the support it provides has been built not by the RP Foundation themselves but by the users - the so called “makers”. It’s this group of people that has made the Pi the great success that it is despite the Raspberry Pi foundation failing in many respects to support them not least in being able to supply the damn things.
But despite this bizarre business model we still rave abut the products and turn a blind eye to superior competitive products. I’ve been guilty of it myself saying stuff like “oh well, they may be impossible to get hold of and the SD card socket is awful and it still only has a 32bit OS and no on/off switch etc. etc. etc. but they are the best supported and we still love them”
What finished me however is that when things got really bad and COVID followed by chip shortages and a plethora of other supply problems came along who did they support? The educational and maker communities, for whom the products were supposed to be for in the first place, the people that created their wonderful “ecosystem” and made them what they are? NO. The lovely Eben Upton threw them all under the proverbial bus and instead supported his “corporate” customers, i.e. lazy cheapskate companies profiteering by charging for perceived high value goods incorporating low-cost off-the-shelf hobby SBCs meant for educational use inside them rather than doing their own R&D. And leaving his loyal core customers having to sit through lockdown either with nothing or at the mercy of eBay scalpers cleaning up by selling the surplus stock at insanely inflated prices. Good call Eben!
Now nearly three years later when they are finally beginning to have some stock to spare they come back to us and say “hey guys you’re really so lucky now because we’ve decided that we can finally let you have some of our stuff again”. Well I say you can shove it up your proverbial Eben. I’ve had no choice but to move on and now I have no reason to come back again.
What goes around comes around. How about if we have a go at actually bursting that bubble and divert our efforts into creating an Orange Pi “ecosystem” instead?
Good post.
Since my interest in SBCs was basically inspired by this channel, I've done many projects and expanded my modern computing knowledge in directions I'd never imagined it heading. I'm appreciative for the gains I've made thanks to the community of enthusiasts. I was lucky to get some cheap pis when the market was good and they've been wonderful companions in my ventures. Sorry we couldn't all enjoy the bounty. Still dislike TH-cam comments like the "pi is dead" "we've moved on" etc. A web search shows me a University of Cambridge stat from 1 year ago - 40 million pis sold. Seems like many of us have working boards and will continue to bask in our good fortune. "Long live the (inexpensive, functional) SBC!"
I think lots of people started with an RPi and while learning to make things with those they also learned that it was overkill to use an SBC for those small projects and they moved to microcontrollers (STM, ESP32, ... etc).
Today we also see people buying a stronger SBC to replace their desktop. When you see what an SBC like the recent LattePanda Sigma is capable ...
If those "high-end" SBCs continue evolving like that it might be the end for the desktop computer as we known it.
Looks like the SBC market is splitting and the desktop users will get their own range of "high-end" SBCs.
A very good point. I had not thought of it that way. Pis got people into GPIO, who then migrated down to more appropriate hardware.
This is exactly the path I've taken. As much as I would love to buy something like a Raspberry Pi 400, it makes more sense to buy a refurb Mini PC and one or more Pi Pico boards. The Mini PC is more versatile (and in some cases expandable), and the Pi Pico can go off and do its own thing once programmed!
This was a great insight into the consumer SBC market! Nobody could do it better than you, really. SBCs have certainly changed in the past decade, where we went from a low-cost SBC to learn coding and do projects on to SBCs that have about as much power as a typical desktop and exciting new things like a RISC-V processor. Keep that Pi bubble floating! (well, at least for as long as possible...)
I felt it in my bones that this would be a possible future, in the short term rather than the long term. It was fun while it lasted. *slides in an i5 Mini PC*. Thank you as ever Chris.
Thank you for another fine video.
I've tried using Raspberry Pi SBCs for 3 different purposes.
1 Temperature logging to the internet. However, I replaced that with an Arduino type board a few years ago and the Arduino is far superior for that application. The Arduino was more reliable, with lower power consumption and better software options.
2 Motion Eye Pi Security cameras. The Pi is very good for this.
3 A lightweight mini PC for using when travelling. The idea was to use a small size bluetooth keyboard and hook it up to LCD TVs in hotels. With the advantage of having a package small enough to fit in any hotel room safe.
However, it wasn't really fit for such a purpose.
(i) The wifi was weak, so where my Android devices worked, in some cases the Pi could not get a signal
(ii) The bluetooth keyboard was laggy and kept dropping out.
One of the main reasons other than price that the Raspberry PI did so well was the community and forums around it that supported makers etc. This is just not available to any great extent with the other Arm based SBC's. Maybe it will come, but getting stuck without that knowlagable support can be a killer.
I think we will see more diversity in the computer resources being deployed with microcontrollers at the edge and sbc and micro pc's as servers and data processing nodes
Yup, it was the community and the support that made it such a smashing success. Similar with the Arduino. We had microcontrollers before the Arduino, e.g. the PIC series was relatively popular. But the Arduino really exploded the market, in great part because they made microcontrollers easier to use through great documentation, and fostering a community around the platform.
I am happy with my orangepi 5, I am using it as my desktop pc for less than 10w with 8 cores
We used RPi’s as a mix of all three categories. Regardless of all the naysayers, it is and was a solid piece of gear for its intended purpose. At first, my son used it to learn about the computing world, coding, and maker projects. I also used it for some of the some purposes but migrated toward some home automation and low-cost video production. Due to the current price point, I’ll use what my money can buy. Thank you, Chris! Cheers
Greetings Chris on this beautiful (at least here) Sunday. I put myself firmly in the enthusiast camp. Whether I need all 50-60 SBCs that I own remains a mystery, especially to the missus.
The SBC market has changed drastically changed since the Pi made its debut and truly launched the form factor into the public eye. The original idea of getting more people, especially chldren into STEM has seemingly been mostly lost.
Supply and demand has had a huge effect on the cost of SBCs along with the demand for more power, demand from users as well as demand from the heavier software requirements aka bloat has raised the need and therefore cost of some of these boards. The Jetsons have that higher cost but you do seem to get more with added GPU capability. LattePanda has also been a bit of a standout with their x86 based boards which comes along with the higher price tags. Their current Sigma release has finally put a line in the sand for me because I cannot justify that cost for my use.
RISC-V, while still in its infancy, should eventually change the landscape of SBCs in my opinion. At least, being open, I hope so.
Well that's it for another ramble. Please stay well and keep the excellent content coming.
I thought of you when I wrote that label! :) I'm very interested to hear that your reaction to the Sigmas was the same as mine. And I also agree that RISC-V will offer new enthusiast opportunities . . .
@@ExplainingComputers I am honored to be in the thoughts of a great mind who thinks alike.
"I put myself firmly in the enthusiast camp. Whether I need all 50-60 SBCs that I own remains a mystery, especially to the missus."
I can relate to that. I can relate to that, yes.
I switched some projects to repurposed min-PCs away from Raspberry Pis as they became harder to get, eg Octoprint for 3D Printing and agree with your analysis. As a maker I am switching to the industrial pi hardware using the CM4 mezzanine cards and IO boards systems for experiments and prototyping with the intention of getting a ULX4S FPGA card when they become available. But I am still benefiting from the vast Raspberry Pi ecosystem of products, software and information which I think will maintain it's advantage over the competition that essentially relies solely on offering a linux capability without putting in the effort to offer further value to the community.
I expect the SBC Market to become more fragmented as different manufacturers specialize in different areas, in response to enchrochement from Mini-PCs (especially fanless) capturing the more general Users and the MCUs capturing more of the Maker market.
The day may come when every SBC comes with a DIN Rail connector.
FWIW: I think you're probably right, but it's also unfortunate.
If one considers the reason for the development of the Raspberry Pi, it's a cheap way to get people in to 'coding' and 'making' with a supportive community. In many ways we are sadly moving toward the situation that existed before - a fragmented space of options, none of which can boast the RPi's focus on leaning. The result will surely be that once again the learning curve for those interested will get ever steeper and many will be put off.
As an aside I'd love to see the Pi foundation do a deal with a big mobile phone manufacturer. Not to make Pi's, but to make *good* phones designed to be gutted for the motherboard and other parts at the end of it's life on to which the Raspberry Pi's OS's could be run. Consider:
-What happens to all those old mobile phones today? Landfill for the most part.
-Most older mobile phones still have more power and capabilities built in than most SBC's, primarily because of economies of scale.
-Recycling your old phone and possibly the screen to use as an SBC is 'free' - selling dismantled phones for 'SBC' use would become a thing.
-An 'SBC' that came out of a phone is an 'SBC' that would likely be very easy to integrate in to other projects.
You Sir, are right on the money ! A phone designed from the outset to be re used as a SBC is a wonderful idea and just the kind of green approach that should be adopted ASAP.
A hat tip from the northern UK to you ! 😃
My first computer was h89 from heath kit. I ran hdos and later cp/m. It was a slow move up to x86 machine from 8080 machine. But I have enjoyed the ride. I get a big kick out of my RaspberryPi machines 4b and 3b. I use them with my trains and DCC++EX. But just recently my old 86 computer died. And I was directed to a mini pc by my mentor Chris and I have been a very happy person. Enjoy your channel your the wizard of acknowledge. Thanks
Ed
The SBC complexity and power has creeped up over the past several years while mini PC's including USB Stick versions have shrunk in size and cost. The decision to select one or the other has become more difficult. The SBC is still a great tool for teaching computer technology even in its simplest forms.
As for SBCs, I transitioned to the Orange Pi, and Banana Pi platforms. I turn them into remote control cars, drones, NAS servers, retro gaming machines, and as microcontrollers. I did successfully turn a Orange Pi 4 into a mini PC, it could stream videos, browse the internet, online shopping, email, word processing, retro gaming, streaming video, etc.
As for emulation, I use a SFF computer hooked up to my television. SFF computers has more power than any SBC, so it is the best choice for retro gaming up to a Nintendo GameCube.
Raspberry Pi have made it very clear what they think of their maker and enthusiast market in recent times. Lets hope that same market lets Raspberry Pi know what it thinks of them!
The Bud Lite treatment...
You totally nailed it classifying the type of people buying an SBC. Pi isn't going anywhere. There are plenty of options that are technically better. The Nvidia Jetson is a good example, however. Economics is the one driving force for everything you'll ever want to do. The Pi might not be the toughest dog on the blog, but when the Pi foundation sneezes, everyone else catches a cold.
I think the Pi Pico with wifi is going to surpass all the SBC as far as being the most popular to use in projects. One of the biggest weaknesses in using a SBC in a control application is the additional layer of an OS. This adds another layer of complexity to the design and increases the chance of having a fault. Running a single or multiple application that directly address the hardware, and using a technique called supper looping will probably surpass applications requiring a multilayered programing environment. The pi pico can give you the best of both worlds within its limitations.
I think the regular raspberry pi is going to be good option for at least the next 10 years, if the company remains stable. As Windows can run on different hardware, Raspbian should be Multiplatform between clone manufacturers. That probably won't happen because once a better platform is out there people will dump the Pi hardware and move to new hardware. Imagine what it would happen if the Orange Pi, or banana Pi foundation built a much better SBC that was cheaper, but had the ability to run the Raspberry Pi OS.
A good overview of the situation. I own a Pi4 that I managed to buy just before they became unavailable. I like the ease of swapping SD cards to try different retro gaming images, and Pimiga3, but I will admit to not using it anywhere near as much as I should. I have not considered using it as a desktop replacement, and I haven’t used it to control external devices through the GPIO. I also own an original Pi that I used to suck the EEPROM contents from an Original Xbox.
As for the future, I think that if the Pi situation gets resolved as promised then the product will continue to be a popular choice for a lot of people due to the community support.
Picked up an AMD SBC, with 8gb RAM, from Amazon for $110. It works great and I use it daily. Picked up another, upgraded to 16gb RAM and use it for the controller of my Fiber laser. Works great!
Finally picked up a third, 4gb version for about $90 and it also works perfectly for what I need it. Streaming music from my server to the sound system in my shop.
About to buy another to put a workstation near my milling machine so I can alter code at the machine instead of going all the way back to my desk.
Started programming micro-controllers back in the mid 90j's with the Microchip 16F84, now i'm using the ESP32-S3 and that thing rocks. About to release my first commercial product based on the ESP32. Wish me luck.
While I have my share of Raspberry Pi, I don't use any at the moment other than a Pi Zero for running PiHole.with 8gb RAM
You've dissected the grave situation in detail. A lot of people are resorting to buying second-hand computers, rejected by government offices after 4 years of use. Consequently, the cost of building a refurbished PC has increased astronomically in recent months. I think people will use a computer in a maker project in extreme cases, where there are no other options like an MCU to keep the cost low. Some projects require a regular computer, such as smart agriculture, traffic monitoring on private roads etc. On top of that everything needs some sort of AI which requires decent hardware. The supply chain problem has created a tremendous amount of turmoil in the consumer PC market here. Things are going to get worse! Computers will no longer be as affordable as they were 5 years ago. You got it right. Anyways, Thanks for the insight!
@gr-os4gd those second-hand PCs are not as cheap as they were in 2021. In fact, these are not cheap anymore. Everybody is looking for some old computer. Yes, recycling old electronic equipment is an excellent way of saving money and it's good for the environment.
Great video, as always.
Even today the Raspberry Pi is still the first thing that comes to most peoples minds when anybody mentions SBCs, despite there not having been any new offerings from the Raspberry Pi foundation for years and there existing many more powerful alternatives on the market. I was quite surprised that the latest Orange Pi boards haven't become the present day rage given how much more powerful and affordable they are.
But then again I guess that the community support that Raspberry Pi has built up over the years is an asset that competitors have no equivalent of.
Another excellent overview; really enjoyed this. Though you touched on it, one branch of SBC users is: the low-power, portable 'hardcore'. For amateur radio, quite a big market for SBCs as we also tend to be into electronics in general, portability and 5V USB, low current-drain are two critical aspects. As the distros become more demanding, so the tendency towards more power-hungry design tends to follow. I suppose the use of 12V for transceivers means we can accommodate greater current draw, though it is very undesirable, given power is always in short supply in the field.
Power over Ethernet seems ideal for these kinds of uses, if the SBC can be powered by it.
David May (UK computer scientist)
May's Law states, in reference to Moore's Law:
Software efficiency halves every 18 months, compensating Moore's Law.
With regards to the categories of "Consumer SBC Buyers" (starting around 1:33 in the Price Changes segment on the video), I suppose I'd consider myself to be in the 2nd category, it'd be nice to have a small and very affordable computing device that can be used for typical office work. As time goes by, Raspberry Pi might have its competitors, and while it is somewhat unfortunate that 32-bit support is being phased out (especially when it might increase the cost of newer SBCs) but we have to appreciate what the Raspberry Pi has done, in terms of laying the foundations for Single Board Computers in this day of age.
they shouldn't be able to decide that 32-bit gets phased out. it's not what we want.
I am getting old but love everything about computers and have been a subscriber for a long time exclent quality video and perfect English easy to understand.. Saying that I i have been bying second hand hardware i waa in London last week bough a Intel cor i5 1151 for under 20 pounds will be trying some Linux software.. Wish you well
I made my own SBC (RK3566) and received it a month ago, it has been very fun. I developed the camera driver (Sony IMX586), and recompiled the Kernel. It was fabricated and assembled by PCBWay. This level of fun and satisfaction couldn't possibly be replicated by buying a Pi
As a matter of fact I wouldn't waste one second of my time developing a project for a Pi, if I can't make a custom board with it's SoC, it's worth nothing to me
Are you a lecturer in life? Man, you really make good videos and explanation is so good. Thank You.
I was a university lecturer and freelance speaker for about 30 years.
@@ExplainingComputers ahhhhhh....there...i knew it!
Raspberry Pi did something amazing, by literally building the market for low cost SBC's.
Now the market has grown, changed and evolved - including the entrance of RISC-V SBCs.
R-pi is now a victim of its own success - especially with supplying boards to industry in bulk.
It could have a party to play going forward, but the market has evolved, so R-pi must recreate itself as well - much like another tech company achieving their initial vision/goals.
Great video. The two features I would love to see in the RPi5 are a return to full size HDMI connectors and a header for a case fan.
I agree with this -- and they would seem such simple "new" features. :)
Good and even handed discussion as always but I’d add a few points:
-Pre chip shortage the SBC particularly the Pi4 and Pi0v2 were very good at doing one given thing exceedingly well, be it a print server, SDR rig,Octopi server, KODI box, signage, emulation station. In such configurations they could be easily paired down to a minimal linux configuration and work well.
-The Pi4 and Pi0v2 in particular were excellent for having an excellent support and knowledge base. The base units were impulse buy pricing, and you could easily bring up many of the basic use cases in half an hour or so if you had an SD image. Since every Pi was nearly identical you could replicate and disseminate implementations easily. You could just as easily repurpose a Pi4 for any other app if one didn’t pan out.
-The low cost of base units avoided the “sunk cost fallacy” in that you could fail quickly and write off the cost of a Pi4 and more so a Pi0. Also this was a boon for STEM in that there was much less anxiety that you would break an expensive computer.
-The path forward may well be the compute module. If you can pick and choose hardware options on a carrier board as well as paired down linux a "CM5" could be a world beater.
I played with Pi's back when you could get one for 30 to 50 bucks, but now the prices have made me look at other sources for projects. If I've got to spend 100 bucks or more then why not look at and all in one PC and better if I can get a renewed one for maybe a little less.
My reason for staying with Raspberry Pi is the software and software support that comes with the Pi. The forum community also keeps me up to date with Pi related things.
Microcontrollers will certainly be useful for projects requiring minimal computing power. Mini PCs can take over the expensive SBCs because any computer can certainly be equipped with GPIO as we have seen in your video. SBCs need to have a reasonable price for surviving in the consumer market as the alternate methods are cheap and still work okay.
You make a great point about heavy distros and prices. With the scalpers having driven the prices up, it's getting as cheap or cheaper to get a mini-PC, or repurpose a five-year-old Dell from eBay.
Sometimes I wonder if we're going to ultimately find ourselves in a cycle of making minimalist devices like this, the software ecosystem gets bloated, until people get sick of it and go start over with some other minimalist platform.
All other issues aside, I feel like it'd be less messy if the SBC world basically fractures in two, one for people who want to use it for a mini PC and another for people who want a minimalist tinkering environment. I don't really like NUCs and whatnot but I wonder if they'll eventually end up taking over that niche.
There will always be a less expensive, better performance per watt ARM chip than x86 can deliver simply because of the instruction set difference. I see x86 taking over the mid to high end of the market, with ARM continuing strong where having the most lightweight hardware possible is a valuable advantage.
@@editingsecrets I think a more traditional microcontroller is a better option at the low end. An SBC is wasted when it sits all day doing nothing but blink lights and read switches, which is what most Arduino/RPi projects amount to. Even an ESP32 is overkill for most of the stuff that people use their SBCs for, excluding the people who want to use them as little desktop PCs.
I do think that once they have the stock of parts again, RPi will take over the maker community SBC space again. It's not about having the most impressive hardware. Other companies have made much more powerful SBCs in the same form factor. It's the software support that RPi offers that's really the winner. I feel like there's been a three year window of opportunity for a different SBC maker to dominate the market, and all they had to do was offer a stable, well supported OS and drivers like RPi does. It's a big task, for sure, but absolutely achievable. None of them really did that though, and I can't understand why. A company like Orange or Radxa could have pulled the rug out from under RPi. Not that I wish a bad outcome for the RPi Foundation, I think they've done amazing work for the maker community. I just am surprised that this didn't happen.
To be fair, I think it didn't happen for the same reason that Pi costs stayed up. Pi costs staid high because people were already in the Pi ecosystem, thus demand for Pis remained high even though they were exorbitantly over priced.
It's the same reason the Microsoft window's environment isn't going any place regardless of how many people grouse about it. Or the fact that it outright spies on its users.
This wasn't helped by the fact that the Pi has been seeing a lot of use as a ready made 'brain' in commercial/industrial products. Just toss into a hardened case and plug it into your product with proprietary software as a ready made solution.
For a lot of industrial products, the added price was peanuts in exchange for a well supported code base.
It's a chicken or egg situation. Yes, someone could have benefited from switching to the cheaper alternative, but the first person to defect would suffer from poor software support and have to start from scratch. And the savings often just wasn't worth that.
Yeah. Turns out everyone can slap together a working Single Board computer (just a matter of slapping some reference designs together on a single PCB) and start selling them.
But when it comes to software? most cop out and expect the users to sort it out for them. proclaiming it as "Community driven".
I have my doubts Raspberry Pi will ever get supply back up to demand again. They've gotten in bed with the wrong suppliers and they're making too much bank from their industry customers (who are probably scalping a lot of the supply they're buying up).
I can confirm for our market in Poland. Recent refurbished offers included:
- Pentium J5005 8/16GB RAM/eMMC Dell Wyse 5070 for 70$ - 10W, passively cooled, with M.2 SATA for storage
- Ryzen 2300U 8/256GB RAM/SSD Thinkpad ultrabook for 100$ (with poor HD screen, but hey, that's 100$ for full fairly modern laptop)
You really have to want ARM SBCs design wins (even lower power consumption, though that Wyse is pretty great; GPIO, form factor) to pick them over x86 HW like that.
Such an accurate recap of SBC's in the world today! Thanks as always Mr. Barnatt for explaining computers to the masses!!! Keep up the good work!
Hello, Chris my first Raspberry Pi was Raspberry Pi 2 Model B V1.1 2014, and as I had mentioned before I now have the Raspberry Pi 4 4GB model and there is a definite difference between the two. Thanks for another great video and the future of SBCs. I also have a Dell Optiplex 790 SFF Desktop. I upgraded to 1TB HD and 16GB of RAM. as my main computer. Also, have a Lenovo ThinkCentre M92P purchased from e Bay for $85.00 barebone, had to purchase the power supply, WiFi Dongle, and wireless keyboard and mouse. 240GB SSD. 4GB RAM Windows 10 Home. On another note, I was using my Raspberry Pi to learn Python to learn to code.
Excellent general synopsis, as always, Chris.
I'd perhaps consider a "spanner in the works", at least in terms of current suppliers, as being that soon many needs won't need their products. In that, I expect that in the next few years, printing one's own SBC will become an additional Maker's choice in some sense. I believe we're currently at the "not quite yet" period (for the _average_ maker's home setup) of being able to download a model of _some open source_ board, like the Blueberry Pi, or suchlike, and have consumer 3d printers which are tuned and capable of printing copper traces on a previously printed plastic base, with a design suitable for home flow soldering ovens.
I say we're not quite there yet, as I'm personally no where near as experienced an electronics engineer or maker as the majority of those who might categorise themselves as such, as my history is mostly in software. But I still think that I've _probably just about_ got the skills to do this, given some investment in current consumer equipment, and I think that feeling improves each year that goes by, simply from new, and usually open source, community advancements in 3d printing methods. Firstly, often as some print-it-yourself printer mod to enable some bleeding edge feature, and then the best of these hacks and mods tend to get brought into new printer models.
If you couple RISC-V, with some consumer printer capable of multi-sided board fabrication, a software package to build the CAD/CAM using a set of board-requirement options, and reflow oven post-processing to make a board with "just the connections I need" (tm) - then I think there's a branch of the future where entry level SBCs will be the "3d printed door latch because my local hardware store doesn't sell one" of the coming years. Or at least I hope so :)
With all the opened source hardware out there, if you were to produce something you’d like based on those open source sbc designs, I don’t think you’d ever have to bother worrying about future revenue streams , ever again.
A successor to the Pi 400 would be great. I really like the two I have.
Yeah the RPi is a good bit of kit, kinda has a Microbee feel to it
Yeah I really love the modern C64 vibes it gives off, just got one on my birthday a few days ago.
First, last, and always, I'm all about embedded controller header boards with some flavor of AVR, ARM Cortex M, Propeller, PIC, RISC V, LX6, etc, onboard. Thanks to the open source community, one can get a great C cross-compiler and IDE for most of them.
The RPi was miraculous to behold, but was something completely different from what I was used to. It was never big or fast enough to be a real desktop computer nor did it provide the full I/O access and throughput I wanted, needed, and expected from an embedded controller. For me, the RPi was primarily a great opportunity to use Octave and Mathematica, and that was worth a lot! 🙂
Greetings Chris.
This video gives an insight into how it's becoming difficult to revive fairly old hardware, even with Linux.
OpenBSD rocks!
How cool to see Chris sitting at a desk rather than just his hands!
All tis is just the evolution of a system - I've been in this business since 4-bit (yes FOUR) and it's amazing how it swings around and changes. Evolution in nature takes millions of years, we see it here in a matter of months.
👍 Very interesting. I ended up in your 3rd category with my last purchase. I noticed SBCs were becoming more powerful, more expensive and paradoxically less capable with recent software updates--very puzzling. I think the software bloat issue is good topic by itself. Many simple task apps have become massive; you wonder what these apps are ACTUALLY doing with most of their cpu cycles.
Another thoughtful video. It is very true that low low prices resulting in no manufacturing taking place helps no one.
Very insightful summary of the situation. In my opinion, an interesting option that few people know about is the possibility of using an Android TV box as an alternative to an SBC. For example, TV set-top boxes based on the Amlogic S905X3 can be easily converted into a general-purpose SBC running the Armbian Linux distribution.
Now that is an idea that i have been mulling for a year or two now, as i have a couple of S905's with 2Gb/8Gb i got for $15 and thought i could reformat them as SBC's with Linux instead of Android. However, i just recently read that they may not be able to be wiped completely clean and could still continue to house the original malware that they came with. I am curious.....what do you think?
Most of my kids had a Pi 2 / 3 and all of em started coding using python as a first language (all but one. She started with C++).
We are thinking of using a compute module on our CubeSat, be-spoke board obviously. I was at our parent company last week, their labs were full of PI’s. The thinking is, the PI platform and software has such a large user base it’s reliable and stable. Just need to consider shielding and rad hardening issues, but some chunky aluminium cases may do the job.
I think your predictions are spot on. Even with the current supply situation, Raspberry Pis still outsell the other SBCs that are more powerful but much more expensive.
I think for SBCs to compete with Mini PCs, ease of use needs to improve. I was recently asked to recommend a low cost computer for someone, and I recommended an Intel N95 based Mini PC. If an RK3588 SBC was easier to use, I probably would have recommended that. But for a novice user that doesn't want to climb a steep learning curve, any of the RK3588 boards would be way too difficult, and would probably turn into tech support time for someone like me. So for that market segment, a Mini PC that is preloaded with WIndows is still the most practical solution. If Arm based boards get much easier to use they might start taking a larger (but still minority) portion of the desktop space.
Thanks for your valuable insights Chris.
I have a range of SBCs and mini PCs which I use mainly for computational tasks like file hashing and audio and video encoding, which keeps my desktop free for daily use. I very rarely use a GUI so use BASH instead. I'm more interested in computational power comparisons between SBCs and Mini PCs e.g. your $100 Mini PC vs the Orange Pi 5. I'd very much appreciate if you could cover this at some point.
I also want to see that comparison.
I think if you need gpio or spi/i2c bus then go to SBC or choose MCU. If not then Mini PC or put it in container or just run in server.
In my case I use RPi4 for ruuvi tags but now also my main computer has BT so I can also run listening scripts there but I still like to run those outside my main rig.
Other stuff is using AVR, ESP8266, ESP32 or MSP430.
Yes i completely agree. I have moatly been involved in the SBC space for its SFF and low power consumption hardware. I don't think I'm alone in that. That said I just bought a very powerful (yet relatively low power consumption) Intel I9 NUC with 64GB of ram and 2TB SSD for around $1000 which fits in my back pocket.
That said i really support the Pine64 and Raspberry Pi, I think there should be a place in the workd for them so i will continue to support projects like these because i believe in them.
Amazing video with a great insight into the SBC market.
Raspberry Pi dominated the SBC marketplace and was originally meant for hobbyists and low-cost gateway into coding.
Seems that isn't simply the case nowadays, after the pandemic Raspberry Pi seems more inclined towards manufacturers and not hobbyists.
Not sure whether the Pi will be in the same position as it did 3-4 years ago given the current circumstances.
The worst thing after all is that other SBCs don't have a good software support or community and will take sometime to come near Raspberry Pi.
Coming to software, like we've always seen before, bulky software absolutely destroys a perfectly capable hardware.
Your last sentence reminded me of a decades-old saying "What Grove giveth, Gates taketh away." Other players are now in those positions, but the game remains the same.
I think you are spot on with your assessment. I had wondered why my Pi 3 units were preforming so dismal compared to how they used to. The pi 1 models are a hidden gem, without modification (LED removeal etc.) it idles at 80ma ! I too am moving to microcontrollers simply for the ability to just cut power, with zero data loss, that fact alone makes me choose a microcontroller over a raspberry pi zero in which bloat is making it darn near useless in most situations. Many Pi zero wireless and bluetooth capabilities can be replaced by a simple esp32. Bloat, cost, the long duration of unavailability have pretty much killed the Raspberry Pi ecosystem for me. These SBC are widely available now and stock is up (July 2023) but I fear they won't sell very many. The combinations of Raspberry Pi burnout, the thrill of finding one being available being quenched (the hunt is over), the extremely unlikely but possible release of a raspberry pi 5, and people moving on to less bloated, lower power, and more stable product unavailability will put the raspberry pi foundation in a situation where they spent a lot of money to get stock, only to find that no one even really wants them.
I had once used a Pi4 as a desktop, but after 2 years of trying to get as much performance as possible out of it, the limitations with TH-cam playback, etc, were just too much to make it workable any more. Now I'm using a X86 SFF desktop, but I do still use my trusty old Pi. It's now being used as my digital desk clock, a task that it performs perfectly.
You can make it workable but you have to invest time and effort learning to build a more efficient Linux on it - I do it that way myself using Gentoo Linux.
Raspberry Pi OS is a great all-rounder OS but on a board of that size and limited power, every CPU cycle and byte of RAM that you can claw back is important - and therefore customising it more can achieve a lot.
And that's just a statement of fact, by the way - it's not my opinion of what I think you or anyone else should be doing to follow my example. I am simple giving you a solution to the problem you state that there is with the Pi.
Firefox wants to implement V4L2 M2M for hardware video decoding for the Pi (and possibly other SBCs?). But I have moved on to the Rockchip RK3588 and 1080p is no problem, even just with software decoding.
I have to say that your last video on buying refurb business systems pushed me over the edge to buy a system just like I use at work. It's a wonderful system with twice the RAM as my work system and it runs Windows 10 just fine. It's my son's gaming system now. All for $110.
One other thing that surely contributes to make SBCs less interesting is that ESPx microcontrollers are getting more powerful and easier to use, and their low power needs make them a better fit for IoT home projects.
The other side is that the new cool thing for tinker with seem to be Large Language Models.
So yes, the space for SBCs has been squeezed a lot.
I think the biggest shift we’re going to continue seeing is away from the rpi as an appliance where the user is expected to just image an sd card, pop it in, and go, and back to projects creating installers/mgmt utilities that just target Linux with the expectation that users install at that level on top of whatever the base distro for whatever SBC they have is. Which to me is a positive both because Linux knowledge becomes more prevalent and because the supply constraints on a single type of SBC will be less impactful in the future.
The chip shortage has come to an end, and Raspberry Pi will soon have enough boards again. Several resellers already have sufficient stock, leading to a gradual decrease in prices. It's important to note that we've experienced significant inflation of 5-10% in the past years, resulting in slightly higher prices across the board. I anticipate that the new prices for the Raspberry Pi 4 will settle around $50 in the coming weeks or months.
And on what basis or authority are you making these claims? I am not saying I don't believe you, I'd just like to understand where your information comes from.
@@terrydaktyllus1320 it's all available out there ... just a simple Google will help you. Inflation has been steadily rising over the past 3yrs due to the effects of the Rona lockdowns all around the world. The fed has been increasing rates to curb it and that has not been quite great for the economy and that doesn't affect only the US, it has ripple effects all throughout out.
@@weiSane You clearly misunderstand me here - I don't do "homework" to make you or anyone else look right.
If you have evidence to back up your point then you present it when asked - because I do not have to take on trust anything you tell me. And the opposite is true also.
You'll forgive me for ignoring most of your "economics lecture" given that it is entirely unnecessary as something that adds to the knowledge that I currently have on the topic.
Did the OP appoint you as their official spokesperson, by the way? I do want to believe what they are telling me but only they can provide the evidence to back up their own claims.
Don't forget price gouging as well.
@@terrydaktyllus1320 damn bro...I was just trying to help lmfao. No need to come out guns blazing. It's not that serious. Enjoy your weekend.
It’s important not to lose the biggest picture here - the OG RPi created this whole space of $5-100 SBCs and there is now a perfect choice at all price points for all applications. That’ll never change so Ebun’s mission is complete. Just because the average price of this market is now eating into the low-cost x86 PC has no effect on the affordability of access at the bottom - quite the opposite as Wi-Fi is now on the $10 SBCs. Love this channel, keep it up
Thanks for a great video. I've always thought SBCs were a really nice idea, but I've also always been able to talk myself out of buying one because I had no particular use for one. I think most people don't realize that for the most common uses - browsing, email, video, music, etc. - a used office computer will do the trick just fine at a much lower cost.
I do everything with scrapped machines- but there's always the increased risk of failure and relatively low power efficiency of older gear you have to plan around. (backups and standbys, etc)
I'd have no qualms about jumping back into text-only office apps and doing everything on an SFF again when this current sub-$100 laptop craps out the rest of the way- I can run a separate STB on the TV for any video I need (I don't do ANY editing anymore) and use less power AND BE MORE PRODUCTIVE with multiple devices. That probably has a larger carbon footprint for manufacturing and operation of all this junk I much prefer to call "my toys", except the only actually new COTS are the SFF and STB...I think I'll have to do a deeper dive on real total costs, but I'm pretty good at guessing after 60 years in tech, and my best wild guess right now is I'm a little less burdensome to the planet than with a spiffy new laptop I can't afford.
My pi 400 does that for cheaper.
For me, it is not so much about whether the hardware is mounted upon a single board or not; or whether it is a microcontroller or computer; it is about the technology being accessible. When TI's MSP430 demo boards first came out, I remember struggling with 900 page PDF files of the smallest readable print, written for embedded engineers. At most I managed to do is blink an LED after a month of effort. The 2 great revolutions came with Arduino and Raspberry Pi in the form of super supportive communities geared to amateurs. Being able to 'creatively play' with the tech massively amplified my learning. Tutorials written for laypeople. Your channel, the 'Swiss-Guy' and Adafruit are massive contributors to my pursuit of technical growth. I hope there will always be a market for inexpensive electronic learning systems. Thanks!
Greetings James. I have somse videos coming up I think you may enjoy. :)
To think what was achieved on the systems in the 1980's and 1990's.... say the Amiga and the Atari ST... with 512k or 1MB/2MB Ram (if you were posh) - and the levels at which the systems were pushed. Every last bit was squeezed out of those machines - no bloat then, I miss that.
I don't know, I loved the video Chris, but crikey, it left me feeling rather depressed at the state of things... is it a case of 'get off my lawn'? possibly....
Yes, it is depressing how much bloat has crept into even simple things like Raspberry Pi OS.
Even if it is a misnomer , people use "SBC" _(perhaps they mean small-board-computer instead of soley single-board-computer)_ for those that are upgradeable such as the compute modules. Upgradeable RAM makes sense because you can replace the defective memory but also increase RAM size. That saves money since it prolongs life. Pi prices did fluctuate and anyway, a HBA or other PCI-e expansion could add cost and yet people do it. The 100Mbps mode for ethernet is worth keeping (even if a port for Gigabit rj45 exists too) because the 100Mbps driver is tiny. There are a great may use cases for that including being able to boot from various sources. This explainingcomputers is a well done channel and the enthusiasm is a factor that makes it even better.
My comment has no hate in it and I do no harm. I am not appalled or afraid, boasting or envying or complaining... Just saying. Psalms23: Giving thanks and praise to the Lord and peace and love. Also, I'd say Matthew6.
I think you nailed it. I started to use esp32 microcontrollers for sensors and led strips/strings as they are cheap and easy to acquire. I still have some pis running some network services. I guess i am mostly maker and part enthusiast. I also agree that if an sbc is going to cost about 200$ i might as well buy a small form factor pc.
With regards to a raspberry pi 5, maybe a cpu update, drop 1 and 2GB just focus on 4 & 8GB units. Maybe a single full size hdmi connector and a focus to keep the cist down. The pi wont compete well against small form factor PCs. I would rather resolve the pi 4 supply issue than get a pi 5.
I think the mini/micro HDMI is the future so more stuff can be fit on to the SBC with the same space.
@@knerduno5942 They're easier to damage and not as sturdy.
Well, they gotta do something in order to get more items on the board. With the newest RPi 5, they added a PCIe - M.2 Slot.@@maxxdahl6062
Great summary and opinio. I've had a number of RPi's over the years and have now settled on using an HP Elite Mini 800 UFF machine running Proxmox. This has given me a significant performance boost, allowed me to run multiple Virtual Machines, Linux Containers, Docker Containers all in a single computer. Busy setting this up as a cluster for failover, etc. These SFF/UFF machines can be purchased for very reasonable prices and fill the needs of multiple RPi's. The power consumption tradeoff is not that big, considering performance per watt as Jeff Geerling points to as a consideration. In addition, these are available in large numbers and not subject to the current RPi supply chain 'tax' or scalping. There are cases for the more industrial focussed use of RPi's as you mention and I can see a use in specific Maker projects where a small board makes more sense (e.g. MagicMirror, or DIN Rail mounted such as Kunbus RevPi)
4:04 there are also a large number of used x86 thin clients available, most starting at just about $25.
They are usually much faster than ARM SBCs and a lot more flexible (choice of OS, upgradeable, PCIe and SATA, ect.)
A good option is the Fujitsu Futro S920 that comes with a dual core AMD SOC (1.5.GHz, I think) and 4GB RAM by default - it will take 8 GB DDR3 SODIMM RAM. It also has an additional PCI-E slot that could take a low profile GPU, though I put an Ethernet card in mine to build a pfSense router with it.
Had been trying to get one of the these long long time. Pointless to bother looking, or it was back when I cared enough to bother looking. The whole Pi situation is far more trouble than a Pi is worth or the solution it could ever hope to provide. I'll spend my money on something avail and fight thru the lack of support. I can learn whatever other SBCs with better hardware specs lack in support LONG before Pi can get their **** together.
Very clear and concise presentation! It's helpful to get an understanding of the current state of SBC computing.
Glad you took this subject, as this could save many from the frustration of adapting these limited devices to aspirational projects, looking for help in the Pi forums only to be ridiculed for 'not knowing' what to do. The acronym SBC makes one think that its a computer, that way, even smart watch is a computer, but with serious limitations one ought to know before investing in them.
Most mini PCs and refurbished SFF computers better suited for a versatile computing environment.
I would like to see the development of an entirely new region/zone/layer of the internet dedicated to legacy, vintage, and low bandwidth computers. The vintage and low spec system world has been growing by leaps and bounds. I think it is time for dedicated internet spaces and or layers that work for anything from a text only 8-bit CP/M or DOS vintage computer on up through the computers of the 90's, 2000's, on up to about 2010. I am sure there are people out there already working on things like this, but it needs promotion and development.
This would be very cool. A "low-bandwidth-computing" Internet. I like this idea. :)
Old protocols already exist. If you can get an IP address and a firewall, you don't need anyone's permission to run a Gopher server or a BBS.
@@ExplainingComputers "You must be no taller than the sign to get on this ride"
@@editingsecretsThat is true, but far too many people don't really see it as an option yet. We need more promotion of interesting internet attractions and content for every tier of bandwidth. The more on ramp opportunities, and attractive options, people have to entertain and amuse themselves with low spec hardware, the more pressure it puts on high spec makers to stop making poor products.
@@markhatch1267 It would seem appropriate for RPi foundation to be leaders in the promotion of lower tech but still very useful free protocols, which could be a good fit for small hobbyist and embedded single board systems. Three minutes to load a full Linux GUI seems like overkill for a lot of great uses for a Pi.
Frankly, my dear, I don't know what to think! Thirty years ago I managed a Web site for a major software company (still in business!), and today I walk around with a computer (Pico-W) literally on a keychain. What I really wonder about the future is when "smartphones" will eventually merge with SBC-type applications --- and sort of blow up the whole thing. On the other hand, I have a friend who does not have a smartphone, and when I gave her a small radio/MP3 player, she couldn't handle the idea of inserting a micro-SD card, eventually buggering it up to be unusable. Admirably, Rasp-Pi has done a great service in aid of education, but more is needed. You're helping in that regard, EC. Thanks.
My use cases tend to fall into the "I need a computer that can run software without driving up the electricity bill" space. Stuff in the vein of OctoPrint, Home Assistant, or custom software like Slack and Discord bots to solve certain problems me and my teams have. For some cases more compute power like you'd find in older SFF PCs is incredibly useful, but a Pi at MSRP just nails the value proposition for so many of my needs.
My gut feeling is that once Pi availability is back to pre-pandemic levels, you're gonna see a shift back to the status quo in some circles at least. There will be some cannibalization for more performance oriented tasks from the mini PC space, but I think the recovery in the other demographics may be stronger than you're expecting.
Until I discovered this new SBC, I would agree with you. IMO the main issue holding back consumer SBCs is drivers and crippled hardware. Fortunately Orange Pi seems to have corrected their ways and just released the Pi 5 *PLUS* with the uncrippled RK3588 and has many supported O/S options including *Armbian*. 8GB model reportedly ~110$US not including power brick or case. Looking forward to detailed reviews ....
I think SBC’s will go a similar route than computers did from the 70th into the 2000th. First the market was very fragmented. Many architectures, many operating systems, and a lot of incompatible hardware and software. This was causing high prices. Then there was the PC and things changed. I hope we will see similar things in the SBC space where we have software that just works and hardware that can be used on many systems.
Except now there is one architecture - arm and one OS - Linux
@@uis246 Does that matter? Linux is entirely customisable anyway and we have RISC-V potentially entering the market. I am looking forward to messing about with RISC-V to see what it can do but I don't think it's going to dramatically change what SBCs do, it will just bring more onto the market with a more open architecture.
@@terrydaktyllus1320 Linux is not "many operating systems"
@@uis246 Define "operating system".
I feel an interesting challenge and debate coming on....
@@uis246 How's this for starters?
Are Arch Linux and FreeBSD the same operating system or different operating systems?
And what about Ubuntu Linux and Debian?
Gentoo Linux and Chrome OS?
AOSP and Google Android?
That's a few to be getting on with, I welcome your responses to those questions - take your time...
100% agree with your analysis. I was an early adopter of the Raspberry Pi when they originally came out - I think I even have the original version with only 256K RAM. I was also lucky enough to get one of the first Raspberry Pi Zero 2 W when they first came out, and I've been trying to get another one, however they have been out of stock ever since. I have therefore given up completely on Raspberry Pi and moved to Arduino micro-controllers for all my Maker projects ... absolutely no problems with supply, ridiculously cheap, loads of clones & a vast amount of resources and support from the maker community. I can fully understand people moving to mini PCs if they want to run Linux & desktop apps etc. but unless Raspberry Pi can fix their supply chain issues very soon I think their days amongst the maker community will be numbered.
In my country, the price tag of Raspberry Pi 4B 4GiB is around 100 USD or sometimes goes up to almost 200 USD and that just a board itself which makes it unreasonable to buy, same as other sbc.😢
Nice overview, thanks. I think the biggest misconception they made for Raspberry Pi was to position this SBC as a desktop replacement. Who really needs LibreOffice on a Raspi for example? A full SW stack on the Raspi - crazy.
I absolutely agree that these days a MC can often replace a Pi - ESP32 has two!!! cores, boots extremely fast, needs low power comes with BT and Wifi and so forth.