58 comments

[ 3.2 ms ] story [ 132 ms ] thread
I am not sure about SBCs, which are typically anemic hardware plus some GPIO pins, but I do have high hopes for NUCs with ~desktop class PCs. I look at my miniATX build which is probably idle 99% of its operational life and continually wonder why I am not using a book sized computer for my everyday life. A NUC with a <30 watt power draw + SSD + up to 32GB of RAM would be sufficient for so many use cases.
I almost certainly spend the majority of my time on an eight year old laptop I keep on my dining room table. I use my office computers for various things--certainly including multimedia--but I'm mostly in a browser and don't need multiple monitors for a ton of stuff. I basically don't use a full-size PC build any longer. The one I have is sufficiently old that my M1 Pro MacBook would run rings around it.
Sure, but you're also paying out the nose for that compactness.
There is a premium price for compactness, but it is not great.

In Europe, the prices, with all taxes included, for an Intel NUC Pro with Raptor Lake vary from around EUR 400 for an i3 CPU with 16 GB of DRAM to around EUR 800 for an i7 CPU with 64 GB of DRAM.

Similar Chinese computers are cheaper. If one searches for a Mini-ITX MB with at least the same number of peripheral interfaces as a NUC and one adds the costs for a CPU of similar performance, for PSU, case, cooler, DRAM and a small discrete GPU, to match the performance of an integrated mobile GPU, it is impossible to build a significantly cheaper computer.

Some economies vs. a NUC can be achieved only by building an even larger computer, with a microATX or ATX MB and by omitting some of the interfaces provided by a NUC, e.g. by not having any USB 4 / Thunderbolt interfaces.

A complete NUC-like computer with a slower Alder Lake-N CPU (e.g. Intel N100) can be found at less than $200, which is a price difficult to match with any bigger computer.

Some people like to point that there are complete laptops that have the same price as NUCs. Nevertheless, those laptops are garbage that I would not use even if offered for free. A NUC is much faster than most laptops that seem to have the same components (due to better cooling) and it has more peripheral interfaces that are also faster. A laptop that matches in features and performance a $800 NUC has typically a price of at least $3000 and it is marketed as a mobile workstation or as a top model of gaming computer.

From where can I get an Raptor Lake NUC for only €400?
Minisforum NPB7 is a i7-13700H for $490 USD. No RAM or SSD included.
At least in Europe, there are hundreds of shops that have it.

For instance, searching NUC13ANKi3 on amazon.de finds it for EUR 325 + EUR 10 for shipping.

Searching 8 GB DDR4-3200 SODIMM finds them at prices between EUR 19 (no name) and EUR 24 (Kingston or Corsair) + EUR 6 for shipping.

So the total is 325 + 10 + 24 + 24 + 6 = EUR 389 for an Intel NUC Pro with Raptor Lake i3-1315U + 16 GB DRAM, including all taxes and shipping.

A NUC13ANKi7, with Raptor Lake i7-1360P can be found at EUR 600 ... EUR 650. Adding EUR 150 for 64 GB DDR4-3200 makes EUR 750 ... EUR 800.

If all you want is a small PC, used thin clients are usually cheap and easy to repurpose.
NUCs are great but they are often completely overpowered for a given application. The additional 200 you drop for it isn't worth it. Also 30w may be chips for a PC, but for battery powered or hybrid powered applications it's a lot of load.

Where NUCs become very attractive in my opinion is when you need reliability. Rpis are cute, but anyone who has ran one consistently for 6months has likely had some sort of issue. Bad SD card, random restart, or what have you. Most NUCs will not do that to you. But there are also other SBCs with a price range in between that wont either.

Just go to Amazon, type 'mini pc" and you will get a list of windows (very likely will run Linux too) small new pcs in sub $100 range. Cheaper and better alternative to RPi. I mean the price that you can by it for, not the declared "imaginary". You need RPi only if you do embedding/robotics, and need GPIOs.
I’ve gone down bit of a rabbit hole the past week looking at low power x86 boxes.

I woke up last Saturday morning with my home router, a PC Engines APU2 that had been fault free for the past four years not booting making a clicking noise. In the end it turned out to be a fault with the powers supply but I didn’t know that at the time and needed internet so purchased a Protectli Intel J6412 from Amazon as it had next day delivery and I didn’t have time to wait a few days for PC Engines support who recently announced they where closing down to get back to me or shipping a new unit from overseas. As it happens PC Engines replied very quickly on a Sunday!

Anyway the Protectli J6412 turns out to be old gen hardware, however I’ve been extremely impressed. It’s idling away running Proxmox with an Opnsense a VM, Unifi, PiHole and NGINX reverse proxy in LXC containers.

That sent me researching some more as I have an old Ryzen 2600 running as a dev server but as idling away most of the time using a good bit more power than the Protectli. That rabbit hole has lead me to Intel N100 boxes such as the CWWK n100 on Amazon/AliExpress. They look like great low powered machines if you can live with a single SSD/16GB memory. They’ll do virtualisation/proxmox, run k3s, 6w max tbp, fanless, multiple nics. Not a cheap option but also not an expensive option all things considered. They even do a N305 which looks to be on par cpu wise to my current Ryzen 2600

It seems the market for small nuc type devices is getting very good right now.

Your power supply woes remind me that I want a NUC to take USB C power delivery. I hate proprietary chargers, and with tiny GaN units delivering 100W, that should be more than sufficient for anything but space heaters.
I bought an Asus PN51[1] sporting a Ryzen 5500U and of course a NVME SSD. Been running KDE Neon on and it's been great. Paid $500 complete with SSD and RAM here in Norway ie including 25% VAT, so $400 ex. VAT.

The fan kicks in under heavy load and sounds like a laptop, but if you want it completely silent you can get a new casing which is made to be fanless[2].

[1]: https://www.asus.com/us/displays-desktops/mini-pcs/pn-series...

[2]: https://www.akasa.com.tw/update.php?tpl=product/product.deta...

Ever since computers got cheaper and smaller, we’ve found different ways to employ them where it would not have made sense before.

I suppose there are two main areas where the RaspberryPi is used: Embedded device and computer replacement.

For the embedded device case, what may have been an Arduino project pre-RaspberryPI became a, say, Python project running on a full Linux system. For this use-case, microcontrollers have gotten much more approachable and many development-grade embedded Linux boards have come out built to the standard of the Raspberry Pi and it’s ease of use.

For the computer replacement, I’m talking about the random media center boxes, tv-as-monitor, network server, or just low power box to use as a desktop. We have the new nucs and knock off or similar boxes because all they did was take a netbook without a screen and keyboard, slice up the main board so it fits in a square, and break out all the I/O ports. These things that can be had for $100-$400 are respectable, low-workload computers.

Has the Pi bubble burst? As long as these ostensibly brain-dead industrial control systems companies keep gobbling up the supply of them, yes. Has the embedded and small-computer bubble burst? Absolutely not.

Software has also improved. Micropython is surprisingly good even on micro controllers.
It seems so wrong to use python in embedded but I know that's me being a turd sandwich.

But I do have to admit, arduinos and rpi zeros are in the same price point. I don't blame people not needing analog for going rpi even when it is gross overkill.

I'm just a minimalist. If I can do my project with 16kb of RAM, I'm not going to throw a GB at it.

That all said... This article is silly. No raspberry pis aren't popping or something. The biggest complaint I think most people have with them is you can't buy them anywhere. Constantly sold out. I would have bought two or three this year otherwise...

MicroPython is not really the same thing. It’s most comparable to an embedded Forth, which isn’t uncommon.
I know that's why I said I was being a turd sandwich.
For those TV/media centre/home server use-cases, I suspect it’s also that enthusiasts might already have a collection of older, good-enough hardware for that purpose as well.
Part of this is likely related to power consumption, particularly how power-inefficient older hardware is. A raspberry pi 3b+ will draw max of 2.5A at 5v, so 12.5Wh would make it 8.64KWh per month, and I pay about $0.13 per KWh, so about $1.21 per month for that thing to be screaming every waking hour. Meanwhile, desktop setups baseline power consumption is in the 100s of watts.
100W for baseline is too much for desktop. Maybe fail of power saving settings. Good tiny desktops can achieve about 12W at idle. https://www.servethehome.com/lenovo-thinkcentre-m900-tiny-pr...
I just threw in the 100 watt as a baseline because my 10+ year old rig runs that while idle and we seemed to be talking about using our older computers as the media player or home server.
According to this, 12yo Sandy Bridge PC uses 52W to 74W. 100W is hungry even over 10 years ago.

https://www.anandtech.com/show/4083/the-sandy-bridge-review-... https://www.anandtech.com/show/7003/the-haswell-review-intel...

You’re forgetting about the rest of the computer. Processors don’t run in isolation.
> Total System Power Consumption
The very machine I had in mind was actually my sandy bridge setup, specifically my i5 unlocked (K-skew) which I had a GTX-660ti (with a reported 110 watt TDP, but at idle it’s quite a bit less). Throw in the two spinning rust HDDs and, at idle, it does exceed 100 watts.
I would gladly purchase many smaller Pi boards if I could get them at anywhere near RRP. I'm not sure why they're so hard get - there seems to still be hobbyist demand, and there's strong industry demand as well. What's stopping the Pi foundation just upping production?
> What's stopping the Pi foundation just upping production?

My understanding of the rumor mill is that a few years ago the foundation started prioritizing commercial customers over hobbyists. This pissed off Broadcom which makes the processor so now they’re not getting preferential treatment at a time when there’s a lot of demand.

You had one job, Raspberry Pi.

Bypassing individual purchasers and attempting to sell to governments and foundations is what tanked OLPC. Raspberry Pi dominated the "cheap educational computer" marketplace simply because you could go out and buy one if you had $35 in your pocket.

Targeting businesses is either a dumb move, or the Pi Foundations version of Doctorow's enshittification.

Depending on your needs, there are a lot of alternatives now. Orange pi or rock boards are fairly similar but much more powerful than the rpi. However they don't compete in the pi zero class, they compete with the higher spec pi boards (4b). If you want a pi zero competitor, I'd suggest the esp32-s3 based boards such as the lilygo t display s3. There are a lot of other options, these are just some that I've used.
(comment deleted)
Remember how easy and fun it was trying to find a PS5 or high-spec GPU card at MSRP in 2020 and 2021? Well you can have the same experience trying to buy a Raspberry Pi in 2023!
>What's stopping the Pi foundation just upping production?

Broadcom

Specifically they for whatever reason refuse to increase production of the chips the Pis use.

But any time you ask why something is wrong in embedded, 50% of the time "broadcom" will be the answer.

The other 50% is "Qualcomm"
> for whatever reason

Probably they aren't making much money on them.

I don't think anybody's learned any lessons about rpi, and that as soon as they return their focus to the consumer the consumer will start snapping them up en masse like they were before. Everyone still thinks it's a chip shortage and not a conscious business decision to target b2b and bulk rather than independent consumers.
Genuine question - what is the lesson that you think should have been learned? I infer some (justifiable) dissatisfaction in your comment directed at the Pi Foundation's choice to target B2B - but, if that trend reverses and Pi boards become available to individual consumers again, do you believe there's a reason that consumers should still avoid them? Explicitly, I don't think that "The Pi Foundation might make consumer-harming business decisions again in the future" is a good reason, because a hobbyist who already owns a board is in no way impacted by their future availability (unlike, say, a company which chooses to rely on them, and would have migration costs if they needed a scale that was unavailable).
(comment deleted)
That's similar to my thoughts on it. I'm not going to cut off my nose to spite my face. If I can get a Pi 4 at MSRP, it still provides me more value than any other SBC. I have multiple RK3399 and even RK3588 boards that are faster than the Pi 4, sure. But in my use cases, those faster speeds aren't doing anything so I'd be spending more money for nothing.
> Genuine question - what is the lesson that you think should have been learned?

There are other competitors in the space (Beaglebone Black, for example) who got starved of oxygen because the RPi ecosystem dumped below-cost boards into the market.

Even worse, some of those competitors (Beaglebone Black, again) are, in fact, more open source that the RPi ever was.

The lesson to be learned: Beware of subsidized ecosystem dumping below cost. The party will eventually stop and then you will be left with nothing because of all the alternatives that you let get driven out of the market.

A very reasonable perspective, thank you!
I actually heard Eben Upton present an inspiring vision of affordable single board computers for both home and classroom education and experimentation. He seemed sincere, and the evidence seemed to back him up.

That's probably a part of why it felt like a bait-and-switch when they pivoted to business sales.

In retrospect it was probably a mistake for anyone to design a course assuming availability of Raspberry Pi boards without a readily available alternative board.

Perhaps some relevant lessons are:

1) Businesses, however inspiring their pitch, are strongly motivated by money; customer support is proportional to the amount of money they expect to receive from you in the future

2) Always have second sources and backup plans

What's your thesis for them returning to the consumer market if there wasn't a chip shortage? They presumably left that market for another reason, presumably driven by choice rather than necessity.
The lack of availability and the associated price hikes have been a big problem. We also now have plenty of devices with RK3588 CPUs which are quite a bit more powerful than the Raspberry Pi 4, its been "obsolete" for a good year at this point and while they have focussed on solving availability issues other SBC makers have moved on with lots of varied devices.
Yes, but the software situation with most alternative SBCs is terrible, which is why the Pi hasn't been superseded yet.
If the SBC is supported by Armbian then the software support is usually pretty good.

https://www.armbian.com/

Yeah I've been running several "Pi-clones" with Armbian for years now, mostly Allwinner H3 based, and only issue I had was due to a faulty USB power cable[1].

[1]: "fast charge" cable with 1 Ohm resistance...

Yes and no. Back in the day having a tiny sbc for any type of small task at home made a ton of sense since they were so incredibly cheap. GPIO, while nice, remained a gimmick for the most part. Fast forward to the 2020's, yes SBC's have become pretty powerful. However their price has skyrocketed and availability is almost none. For the price of a raspberry pi(or less), I can get a second hand office mini atx with 16 gigs of ram, 500gb ssd, a 5-6-7-th gen quad core i5 and a good amount of USB's and video outputs. And although 4-5 years old, it is stupid fast in comparison to any pi. I still have a bunch of pi's. One is an ssh server from the outside world to my home network but that's all it does. For anything else, I have two similar mini pc's in the corner of my flat and they do a much better job than any pi. One is processing a ton of data from a subreddit I moderate, along with backups, a small ml model to quickly filter out crap and whatnot. The other hosts dozens of docker containers running self hosted stuff - time management, documents, backups and so on. And there is no denying that x86(with all it's horrors) will always offer a ton more than ARM. All of which is horrible news for the pi's.

There are some potential applications still for consumers. Namely something that recently came to my attention - the beepberry. They are currently out of stock but this is something I'd adore. Get Kali running on it, and I want two immediately: toss one in my car and one in my backpack. I cannot even begin to describe how useful this would be. Then again - kind of a niche problem which is a subset of a relatively small niche to begin with - consumer sbc.

Based on price/performance alone an 8GB RPi 4 doesn't make sense even in its own category--it can't compete with an Nvidia Jetson. But if you're doing a lot of Pi work then the more expensive models earn their keep as dev machines. (Multiple compiler threads can actually use that RAM, and there are situations where transpiling on a desktop can't help.)
As the Pi got more powerful it moved into more areas. Some unassociated commercially based solutions now have "..or run on a pi if you have one" where you're really paying for some service fee or s/w costs (plex for instance) -they target other devices too of course.

If the Pi CM4 came with more motherboards which did PCIx4 (Taco from RADXA, the rockpi people) then you'd see it being used as the foundation for NAS builds and the like. Right now, Aside from the taco, I don't know what I can buy which exposes the PCI bus in a way I can buy semi-random cards to do eg SATA and work them reliably. Or some PCIx4 function which might suit another role.

I've found myself looking at breakout daughtercards for the M.2 slots (or whatever it is, which maps PCI into that format) -most of them seem to come with consequences: its not as fast as the single bus speed suggests if you multiplex anything behind them.

And, booting off devices which live on the M.2 slot sometimes seems painful from the write-ups: you have to fenagle the <whatever pi-like clone it is>/Pi boot state to respect that device as a boot device. Same with the CM4, there's a lot of "be brave and know how to do things with uboot" about driving that in any way more than the simple.

Pi "Clones" now mean they do uBoot and they have the GPIO grid. pretty much everything else is not necessarily very much like a Pi. Really? I think it's become "ARM core as a service" combined with some industry standard/semi-standard IO choices.

I'm running one Pi4 as a NAS, one Pico as a JJY watch time setter, one pi3 as a GPS tracker for Bert, and just retired one pi3 as a plex server for a chromecast audio: it turns out it's simpler to use closed-box solutions for some things, which makes me sad because I liked the IQ DAC I had on that Pi but it was not very nice to work with, in s/w.

I'll always play with small devices I think, but its more and more simply playing with s/w running on a small device which could be a pi, or a disused Mac, or an old laptop. Same-same. Other people do a LOT of GPIO based stuff or use the pi to emulate something else which is a different ballgame entirely. If you e.g. want the pi to run a slot car timing rig using hall effect sensors, you REALLY want a pi. Or a GPIO equipped equivalent. Being behind some kind of USB attached serial emulator just feels even more "janky"

I know people using Pi to emulate being an older CPU or an old I/O or storage device for some other old machine.

I don't see the Raspberry Pi going anywhere anytime soon. It has a massive community and excellent Linux support I'm sure it will popular for a while. The problem as I see it is, cost and availability. If I can get a x86 based mini PC for less than a Raspberry Pi, then I'm going to do that. I'm not going to buy other sbcs though because everyone I've seen using them has just had a headache getting the thing to work and keeping it up to date (not mainlined kernel)