hckrnws
>"...should began this July..."
Seriously? Should?
How. Many. Times. Have we heard it already?
In the timeless words of George W Bush Junior "Fool me once shame on you, fool me twice... Wait, I can't be fooled again!"
No way I'm going to get hopeful here and then pi locator will show one pi zero 2 W in entire Europe, in the one shop in Belgium the owner of refuses to ship out of his country...
How many cool open source projects could we have by now on other boards that remained available throughout this time if the liers at the Pi foundation (and people that believed them) didn't tell us every few weeks for last X years "it's only few weeks. We're making millions, this will all resolve the next quarter the latest!" All Bull**!
I've been saying it since year 2 of the shortage. Those Pis are not coming back. If your favourite project uses raspberry pi encourage them to support another board. If they(the foundation) was honest from the start and said "look guys, we've burned through all the charitable donation money and this business model is unsustainable, we have to pivot" I would have some respect for them, but not this.
Having said that I also have to submit some self critique. In my own projects it has been oh so much easier to just pull one of my remaining rpi zero 2w from the drawer (it has to be zero because I'm mostly doing RC stuff) and put something together in a day vs developing software for months. Still Rockchip has gotten a lot better with their documentation. It's not a documentation a not so technical user can take and run with it, but for developers it is perfectly usable. Now if we (developers) just wanted to use those other boards seriously...
> How. Many. Times. Have we heard it already?
Personally, it's the first I've heard of this.
I have heard them saying that the problem won't be solved at least until $SOME_DATE a lot of times. But this is a completely different statement.
> "look guys, we've burned through all the charitable donation money and this business model is unsustainable, we have to pivot"
AFAIK, the problem is that the chip manufacturer decided not to build it anymore. Again, AFAIK, none of the decision falls into the project. Besides, clones have been unavailable too, and if it was just about the project, the clones would still be created.
Yes, it's all a shame, and it had many bad effects.
Not coming back? But they clearly are? Pimoroni was shipping Pi 4Bs last week.
It has nothing to do with "burning through the charitable donation money". The Pi is sold by a (profitable, AFAIK) wholly-owned subsidiary, the Raspberry Pi Trading Company. Manufacturing them does not absorb Foundation money at all.
They have been supply-constrained on components. It's that simple, surely. Eben Upton has said it again and again, and sometimes gone into detail about which components. They chose to try to keep the whole ecosystem alive, which meant prioritising commercial customers, they've been manufacturing millions of devices per year, and it still isn't enough to meet demand.
FWIW, if I was in their shoes I would have discontinued the 4B series entirely, by early 2022. I personally think they should not be operating in that segment with an SBC. I think there should just be a Pi 400 that takes a CM4. Both to simplify things in the educational sector, and to get out of the "high end SBC I am not using that is in my desk drawer" segment entirely.
They should focus on the Pico, which is clearly now their most interesting product line and their most interesting educational device.
> Pimoroni was shipping Pi 4Bs last week.
How many? Four? Six? Did they go to scalpers as usual?
We shipped thousands last week and we didn't ship them to scalpers: https://shop.pimoroni.com/pages/ordering-restrictions-for-ra...
Plenty more will come in short order, the past couple of years have been a real mess on supply chains - it's been hard on all of us.
It looks like it's not possible to order one unless you've ordered something from Pimoroni before 12 May 2023. The only one in stock currently is the Pi Zero. So unfortunately it's still the same story. Yeah, we had lots of stock, but it's gone now and you couldn't order it anyway. I hope it gets better soon, and all this pent-up demand doesn't collapse once stocks are full.
AFAIK Pimoroni have had those "existing customer" measures in place for some months for the Pi boards; it's not a new measure. It's a good way of resisting scalpers, IMO.
As is allocating some of the stock to the "essentials kits" that most new customers actually need anyway (particularly Pi 4B customers; the official USB-C PSU is one of the few safe bets).
This is correct, and we have been able to ease up on the restrictions a bit recently:
- we reset the date before which you needed to have a previous order
- the Pi Zero W is now available to all customers up to 10 per order
I appreciate it's been frustrating but things are improving and will continue to improve - you'll see a lot change over the next couple of months I'm sure.We've done our best during this time to combat the scalpers - no solution is going to be perfect but we did what we felt was right to give our loyal customers as much support as we could.
I've ordered something now to at least get in the queue for the next time. Hopefully by then, some Zero 2 boards will be available.
Nope. You could actually order Zeros in packs of 10. There is increased availability of many models - the Zero 2 W, however, has an in-chip memory package that makes it a rarer item in general, and will take a while longer to ramp up.
Raspberry Pi Ltd is not wholly owned by the RPF anymore. They did raise outside funding a while back.
Discontinue 4B? Why?
Should Sony discontinue the PS5 when they failed to supply enough to meet demand?
Not sure what you mean about the PS5 unless you've misunderstood what I mean.
The CM4 can do _most_ of the work that the Pi 4 and Pi 400 can do.
A hypothetical future CM5 might feasibly be designed to be the CPU board of a still-tiny Pi 5B, Pi 500 and maybe even a smaller form factor.
The CM units are shipped in a bunch of different varieties, which would allow e.g. schools and consumers to decide which variation of a Pi 5 or Pi 500 they want.
It could eliminate all the "5B" SKUs and allow them to focus on manufacture of fewer SKUs at larger scales.
And this way, the market gets to solve the endless "I wish it had USB-C", "I wish it had two gigabit ethernet ports", "I wish it had two CSI ports", "Nobody serious will use it without some onboard flash" criticisms (that proponents of other SBCs have), with variations of the CM and the host hardware.
There are already some good reasons to buy a CM4 and a host board instead of a Pi 4B.
I wonder if this is a decision that could have been taken two years ago; there were already host boards on the market, and they could have produced one themselves by now.
They were churning out half a million CM4s a month but there was no way to use that to serve the consumer market.
I see your point. Perhaps they will move in that direction given time.
> How. Many. Times. Have we heard it already?
AFAIK what they've been saying so far has at best been "we're able to continue making them at our current pace". They've also been saying they were focusing on commercial partners over consumers (which IMO is better because commercial customers need some level of assurance of future stock to actually use pi's in their designs).
Neither of those things particularly screams "Regular customers will soon be getting pi's in their hands"
December 2022 They said they expected better stock this year, which appears to be accurate.
I'm more curious to see if they can manage to make a PI 4 successor given how many years the pi 4 has been around already.
"Those Pis are not coming back."
Yes, they are.
Disclaimer: CEO of Pimoroni - an approved Raspberry Pi reseller.
> In the timeless words of George W Bush Junior "Fool me once shame on you, fool me twice... Wait, I can't be fooled again!"
I agree with your comment, but I can't let such an eloquent presidential quotation be so misquoted.
Bush said, "Fool me once, shame on – shame on you. You fooled me, [I] can't get fooled again."
The last I remember was them saying that they thought things would be in good shape by the end of 2023. I'm pretty sure I heard that about 6 months ago.
This announcement seems to be consistent with that. Not only that, but I'm seeing more availability now than earlier this year.
I see no need to call them liars.
I never knew GWB used Raspberry Pis.
Each time this topic comes up, some broad threads emerge:
- people saying that there are "more performant" devices. Sure, there are. How do they perform in terms of kernel support and documentation?
- people (often Americans) saying that the Pi has been "supply-constrained for five years". It really hasn't; it wasn't particularly supply-constrained in the UK until about five or six months into the pandemic. Yeah, some models have been more popular than they expected.
What I don't understand is this: how is it, if you've not been able to get hold of something for _five_ years, there's been no obvious co-ordinated attempts to get mainline support for the alternatives, or even any attempt to match the documentation? Five years was plenty of time to pick a single series of devices and match the same level of kernel support, surely? Couldn't it have been crowdfunded?
It's almost as if most of the people whining are just upset there isn't a cheap device they can drop into a Kodi box without any effort.
The truth is that the Raspberry Pi SBCs are unique because they are the product of their funding model. If you want "more performant" devices that are as good, you might need to front up the cash.
> - people saying that there are "more performant" devices. Sure, there are. How do they perform in terms of kernel support and documentation?
One of the main options pointed at is low-power x86 boxes, which are better than Pis ever were. The rest of the ARM ecosystem is, I agree, poor.
> - people (often Americans) saying that the Pi has been "supply-constrained for five years". It really hasn't; it wasn't particularly supply-constrained in the UK until about five or six months into the pandemic.
Then it really has; there being a supply in the UK isn't exactly useful to most of the world.
> What I don't understand is this: how is it, if you've not been able to get hold of something for _five_ years, there's been no obvious co-ordinated attempts to get mainline support for the alternatives, or even any attempt to match the documentation? Five years was plenty of time to pick a single series of devices and match the same level of kernel support, surely? Couldn't it have been crowdfunded?
Partially, nobody was expecting it to actually take that long. If at the start of the shortage we'd known it was going to be 5 years, there probably would have been more of a push. OTOH some of the alternatives have been getting more polished.
> It's almost as if most of the people whining are just upset there isn't a cheap device they can drop into a Kodi box without any effort.
Okay. Yes? That's a valid usecase.
> Then it really has; there being a supply in the UK isn't exactly useful to most of the world.
There being a shortage in the US isn't exactly relevant to most of the world.
lol , nowhere its available.
I've bought several over the past few years, including during the pandemic, without any issues from some local distributors in the EU.
> One of the main options pointed at is low-power x86 boxes, which are better than Pis ever were.
In terms of what? Battery usage at idle? BOM to operationalize the design? Surely not.
> > It's almost as if most of the people whining are just upset there isn't a cheap device they can drop into a Kodi box without any effort.
> Okay. Yes? That's a valid usecase.
Funny enough, all throughout the pandemic the one thing that kept being made was Android streaming boxes. I'm pretty sure it's dead simple to wipe those and put Kodi on them. So it's not much of a valid use-case. (I'd instead expect most hobbyists who want Pis, to actually want them for running game-console emulators — specifically, emulators for consoles new enough that those weaker streaming boxes would struggle with them.)
The Pi is not stellar in terms of battery usage at idle, either. Its CPU doesn't have proper power management handling, which is why all homemade raspi-based phones/tablets/etc have a <6hr battery life.
>>> - people saying that there are "more performant" devices. Sure, there are. How do they perform in terms of kernel support and documentation?
>> One of the main options pointed at is low-power x86 boxes, which are better than Pis ever were. The rest of the ARM ecosystem is, I agree, poor.
> In terms of what? Battery usage at idle? BOM to operationalize the design? Surely not.
I was responding to kernel support; x86 boxes can usually just run straight mainline Linux with no problem, which is far better than most of the ARM ecosystem (even Pis aren't running mainline, they're just a well-supported downstream).
> Android streaming boxes. I'm pretty sure it's dead simple to wipe those and put Kodi on them
There are alternatives, but then we're back to spotty support, fragmented ecosystems, and using a Pi for video streaming is still a valid usecase.
Some of those Android TV boxes actually had pirated apps of various kinds - and SoCs that were gimped to the extent where you really needed the source to rebuild that particular version of Android.
Can you give some examples of the low-power x86 boxes? As of my last knowledge update :) none of them took off
HP T620 is one of the popular ones.
> - people saying that there are "more performant" devices. Sure, there are. How do they perform in terms of kernel support and documentation?
And Pi isn't even supported fully in upstream kernel! And it's so much worse in competitors...
> The truth is that the Raspberry Pi SBCs are unique because they are the product of their funding model. If you want "more performant" devices that are as good, you might need to front up the cash.
Not really. The key point why people want it is that they are by far most seamless product out of the competition (sans maybe some x86 ones I guess). 3rd-party patched kernel is always PITA especially over long lifetime but rPi got theirs in "near vanilla" debian-derivative so there is very little friction. And on top of that of course ecosystem that grew around it.
*if* there was competitor that have similar featureset yet work at 100% vanilla kernel there would be a chance but there is always some problem to deal with.
Right. I guess what I mean about "product of their funding model" is that you do not get a seamless, joined-up-design product at this price point, with that much community buy-in, that much documentation etc., without the Raspberry Pi's origin story.
Any number of companies have tried to do this and failed, because you don't get to _here_ by just launching a new commercial, cost-effective, powerful small product and hoping it gets supported. You get to what the Pi is now by starting the way they did.
It sometimes surprises me that the US market, with its many tech billionaires, has not managed to achieve anything similar. To the point that at least one US tech billionaire (Eric Schmidt) has backed the Pi Foundation instead, because it's doing the actual work.
People doing real projects with deadlines find another solution and don’t whine about it on message boards
You forgot all the people who forget there are N5095 (and now N100) mini-PCs galore at pretty reasonable prices these days.
Prices over $100 are not really competing with Pis.
For a living room/desktop use case, where you’d have to buy a case and storage, they are.
A pi case is $3 plus shipping.
If my goal was attaching several terabytes I'd be even less inclined to spend $120 rather than $40 on the computer, but for living room/desktop use case half a terabyte is plenty and only costs $30. So that's $150 vs. $70 total.
The shortage ruined the brand for me. Any sort of commercial product based on this cheap platform had supply issues, which caused a pivot. I had an old boss balk at me for suggesting we use Pi's back in 2013, I now see why.
Glad to see the gouging finally come to an end!
Just to reinforce what another commentor said, Pis were still being made available to sectors like business and education, so vendors who built products around rpi were still able to operate. We worked on a project in 2021 where we needed around 150 and had no problem sourcing them from a single vendor at MSRP alongside their product.
Of course, that's not what I expected since I'd been personally affected by the shortage to retail, and because the others weren't aware of that, I probably looked worried over nothing when I suggested we explore other SBCs.
> vendors who built products around rpi were still able to operate
> We worked on a project in 2021 where we needed around 150 and had no problem sourcing
We had minor sourcing difficulties ordering much fewer for production (10-15).
That's because Eben was telling everyone it was about "protecting jobs" when it was really just about selling to the largest companies. All the cottage industry folks who needed Rpis here and there? They could get fucked as far as Eben was concerned.
Basically: greed (lower overhead from larger orders) and more greed (chasing large companies, I guess in the hopes of securing fat, profitable long term contracts, or maybe donations, or both) and probably trying to cut out distributors, too (the "email us and we'll work something out" he kept telling everyone), again because of greed.
Raspberry Pi foundation burned their brand image and good will to the ground. Everyone has moved on to using clones, other ARM devices, embedded systems that have AI accelerators, old PCs and laptops, ESP32's, and so on.
Half the shit people were doing with Rpis were better suited for things like ESP32's anyway.
As a counter point to this we, Pimoroni, supported a few smaller businesses that relied on Raspberry Pi during the shortages with modest numbers in direct coordination with Raspberry Pi.
These were not "the largest companies" by any metric. They were often orders of just tens or hundreds of units. It was absolutely about trying to keep commercial users who had locked into the Pi ecosystem viable during the supply chain disruption.
I think I've only ever used the row of GPIO pins on the Pi for UART, and for running the dog-slow screens that use up most of them. These days, I'd put Micropython on a Pi Pico or other MCU for GPIO.
What was the path you took to get volume from vendors? Any links would be most appreciated.
What I did was email business@raspberrypi.com, as suggested at https://www.raspberrypi.com/news/production-and-supply-chain...
To clarify, it was from a single vendor who sold a product that was designed around use with particular models of raspi, not bulk individual boards (even though that's how they shipped).
This was also two years ago, although we've been able to get more from them since, so I'd recommend calling companies that have any kits or products that include pis and asking them for info on ordering. If you have a procurement office or purchaser, they can help.
You'll have to forgive me for not wanting to threaten our golden goose just yet :)
The reason hobbyists had shortages was because they allocated supply to businesses to help them stay afloat. Possibly your old boss was not right.
You are both right.
Supply was allocated to businesses, but there was not enough for every business, so quite many of them could not buy Raspberry Pis for their products.
Then again, everyone was having major supply issues. Yes, you could pivot from the "hobby" R. Pi to a "serious" industrial brand that had sales people talk big words about how they don't have these "beginner player" issues - right up to when it was time to deliver and suddenly they start talking about 100+ week lead times and 2025 delivery dates...
I thought the raspberry pi was meant to be an educational diy system? Prioritizing businesses over hobbyists seems to go against that.
Lots of small businesses sprang up that used it - perhaps some of them were even started by hobbyists. If they didn't get their supply it would have cause a loss of jobs and hardship.
The Pi Foundation decided to help those people keep their jobs during COVID etc which seems to be the clear moral choice.
They surely could have pivoted to an alternative, potentially creating more jobs in the progress. Most commercial raspberry pi projects should be microcontrollers anyway, and certainly not be running on the sdcard.
Obviously they didn't think so - whereas hobbyists can do so too and have no risks to worry about.
How is that the brands fault? For better or worse they even prioritized commercial products despite it not being their core audience.
This is the first I am hearing about the commercial support on this. I had no idea my EIN was so powerful. The brand in my mind is 100% hobby because if I was able to source enough for prototyping (which was not possible over the last 18 months without getting gouged) then sticking with tried and true platforms was the only path forward. Not trying to be cheeky, just surprised the prioritization for commercial products was not publicized more.
I don't know what quantities you wanted for prototyping, but on the actual "we have a product and need rpi to ship" this was well enough advertised (imo). I certainly knew about it just from looking into acquiring some hobby-use Raspberry Pis.
How is the impact of the brands choices the fault of the brand? Did you forget an /s?
Their choice can most certainly be questioned, but in this case the choice favored the use-case OP was talking about. So if anything it should have strengthened the brand image from that point of view.
It was mentioned as a footnote in the Dec 2022 update. Seems like a miss by the marketing team. My bad for missing this and only seeing the headlines.
Pivot to what? Any sort of product based on SBCs had supply chain issues. The supply chain risk with comparable board was and still is higher. Maybe if you target a common denominator you can amortize that, but that also has a bigger upfront cost.
I need to give it two more years to see how stable the supply really is, before planing to extend on it.
Right now it's basically supporting the old systems to keep those at least running.
I'm sure your old boss magically predicted pandemic and chipaggedon happening /s
Not sure if rpis are that relevant as of now when there are so many other, usually more performant and overall compelling offers in the same form factor and upstream support to a certain degree.
I agree completely. I've ranted on this a bit on HN before but in many ways the Raspberry Pi is a victim of its own success. It more or less became the "jack of all trades master of none" for a staggering variety of projects and applications - even when it is far from the best approach. I personally cringe when I see people trying to string together a bunch of drives on a shared USB bus for a NAS application, or trying to hack something around the single lane of PCIe or whatever... When by the time you get to kludging that together you could have bought a small used NAS with drive bays for equivalent or less cost that would be at least an order of magnitude faster and significantly more reliable.
That said "you do you" and if you want to (poorly) run a bunch of stuff on a half-dozen Rasperry Pis (or whatever) then more power to you.
On the flipside of this, the wide ranging and very robust Raspberry Pi ecosystem kicked off a surge of people self hosting all kinds of applications and getting people to leave cloud services. For whatever reasons the "look for a thin client on eBay" approach was never going to kick off the explosion of these self-hosting options and projects. So we should all be very thankful of the Raspberry Pi for that.
I'm glad that with the shortage people were forced to re-evaluate some of these approaches and projects to realize things like "Oh, this $100 used Intel XYZ is 20x faster and still low power enough to not mean anything".
With the launch of Willow and the Willow Inference Server I'm learning a lot about how people self-host and the variety is staggering. We see everything from "Well I run ABC on Pi 1, DEF on Pi 2, GHI on an old mac mini, and XYZ on an Intel NUC" to single Raspberry Pis to extremely complex Proxmox configurations. I understand a lot of this self-hosting infra grew gradually over the years but needless to say a single, cheap, reasonably powered machine can do all of that and more.
The answer to all of this is "where's the fun in that"? I can afford to pay Amazon <$10 a month to host progscrape.com given the absolutely minimal amount of traffic I get, but there's something fun about using an underpowered, popular platform to build something that a handful of people use, and keep it up over time.
The original home computing explosion didn't come about because of the power of those early platforms -- it was because the thrill of accomplishing something under the limitations of a popular platform is the reward early pioneers needed to keep going.
So yeah, the Pi is not ideal for so many use cases, but successfully arm-wrestling this well-known-but-limited platform into serving web traffic or doing something equally useful is why people keep using it.
> The answer to all of this is "where's the fun in that"? I can afford to pay Amazon <$10 a month to host progscrape.com given the absolutely minimal amount of traffic I get, but there's something fun about using an underpowered, popular platform to build something that a handful of people use, and keep it up over time.
How is this any different from self-hosting on any other hardware? Other than the admittedly fun and engaging aspect of putting the work in to string together a bunch of USB hubs, cabling, and SATA bridges, etc. I think a lot of that thrill comes from the challenges of pounding a square peg into a round hole along the journey.
Again, whatever works for you but it's been curious to watch it progress over the years.
You're right about the square peg, but the big difference is the size of the community. I could do this work on any platform but if there's only a handful of people that know what a MangoPi or KiwiPi (names invented) can do, there's a whole lot of context to understand how cool a hack is.
Ditto for reading about projects -- it's great that Alice's Intel nuc can serve X pages per second, but it's easier to read and understand the technical feats of someone hacking off traces on a Pi 4 to expose PCIe because we all know the limits of the platform.
The comparison to other ARM SBCs is kind of a non-starter to me considering what the software support and ecosystem/community looks like. I have an entire graveyard of ARM SBC Raspberry Pi alternatives and I've given up on that approach due to the horrendous software support and fringe ecosystem.
However, I'm primarily talking about x86_64 here and is has the broadest software support and biggest community of all.
> I personally cringe when I see people trying to string together a bunch of drives on a shared USB bus for a NAS application, or trying to hack something around the single lane of PCIe or whatever... When by the time you get to kludging that together you could have bought a small used NAS with drive bays for equivalent or less cost that would be at least an order of magnitude faster and significantly more reliable.
I don't think this is a fair comparison. In principle, anyone could follow the hacky instructions for the Pi and complete the same project. In some cases (such as the PiHole), it might even be worthwhile.
If your instructions begin with "find a used thin client on eBay," however, then there's no guarantee that the subsequent instructions are useful. My thin terminal might not be the same as the author's, and the author certainly can't specify a single particular brand/model because the eBay stock is limited and changeable.
"Mainstreaming" a hobby project needs the standardization at least as much as it needs the performance, and the Raspberry Pi's ecosystem remains an advantage. Ironically, this advantage is the same reason that the company prioritized business deliveries during the shortage, since business use depends even more critically on the stable, supported platform.
We're in agreement. From my original post, one sentence after your quote:
> For whatever reasons the "look for a thin client on eBay" approach was never going to kick off the explosion of these self-hosting options and projects. So we should all be very thankful of the Raspberry Pi for that.
I'm looking for a noiseless low power NAS solution. Any recommendations?
After my transtition to an M1, now I can't stand any fan noise running on the background.
Get a NUC or mini-ITX system and a fanless case[1][2]?
[1]: https://www.akasa.co.uk/update.php?tpl=list%2FCHASSIS+POWER....
[2]: https://www.akasa.co.uk/update.php?tpl=product/product.detai...
Where do the drives live? Drives, as in at least two, but more likely 5, if you're taking fault-tolerance seriously.
Since you want silent it'll be SSD's. One internally, you can get external USB housings for the others. Shouldn't be too shabby if using USB 3.1 or better.
If you want HDDs then a fan is a non-issue, the HDDs will drown out the fans.
An Odroid HC4 https://www.hardkernel.com/shop/odroid-hc4/
> usually more performant and overall compelling offers
True.
> compelling offers in the same form factor and upstream support to a certain degree.
But that degree matters. What SBC has the best upstream support these days? Orange? Banana? Pine? I honestly don't know, despite taking a casual interest. Also, upstream and driver support gets worse if you need a newer chipset, to support features like 4k or H.265 or USB PD or 2.5GbE. To be fair, Raspberry Pi doesn't offer the latter.
Quantity has a quality all of its own. Nothing else has quite the same scale.
It doesn't hurt that the OS is a very well-supported Debian derivative, but the uniformity of the ecosystem is a massive benefit you don't really get anywhere else.
Raspberry Pi OS is a well-supported Debian derivative but there are a wide variety of operating systems available for the Pi. I don't run Raspberry Pi OS on any of my Pi's. That being said, I do run a Debian derivative on a few of my Pi's, Ubuntu server. I run OpenWRT on 2 Pi's and Fedora Desktop on another.
Chromeboxes, ThinkCentre M-series, OptiPlex Micros and 7070 Ultras, Intel J-series boards, etc have been around since before the supply chain disruptions, but I have so many projects built around and parts for rpi that I can't as easily scale or retool.
I think their scarcity has made us all appreciate what having consistent low-power SBC platforms with convenient programmable interfaces is. I found uses for every Pi 0/2/3/4 I had that had been sitting disused because I was too excited to move to 4s when I really didn't need more power.
On the contrary, the Pi has gone from a small computer to the base of many projects. Where for many self-hosting is synonymous with the raspberry pi.
I don't necessarily agree with that, but it is a pretty great starting point. The competition has a very long way to go to be comparable. Performance is also seldom the limiting factor.
What are you using these days?
Personally I switched pihole and a few other servixes to a docker container on a USFF pc. Still using a pi4 for retro games
I'd like to know also. I've been looking to upgrade a Raspberry Pi 3 that runs a dashboard display and have been checking Raspberry Pi 4 stock, but if there's another device I should be using in the same MSRP range I'd like to have a look.
List 3 please. That is without a snark. I needed couple of pi-s for some personal project and it was EUR 100 or more per piece. Up to the point where cheap android rooted phone looked like a better option.
Which compelling offers did you have in mind? The similar boards I can get at 35$ are Orange Pi 3, Orange Pi Zero 2, MangoPi MQ-Pro, Le Potato. None of which are better than a Pi 4 (at least in my use cases).
Could I spend more for a better board like the RK3588 boards? Sure, I have a Rock5B and Orange Pi 5. But I don't want to spend more for performance I don't need.
To me, the frequent posts about RPi projects, sourcing, and the like that reach the front page of HN suggests otherwise.
What is your suggested favorite replacement? All the less that I’ve looked into are either just as hard to get and expensive or are discontinued before they gain ground. Happy to consider other options.
The RPis is ubiquitous for cheap and plentiful supply while not being the most performant. Their widespread availability and support makes them the de facto base model for almost all cheap electronics projects nowadays.
Software support.
Comment was deleted :(
I have a small stack of Raspberry Pi boards that I bought one at a time over the last few years. My issue with them is I could never buy three or four at once. I was always hit with the full shipping cost on something that weighs an ounce. As far as I know this has always been the case in the US.
I want to be able to buy a few at a time, like any other product.
ARM SBC entry costs have jumped upwards of $100 and low powered Intel machines (used thin clients, nucs, workstations) have become more accessible than ever. When the 4 was brand new, 4 cores/4gb was pretty killer for the entry price.
The Pi had an incredible impact but it's pretty obvious it doesn't fit very well into a lot of usecases any more given other options. I hope that over the next few years the Pi foundation can focus on its original goal for the Pi: a cheap and accessible device that could be used in schools to teach programming
Yes but when you compare the TDP and compute/watt of such a system compared to an arm sbc it pales in comparison. Being fan-less hence much more compact is another thing they lack as well.
NUC and other platforms get down to single-digit watts idle. I have a much larger, very performant Ryzen 7 system that idles at 30 watts with an Nvidia GTX 1070. The GTX 1070 alone is 10w idle and tiny (under 1 sec) 70 watt peaks for inference in my various uses (the fan doesn't even start). The x86_64 architecture has come a long way in terms of energy usage and even the much older NUCs, thin clients, etc can pretty easily get to the 10 watt range.
I'm not one for wasting energy but 30 watts at 24/7 usage at some of the highest energy costs in the world ($0.23/kWh) comes to roughly $5/mo and enables many, many, many things that just aren't fundamentally possible (or very ill-advised) with a Raspberry Pi.
For me personally I'm more in the $0.12/kWh range and I can run this system, with a crazy amount of functionality, for around $3/mo.
Yes a Raspberry Pi would be something like $0.26/mo but a few dollars a month is completely unnoticeable in a typical household utility bill. The level of your toaster or how long you microwave leftovers adds up to be significantly greater.
That's all fine if you're powering the thing off mains power. A lot of Pi use-cases involve running them on batteries, e.g. any of the Pi-powered "retro handheld" devices.
I'd say ADL e-core beelink esque smallboxes for $150 offer good value and decent tdp. No where near pi, but also not like the old Skylake thin boxes, either. We're talking probably sub 65w all in, maybe even closer to 35w.
> I hope that over the next few years the Pi foundation can focus on its original goal for the Pi: a cheap and accessible device that could be used in schools to teach programming.
FWIW I think they really are. In the middle of last year they deliberately pivoted both MagPi and Hackspace Magazine to projects involving the Pico, and they've done, IMO, a very good job in demonstrating that many of the educational projects around the Pi Zero are little more than simple Python (usually) scripts that toggle a few GPIO pins. Not only do they run just fine on a Pico or Pico W, it's arguably of more obvious educational value to see them running on such simple devices.
[flagged]
Please don't post like this here. It's not what this site is for, and destroys what it is for. You can make your substantive points without snark, flamebait, or name-calling, so please do that instead.
I keep reading how "it'll get better soon" but still can't buy them like I used to. Looking forward to "here is the link you can buy today for low price". Feels a little marketing PR otherwise.
I keep reading how "it'll get better soon" but still can't buy them like I used to.
Their messaging has been consistent going back to at least sometime last year, from what I can tell: things will be tight at least through the end of Q2 2023 (albeit improving towards the end of that range) and will get much better through Q3 and Q4, with normal (pre-pandemic) availability by sometime in Q4.
So far things seem to be tracking exactly along that path. Prices on the scalper-resold Pi 4's on Amazon started dropping precipitously about a month or so ago, and have dropped from north of $200 USD to around $120 USD (for the 8GB model). And rpilocator consistently shows far more availability of different models than it did a couple of months ago.
Clearly we aren't "out of the woods" yet, but one can absolutely start to see the lights of civilization in the distance if you look.
https://www.amazon.com/Raspberry-Pi-Computer-Suitable-Workst...
I feel like the Raspberry Pi foundation hasn't met its objective of computers being accessible to everyone (eg. children). The Pi itself is prohibitively expensive, and it looks like they cater to companies first.
Arduino and it open sourcing the hardware meant very cheap but less powerful SBCs. I hope to see something in a similar vein from them.
Until there was a chip shortage, they really did meet that obligation.
The decision that nobody likes that they had to take, surely, was to try to preserve the commercial ecosystem that exists around their products, not least because a) it helps them achieve the economies of scale that make the devices cheap, and b) so many of those commercial products have educational implications.
The Foundation is quite literally not responsible for the devices being prohibitively expensive, and indeed as far as I can see, when official resellers regain stock, they do so at prices very close to the pre-crisis prices, when you take into account rampant general inflation. The Zero 2W is going to be slightly more expensive. The Pi 4B seems to be at the same price.
And the Foundation's goal is more about access to computing _education_; I would argue they are still achieving things on that level, because many of their activities are more general than the Pi.
> And the Foundation's goal is more about access to computing _education_
A couple of months ago I attended a talk given by Eben Upton. He talked through the history of the entire RPi initiative, starting with his observation that Computer Science admissions at Cambridge University were falling away at quite an alarming rate, and how that inspired the whole thing.
Before attending the talk, I had understood the proposition to be that a cheap home computer put into the hands of kids might spark some natural curiosity, like what happened during the 80s and early 90s home microcomputer era, which would then ultimately lead to more tech-literate kids who would go on to study Computer Science and ultimately have careers in the field.
I'm not sure this was wrong, but it definitely wasn't the whole story.
Eben described the work of the foundation, and it's far more extensive than I had any appreciation for whatsoever. For example, both domestically and internationally the foundation are heavily involved in teacher training, and have played a role in developing curriculum across the globe to further or even initiate IT/CompSci education.
The structure of the Foundation and Trading, he explained, was the most appropriate vehicle under UK tax rules for funding charitable educational work with commercial enterprise. I'm probably simplifying this massively but in essence the flow of profits/dividends/whatever up from the for-profit Trading arm to the charitable Foundation helps to pay for the significant amount of educational work that Foundation does around the world.
I hadn't appreciated many of the forms of educational work that the foundation are involved in at all.
From this perspective, it is perhaps easier to hold a charitable view of Eben's comments around preferential stock availability for small businesses. In the position that they have been in, whereby they're going to sell every unit produced easily whatever the channel, and keeping in mind funding the education goals of the foundation, I think it's entirely reasonable to also help keep some small businesses afloat along the way at the expense of hobbyists.
Prior to this talk, I think my blind spot had been that the entire mission was to put cheap home microcomputers into the hands of kids around the world. I can understand that throttling this home or hobbyist supply in favour of commercial sales takes a bit more to interpret charitably.
But the Foundation is doing so much more than that, and that was entirely lost on me, that failing to fulfill hobbyist market demand no longer looks to me like failing in the original mission.
Thank god! I’ve been so frustrated I couldn’t get one of these and then put it in a drawer and immediately forget about it.
Same. I used to have an alert to notify me when pretty much any Pi was in stock. I’d purchase and stick them in a drawer. I have a drawer full of Pi’s that have never been turned on.
Got any 8gb CM4? I'll buy one from you...
I haven't had many problems buying one if I need one, but Buy a PI's Canadian store is selling them for $185, which is a little much. At least they throw in a case, some heatsinks, a power supply and some cables.
For those who have been having problems getting one, is it because they've been hoarding them? I'd really like to know what people are doing that needs more than two or three Pis.
> Buy a PI's Canadian store is selling them for $185
> For those who have been having problems getting one, is it because they've been hoarding them?
They retail for well under $100. You’ve been buying from scalpers who are jacking up the prices.
MSRP-priced retailers haven’t been able to keep stock in years, and often sell out within minutes of getting inventory.
Some people just want their first or second RPI, but aren’t willing to pay scalpers. I don’t know how many people are hoarding them, but I know some people prefer them to arduinos and more simple microcontrollers. Many people are more familiar with Linux development instead of embedded development.
When they were cheap $35 boards, the advice was buy one per project.
> scalpers
Overpriced or not, they are an official reseller according to the RPi website.
I haven't been able to buy one in New Zealand for years now, at any price.
To be fair, isn't that pretty normal for electronics in AU/NZ? I've been watching Hardware Unboxed for years and they're always lamenting how hard it is to get parts to review.
No, that's not accurate.
Excellent news. I just bought a Fujitsu thin client for €18.00 to scratch the Raspberry itch but will definitely try to get my hands on a Pi 4.
Can you run Linux on a thin client? Also where did you buy them for so cheap?
As far as I understand, most thin clients nowadays are just PCs, usually AMD embedded SoCs with very good support for Linux.
No problem running lightweight distros on them, though the really thin clients tend to be underpowered for web browsing. Available very cheaply on ebay, and pretty much all support remote boot so easy to run diskless (assuming some minimal local server).
This person has had success with the idea: https://weisser-zwerg.dev/posts/dell-wyse-fujitsu-futro/
> Can you run Linux on a thin client?
Yeah, these originally came preloaded with eLux.
> Also where did you buy them for so cheap?
Second hand on Ebay, I'm located in Europe.
After the only ones being low spec for ages, there do seem to be some higher spec ones becoming available - I nabbed a Pi4 8Gb yesterday. No sign of Zero 2Ws yet though
2Ws require a specific fabrication process that will take a while longer to ramp up (due to the RAM-in-chip)
For those that love the pi and electronics I’d suggest checking out the raspberry pi pico which is not a full os but a microcontroller with tons of features including WiFi that runs Python or c scripts. Replaced the pi for me in the projects I was working on, especially paired with a nice LED matrix display. My favorite combo at the moment is this led matrix with presoldered pico w: https://shop.pimoroni.com/products/cosmic-unicorn
I wonder how many people have several older models of Raspberry Pi sitting in their drawers and collecting dust, as selling them seems to be too much of the hassle?
I have a few of the model B version that was released in 2012, somewhere in my basement. I’m not even sure how usable these would be today.
I'm using several original BCM2835 Raspberry Pis and Pi Zeros (original) to compile pkgsrc packages for NetBSD on earmv4 and earmv5 because the newer ARM CPUs won't run anything below earmv6. There are some instructions the newer CPUs don't support that'd need to be emulated.
With a USB attached SSD, they're quite reliable and run for many months at a time, compiling non-stop with no issues.
Yeah anything older than the Pi 3 with armv7 is also more or less unusable in terms of OS deps, since there just aren't many package releases being maintained for v6 and below.
I'm using them precisely to make packages available for them, so I don't think it's fair at all to say they're unusable.
The OS is fully, 100% supported, so we're not even talking about that. See:
http://nycdn.netbsd.org/pub/arm/
The number of packages available is decent and is only improving:
Ah I'm not talking about NetBSD there specifically, mainly Ubuntu and Debian since that's what I typically use. But in general the OS support isn't very good.
You're replying to a post about NetBSD ;)
Your statement, "But in general the OS support isn't very good," would be better framed as "But in general the Linux OS support isn't very good," because OS support for pre-earmv7hf and pre-aarch64 is excellent in NetBSD.
Yep, I think we understand each other now.
The original one with 256MB of RAM? The later models upped that to 512MB. I ended up breaking the full-sized SD card slot on mine.
The Pi 4 is really only a moderate step up in performance from that model.
Ars Technica had an article about this a few months back, used thin-client PCs can fill in some use cases https://arstechnica.com/gadgets/2022/11/used-thin-client-pcs...
I couldn't get them for use as ambiance devices so I bought some cheap Lenovo Ideapads, they work great.
There are also the ThinkSmart View devices: https://taoofmac.com/space/blog/2023/04/22/1330
Those look cool, I wasn't aware of them. I feel more comfortable with the laptops since they usually undergo a grueling test cycle at Lenovo. I hook them up to an external monitor which hides their presence and use a wireless air mouse to control them.
Any rumours of a 5?
Alas, no.
“Don't expect a Pi 5 five next year,” Upton said [in December 2022], explaining that the Pi team sees 2023 as a year to spend “recovering from what just happened to all of us [COVID, supply chain issues].” https://www.theregister.com/2022/12/21/no_rpi_5_2023/
The raspberry pi becomes even more outdated....
That was never a problem for them.
I'm wondering why other manufacturers don't look like impacted as much from the covid supply chain disruptions as they pose to be in every given chance.
I think it has to do with the fact that the majority of their product line is based on a few very specific SoCs that are quite hard to replace so fast.
No need. The CEO said (in an interview with the "explaining computers" YouTube channel) 2023 will be about fixing supply of existing models.
It's reasonable to expect the 5 to arrive in 2024 at the earliest. Personally, I wouldn't expect super fast m2 ports and whatnot.
Frankly, at this point I am ready to spend a bit more and get a proper 5800u mini pc or something. Waiting for AMD's Zen 4 APUs to hit the market. They are not going to be cheap, though
I really wish they would decide not to do it at all.
Put all the high-performance work into the CM units of different specifications. Make (or encourage the market to continue to make) mainboards that take those units. Maybe certify a couple of them for the Foundation's use.
And then focus on making microcontroller-level chips at different performance levels, up to an "alternative home computer" device. Something small, which can be learned completely. Enough power for something equivalent to a Minix or a CP/M or RISC OS. The RP2040 is almost there anyway.
I have always been slightly annoyed that the RP2040 didn't feature a RISC-V core to being with, with a 2nd gen that includes the privilege mode specification. RP2040 as it stands is nearly capable of being "learned completely" and providing something like an OS, but having a hardware MMU really makes a different experience. Of course, the ARM architecture has an MMU spec too, but I've got a soft spot for RISC-V.
I wouldn't hold my breath for small AMD based boards. For example PC Engines (they make the APU and ALIX boards) has stated that AMD is increasingly difficult to work with as a small manufacturer. And all their old SoC's have been EOLed.
> It's reasonable to expect the 5 to arrive in 2024 at the earliest
My expectation is 2025 at the earliest, tbh. Mostly based on the CEO's repeated statements, and the complete lack of rumors about any developments. If a new lineup of PIs were going to launch next year, they'd have to be pretty late in development already.
I'd be happier if they ditched pi 4 production and just released a 5 with whatever SOC manufacturer has the best availability, hopefully that isn't Broadcom.
There are other SBCs with SoCs from other manufacturers out there. They've been out there for a decade. Think BeagleBoard or Wandboard.
Sure, I own a few pine64 devices and an RPI4 too. I would like the raspberry PI community to synchronize on a manufacturer like rockchip for a while. I simply never see a useful broadcom chip in the wild that isn't in a RPI and the RPIs omit features I want.
Broadcom doesn't put chips "in the wild", which leads you to the paradox that is RPI. It was a scrap chip that Upton made into a cheap educational computer, the rest of you ran off with it as a supercheap COTS computing module and now nobody can figure out how to ship their products without it. Or, most distinctly, nobody wants to believe that the RPI isn't subsidized and all SoCs are this cheap.
Broadcom doesn't deal with small fish, much less hobbyists. I've dealt with their products working for a large product maker and it wasn't any better on that side of the fence.
Are you saying that you think Broadcom are still subsidising the Pi? Even the Zero 2W, which has a custom chip designed by Raspberry Pi?
I mean, as I understand it, they ship about half a million CM4 units alone every four to six weeks, and they sell Pi hardware to businesses like Citrix: why would Broadcom need to/be willing to subsidise chips at that scale?
I'll believe it when I see it
[flagged]
Sorry, but the company's decision to hire a former cop (nothing wrong with that), then glorify his surveillance work (not great, but not the worst thing in the world), then pretend that the people upset with that were actually vegans who were upset with a picture of bacon or some bull like that convinced me that the company is completely, utterly out of touch with normal people and with the zeitgeist of the hobbyist movement.
They can go act like they're above reproach with their new "business" community all they want. Good riddance.
I'll use what Raspberry Pis I already have, but I'm happy to buy Orange Pis and Nano Pis for other uses.
Crafted by Rajat
Source Code