> "...: It sounds like the key feature will be 'more': a faster CPU and faster IO, rather than new features."
Raspberry Pi Holdings is a embedded systems manufacturer for pity's sake; we don't need more from them, we need less. [EDIT] A faster Raspberry Pi 6 is encroaching on the territory of the Intel N150 and its successors and mainstream Linux distributions and that is a battle they would lose in terms of price and performance.
Give us a Raspberry Pi Zero 3W with proper sleep states to reduce sleep power consumption, lower idle power while awake, and 1 GB of RAM even if it doubles the price.
^^^ when I tell people tangential to the field that the latest pi needs considerations of cooling solutions and a beefy power supply (no more just any old micro usb cable into any old usb port), they're astonished. It was a "microcontroller" you could program in Python with a friendly Linux environment and is now an expensive, power hungry, hot computer with a microcontroller hanging off of it
I agree that Raspberry Pi is not a good general purpose computer, but some of these criticisms are starting to feel like a pile-on with partially incorrect information.
> the latest pi needs considerations of cooling solutions
FYI you can run the Raspberry Pi 5 without a fan or even a heatsink. It will safely throttle itself if it gets too hot.
If you're trying to get maximum performance out of it all the time, you will want a heatsink and fan. If you want to run some Python scripts in a Linux environment or even if you're doing heavy work and waiting longer is not a problem, you don't need extra cooling.
> and a beefy power supply (no more just any old micro usb cable into any old usb port)
This hasn't been true in 10 years.
Powering something off of any old USB port means it would have to fit within the 5V 500mA basic specification, which the Raspberry Pi 3 exceeded long ago.
> It was a "microcontroller" you could program in Python
It was never a microcontroller by any definition of the word.
> > It was a "microcontroller" you could program in Python
> It was never a microcontroller by any definition of the word.
I think the poster means people treat the R-Pi like an MCU that runs Python. The Arduino was popular at the time the Pi came out but limited. Once the Pi landed, it quickly filled the gap and the Arduino's popularity diminished to the point where it's now a corporate Pi clone.
> Powering something off of any old USB port means it would have to fit within the 5V 500mA basic specification, which the Raspberry Pi 3 exceeded long ago.
Theoretically, devices like the iPhone have lower power consumption than that and loads of performance features like recording 4k 120fps video.
Of course, an iPhone costs much more than a RPi, and has much better economies of scale, so they’re not truly comparable.
> FYI you can run the Raspberry Pi 5 without a fan or even a heatsink. It will safely throttle itself if it gets too hot.
What's the point of doing so though? If you're doing this, you're obviously using the wrong device. If all you need is to run some python scripts in a Linux environment, you should use a Pi 3 or Pi 0w2.
Even for extended workloads, a Pi 5 without a heatsink is still a hell of a lot faster than a Pi 4. And as sibling says, most users appreciate bursty speed while not doing prelonged compute (see also fanless laptops).
(Disclaimer, I work for raspberry pi ltd, not views of employer etc.)
Most light workloads are very bursty. When you type a command or click on something you want latency to be low. Having the overhead to get it done quickly at the full clockspeed is good if you are latency sensitive.
Throttling has become a bad word. Some feel compelled to avoid it at all costs, doing things like buying big coolers and running synthetic benchmarks to avoid it. Unless you're doing sustained workloads where you need all of the performance, allowing a little throttling is fine.
If you're consistently doing bursty loads that throttle, the thermal cycles are going to do some damage to your chip and it'll fail sooner than if you ran it cooler.
I've been doing some heavy SDR lifting with a couple of my Pi 5s, and my own experience is that the active cooler works extremely well, and more often than not the fan can be shut off and it will work well as a passive cooler.
> It was a "microcontroller" you could program in Python with a friendly Linux environment and is now an expensive, power hungry, hot computer with a microcontroller hanging off of it
The Pi project was never originally a microcontroller - it was always a full-blown SBC you could program any way you want with some GPIO pins attached. People literally used them as (slow) home computers.
The company didn't sell its first microcontroller until years later in 2021 with the Pico, by which point we already had Pi 4. I do though think its a real shame prices for the SBCs have risen as they have.
At one point, the raspberry pi was a decent option if you wanted something hobbyist friendly that could toggle GPIOs and connect to the Internet (and later Bluetooth).
I suspect Espressif has mostly taken over that market now
It's still true that people, out of convenience and familiarity, used Raspberry Pi for tasks where a microcontrollers would have been perfectly adequate
There was definitely usecase overlap due to the presence of the GPIO, but huge numbers of Pis ended up doing things a microcontroller can't - stuff like the PiHole and Retropie projects, and never used their GPIO pins at all.
Thinking of any of the early Pis as microcontrollers ignores a huge amount of the ways in which actual end users interacted with the thing, and even the way it was sold and marketed. Upton was trying to replace early hacker-friendly home computers like the BBC Micro/Apple II, for a new generation.
This just clarified something for me. I've always been annoyed when I see a Pi with nothing connected to its GPIO header; why not just use a cheap thin client? Or an old laptop, for that matter? But that's missing the point. Here's the point:
Pre-Beagleboard-and-Pi, if you wanted an programmable thing to work with GPIO, you used an Arduino or a BASIC Stamp, or just a plain old PIC. But they wouldn't run a real OS.
Pre-Beagleboard-and-Pi, if you wanted an embedded Linux box, you used a WRT54G or a Soekris or an old laptop. But getting GPIO out of them was a PITA. (And often involved lashing an Arduino to the side.)
The Beagleboard (released in 2008), could finally do both. It had gobs of I/O and first-class support for it under Linux. It was pretty affordable. Then the Raspberry Pi came out in 2012, with a similar amount of GPIO, but demolished the price point to where it made sense to use it in place of a microcontroller.
That's really the magic of the Pi. You can keep one cheap gizmo around, and use it to solve (a large fraction of) two classes of problems. It doesn't fully replace everything a PIC or a PC can do, but it replaces an awful, awful lot of them.
On the other hand, the RP2350 actually is a microcontroller, and IMO a nice one for many purposes. PIO, high-quality datasheet, nice ecosystem, etc. And the Pi Zero 2(W) can do most things the Pi/Pi 2 could, with a smaller footprint and less power consumption. Variety is nice.
Do you have a reference for this? Looking around, I see it being beaten by other ARM SBCs, and even low end Intel devices.
Many years ago, I measured performance per watt of the original Raspberry Pi when they were still relatively new. The performance per watt lagged behind even a beefy Intel box since the original Raspi was so slow that it destroyed any gain it got from using so little power.
They've never been particularly low-power, in a performance-per-Watt sort of way compared to other offerings at whatever present day. In recent times, I've seen completely-believable reports that N150 boxes walk all over it.
They've never been particularly cheap, in a performance-per-dollar sort of way. Used machines from eBay, yard sales, and old broken laptops (that still compute!) have always been better. (They usually come out OK when new is compared to new, though, which is IMHO the only valid comparison.)
Those comparisons were never very favorable.
---
The parts where it shines are: Small; they come in two sizes, and those sizes are small and smaller. That was new, but it's been cloned all over the place.
Built-in GPIO that's meant for people to actually-use and tinker with. That was new, too, but it's also been cloned. (Also: These days, anyone can plug a cheap Pi Pico into any PC with USB and get a fairly intense amount of GPIO to goof with.)
Standardization, and the appliance-like behavior this can enable. Lots of folks, including kids of all ages, just download pre-built images and swap SD cards like they would cartridges in a Nintendo. That's not for me, but it's pretty neat.
Community. They've still got a lot of momentum by being first at these roles. That's good. It helps newbs (who at this point may have never had anything resembling a "real computer" to play with ever before in their lives) to get started.
Low-power enough: It's not ideal when chasing tiny Watts for battery or small solar power, but you probably won't notice it on an electric bill (and despite the bizarro-world cooling rigs people put together, a passive heatsink really is good enough to keep it running in-spec).
raspberry pi has terrible power management as well. turning off a core was impossible on previous generations for example. a few years ago rpi was the worst of all sbcs we measured for battery powered usages. this was for an actual embedded product so it wasn’t going to be used either way (ask broadcom for some chips HAHA)
but there is very little reason to use a rpi over other sbcs if you have a remote idea what you’re doing beyond hobby use
Performance per watt is nice but I’d be more inclined to talk about “problems being solved per dollar.”
If you don’t specifically have a project where you need the GPIO pins built in, I struggle to understand the use case proposition of a raspberry pi compared to a typical x86 mini PC or even just grabbing a think client desktop like a ThinkCentre.
Almost everything that is unique to a Pi compared to an x86 mini PC seems like it makes more sense with an ESP device.
When the Raspberry Pi was $35 and it ran a desktop OS and the cheapest alternative that did that was 5x the price that use case made sense.
Maybe a tick-tock release cycle (one with new features and some speed, the next with the ~same features and more speed) is where they're headed, and maybe that makes sense. They wouldn't be the first.
I'd love to see even-lower-RAM versions, though. Most of what I use Raspberry Pis for at home for is not RAM-hungry at all.
My Pi4 network router has 2GB because that was the smallest/cheapest version at release when I got it, but the system itself consistently only uses about 64MB of RAM. It'd do perfectly well and have a ton of breathing room with just 128MB of RAM (which will never happen, but if it did happen...).
I suspect the Pi4 that I use as a set-top box with Kodi would be fine with 512MB.
I've used Zero Ws for all kinds of things over the years and never felt RAM-starved with their little 512MB of RAM.
So I'm learning towards 512MB.
But sure: 1GB options would also be fine even if it does double the price. Our comments serve to demonstrate that there's room in the marketplace for different SKUs with different memory capacities. :)
Ideally each RPi generation should keep the same price (or lower now that it's gotten so high) but with better performance. If they can't do that they just shouldn't create a new generation.
Raspberry Pi isn't in direct competition with N150's.
Their niche is the industrial/embedded space. For that market, power consumption doesn't matter. What matters is that each model is guaranteed to be available till a specific date.
Among the people making things like a DIY NAS, who want fast USB, lots of cores and RAM, small-ish, not-too-bad power consumption, running Linux; and not caring much about GPIOs or passive cooling, it’s in competition with the N150
Among people who want GPIOs and network connectivity, a low price and an open, microcontroller like experience, not caring much about USB speeds and lots of cores and running Linux and suchlike, it’s in competition with the esp8266 & esp32. And the previous generations of RPi.
> Raspberry Pi isn't in direct competition with N150's.
They may not have intended to be in direct competition, but in the current crisis conditions they are priced about the same as equivalent RAM/storage N150s, have even worse supply issues than the N150s, and have worse performance/watt than the N150s.
Its mighty hard to recommend them for new projects at the moment (nor any of the Pi clones, which are also rocketing in cost and dwindling in availability)
Is there some serious astroturfing going on with the N100/N150, or am I just jaded?
I have a bunch of old intel atom boards laying around. The Intel Compute Stick (TM) burnt out its flash root drive in a few months. The C2000 board I had burnt out the clock pin to drive the bios. I have a Clover Trail with a PowerVR GPU (I thought I was getting an intel GPU because it was branded Intel Graphics or similar, but nope!) that lost Windows support very quickly after launch, and has no GPU drivers for any other OS.
Instead of being fooled 4 times in a row, I looked into using an N150 for a NAS, but this time I held off a bit until after launch so I could research it first.
Lo-and-behold, they all have crazy PCIe / memory subsystem data corruption issues. I guess there are some chicken bits for the OS developers to set if the kernel can stay up long enough after boot without a panic.
Why would anyone buy this for a NAS / embedded use case?
All CPUs from all vendors have tons of bugs like this, which are mitigated in the operating system kernels, e.g. in the Linux kernel.
I am pretty certain that the Linux kernel must also contain specific code for various quirks of all Arm CPUs that have been used in the various Raspberry Pi models.
Intel had indeed several bugs that were more ugly than usual in their recent CPU models, like also MONITOR not working correctly in Lunar Lake, but even so, Intel still has better documentation for their CPU bugs than most vendors of Arm-based CPUs.
In any case, the bug that you linked was solved in the kernel years ago and it affects a privileged instruction that cannot be used in user programs. It does not have any direct relationship with memory and PCIe corruption.
I've been running an N100 for 3 years with a 5 bay external enclosure over USB 3.2 Gen 2 and ZFS, and have not had any issues. It is pretty phenomenal, pulls about the same power, and costs around the same as an RPi 5 but provides substantially more compute and throughput.
I think exactly the opposite: we have no shortage of embedded crap we can buy; what is useful is dismembering intel. It would be better if the pi were risc v but this will do for now.
I have bought an rpi at every generation. And I still have yet to find an actual use for them.
Everything they do from a compute perspective is just better with a mini pc or old laptop with a mobile spec chip.
Everything they do from a programmability perspective is just better with a microcontroller specific to the task.
I just don't see the actual market position for these things. They were supposed to be a cheap board, but you can't actually buy them cheaply because the vendors upcharge so much.
Rapid prototyping: I created a PoC to take a webcam snapshot every half hour and upload it to a server in an afternoon. Freelance project. Could it have been done with a microcontroller? Yes, but not in 4 hours.
Local digital displays in a gym: we built a system with a number of overhead 60" TVs, each with a pi on a VESA mount to show scheduling and workout information in a gym for a client.
HVAC controller: bid on a project where the customer's original concept was a Pi managing a rooftop HVAC system for large buildings. They outgrew the pi and wanted a new solution.
Data aggregator: collecting sensor information via BLE (bluetooth) and uploading to an internet server over Ethernet.
Remember that "cheap" to a consumer doesn't have the same definition for a business. To most profitable businesses, a $100 computer that fits on the back of a TV and consumes a fraction of its power is cheap. In fact, one of the reasons that Pi's were so hard to find for a while is that the Raspberry Pi foundation was prioritizing industrial/commercial customers over hobbyists.
I think I found a use case, but would love to be proven wrong with some faster / cheaper approach:
I've tired of buying S1 compatible sonos speakers on ebay, so I'd like to build a speaker enclosure with a WiFi device that has a high-quality DAC and the ability to use pipewire or similar to do real-time DSP and multi-room audio sync.
A RPI + third party hat should work well for this, or so I am told.
I think at this point the brand reputation and software quality are a big selling point.
If you're trying to build a couple of units of some embedded thing where you need to toggle some GPIOs or serial devices in response to requests over the network, but don't have the expertise or resources to do it with a microcontroller, a Pi is a great option - you know you'll have software support, and you know that the vendor will be making the exact thing you bought for 5-10y.
For hobbyist stuff at home, I agree, though. A mini PC is probably better for homelab stuff, and an RP2350 or ESP32 is probably better for anything embedded or battery powered that you want to do.
Development speed is also one of the forgotten axes. I mentioned upthread a system I built that was done in less than a day. There's no microcontroller solution I know of that would have let me deliver it that fast.
> I have bought an rpi at every generation. And I still have yet to find an actual use for them.
It's amazing how well these fit into the category of products that people feel compelled to buy, play around with, and then forget about.
Flipper Zero is another product that landed in the same space.
What's sad is that Raspberry Pi does have a lot of legitimate use cases and people who want to use them, but the supply has always been swamped by all of the demand.
* Replacement controller for my UFO Catcher - It has WiFi, easy to update, and I can operate the machine remotely with it. It's bolted to the back of small touchscreen that lets me change the machine settings as well.
* Remote printer access - I can monitor from the USB cameras and gather statistics about the prints.(I suspect a lot of 3D printing enthusiasts use them for this purpose.)
Having a small low power computer has been useful for me in those instances.
Obviously the business model is selling a $30-100 SBC to jerks on Hacker News who simultaneously brag about buying literally every single model, but apparently have no use case for them. Even better, these schmucks will never plug the product in, never submit a warrantee, and never harass support. Brilliant model!
The 8GB Pi 5, at $170 [1], is encroaching on Jetson Orin Nano Super's $240 price point [2]. But the Jetson has a faster CPU (newer a78ae cores rather than a76) and, obviously, a whole-ass GPU.
Nvidia's software platform for the whole Jetson series was, at least in my experience, absolutely awful on the Jetson Nano and Orin boards I worked on. Has that improved at all? I did not appreciate that the only option they provided was a full desktop version of ancient Ubuntu... and even flashing the OS image was a bizarre process.
Edit: looks like they at least have a better headless option now.
Nowadays upstream Linux with UEFI mostly works, with their out of tree drivers. I’ve managed to make it work in NixOS with the stock kernel. Look at the open embedded L4T project, they have some recipes for building that. No need to use nvidia’s kernel anymore!
Also, supposedly on the second half of 2026 they were going to be moving even more stuff out of their Jetson-specific drivers as they already do for their slightly newer chips (so you could use the standard drivers, and standard CUDA builds). Let’s see how that turns out.
Yeah, one of my bigger complaints especially on the Nano was the GPU only had really limited model support (iirc, mostly tflite but maybe I'm misremembering) and it sounds like the newer ones are more normal. That and what seems from the docs to be better headless support would be major improvements. Going further to mainline distro support would make them interesting to me again.
I was always disappointed by the Nano as it was a pretty capable device, but it seemed like not many people picked it up as a platform for cool things which I always attributed to the software.
A very small point, but pulling from a feather form factor BOM to compare.
$0.12 for microUSB female connector (rated 1A)
$0.26 for a USB-C female (rated 3A). Needs 2 x resistors (< $0.01), 20% larger board area
I think the power capabilities are the biggest item. If you want to pull higher current from a laptop for development or supply from a wall, you have to switch to USB-C.
I don't think either of these prices are that aggressive - pretty sure the cost comes down at volume.
I wonder if it would be worthwhile for them to produce both. Well, it will be hard to compare because the design cost doesn’t show up in the BOM, haha.
But it seems like it would be useful nowadays, since some laptop have mostly USB-C connectors, and USB-C to USB-C is pretty common. I’ve never seen a C to Micro. Do they even exist?
I have an unfair bias because I design PCBs as a significant part of my job, and switching out to USB on this board appears to be a non-issue.
I have a Pico in front of me, and there's plenty of room there for a USB-C footprint and the two 5.1k resistors. Given that, I cannot reasonably agree that the "design" stage is significant.
In other words, it's a change that I would make to my own board in 2-5 minutes because the stakes are low. My ballpark guess is that such a change at RPi would have to go through a proposal stage, a PCB change review, and then there would be dozens of places to update documentation.
Since backwards compatibility is non-optional, this would result in a separate SKU, which means that the whole distribution chain needs to be updated with a new product.
So, I acknowledge that when you're working at their scale any change like this is the definition of non-trivial. What I don't agree with is the conclusion that it's not still clearly the right thing to do.
For what it's worth there are third-party rp2350 boards with USB-C connectors if that's important to you. Heck, WaveShare has one with two USB-C connectors: https://www.waveshare.com/rp2350-usb-c.htm
I do think that you're missing my point, which is that we're significantly past the point in this wretched timeline where they should offer a USB-C version of the reference board for this MCU.
Would love to see actual security focused hardware/software features, like full OP-TEE, fTPM (or a more ideally a real physical TPM), and similar. For example, so that the OTP isn't the only way to store a disk encryption unlock key.
The existing secure boot mechanisms aren't bad, but allowing for more than one public key hash in OTP would be nice, too.
These kinds of things are expected to be on modern embedded SOCs and SOMs now.
You do know those are trivially bypassed with a signal processor, right? If physical access is outside your threat model, that's OK, but it makes (for example) the forced Win11 upgrade for DRM^H^H^H boot integrity enforcement seem ridiculous.
I recently found out about the Radxa Dragon Q6A. A Qualcomm chip with faster CPUs, a good GPU, a DSP and AI accelerator, and a hardware video encoder seems very compelling. It even supports Windows if you want that for some reason.
I run the media lab at one of Europe's must prolific art universities. The variant I tend to use most is the 3B+.
Reasons:
- full sized HDMI connector
- headphone connector
- good bang for the buck
If I had one wish for any new product in the Raspberry line it would be: Do the Raspberry Pi 3++ or something. Same thing. Faster, but with USB-C power connector, 4K Video resolution, 2× USB-C I/O, 2× USB-A peripherals and maybe M.2 support.
Same here. The 2b, 3b and Zero 2 have the best cost-benefit ratio. I recently migrated the Oberon System 3 to those three boards and particulary the Zero is a great host for this bare-bone system: https://github.com/rochus-keller/oberonsystem3native/. I hope the Raspi company postpones the end of life date of the 3b and Zero 2.
As someone who has just had to migrate a project off of radxa due to ongoing supply issues... good luck getting your hands on more than one or two samples of those :/
The only way I'll buy another raspberry pi is if they come with a power supply that's guaranteed to work with them. I got tired of the random reboots in the night and replaced my media center/NAS with an old Nuc.
Official power supply is sold separately. Pis work fine with unofficial USB-PD power supplies but will limit maximum current for USB devices. I have two Pi 5's and one 4 and they all are perfectly stable when powered with an old Lenovo laptop charger.
It's still a huge pain point. I keep a magic "correctly sized, but provides clean power" USB charger with my Pi in a drawer, because I know that, if I lose that charger, I'm buying a replacement. It's unclear to me why they're so borderline. Maybe USB-C would help?
> "...: It sounds like the key feature will be 'more': a faster CPU and faster IO, rather than new features."
Raspberry Pi Holdings is a embedded systems manufacturer for pity's sake; we don't need more from them, we need less. [EDIT] A faster Raspberry Pi 6 is encroaching on the territory of the Intel N150 and its successors and mainstream Linux distributions and that is a battle they would lose in terms of price and performance.
Give us a Raspberry Pi Zero 3W with proper sleep states to reduce sleep power consumption, lower idle power while awake, and 1 GB of RAM even if it doubles the price.
^^^ when I tell people tangential to the field that the latest pi needs considerations of cooling solutions and a beefy power supply (no more just any old micro usb cable into any old usb port), they're astonished. It was a "microcontroller" you could program in Python with a friendly Linux environment and is now an expensive, power hungry, hot computer with a microcontroller hanging off of it
I agree that Raspberry Pi is not a good general purpose computer, but some of these criticisms are starting to feel like a pile-on with partially incorrect information.
> the latest pi needs considerations of cooling solutions
FYI you can run the Raspberry Pi 5 without a fan or even a heatsink. It will safely throttle itself if it gets too hot.
If you're trying to get maximum performance out of it all the time, you will want a heatsink and fan. If you want to run some Python scripts in a Linux environment or even if you're doing heavy work and waiting longer is not a problem, you don't need extra cooling.
> and a beefy power supply (no more just any old micro usb cable into any old usb port)
This hasn't been true in 10 years.
Powering something off of any old USB port means it would have to fit within the 5V 500mA basic specification, which the Raspberry Pi 3 exceeded long ago.
> It was a "microcontroller" you could program in Python
It was never a microcontroller by any definition of the word.
Raspberry Pi foundation has released microcontrollers that run MicroPython in a very user-friendly format https://www.raspberrypi.com/documentation/microcontrollers/m...
> > It was a "microcontroller" you could program in Python
> It was never a microcontroller by any definition of the word.
I think the poster means people treat the R-Pi like an MCU that runs Python. The Arduino was popular at the time the Pi came out but limited. Once the Pi landed, it quickly filled the gap and the Arduino's popularity diminished to the point where it's now a corporate Pi clone.
> Powering something off of any old USB port means it would have to fit within the 5V 500mA basic specification, which the Raspberry Pi 3 exceeded long ago.
Theoretically, devices like the iPhone have lower power consumption than that and loads of performance features like recording 4k 120fps video.
Of course, an iPhone costs much more than a RPi, and has much better economies of scale, so they’re not truly comparable.
> Theoretically, devices like the iPhone have lower power consumption than that and loads of performance features like recording 4k 120fps video.
iPhones use cutting edge (expensive) manufacturing processes for their chips.
They also have batteries inside to cover the power demands while doing intense work like recording 4K 120fps video.
I don't understand what point you're trying to make. The Raspberry Pi is not in the same class of device as an iPhone.
An iPhone will absolutely draw more than 2.5w under load.
> FYI you can run the Raspberry Pi 5 without a fan or even a heatsink. It will safely throttle itself if it gets too hot.
What's the point of doing so though? If you're doing this, you're obviously using the wrong device. If all you need is to run some python scripts in a Linux environment, you should use a Pi 3 or Pi 0w2.
Agree with your other points.
Even for extended workloads, a Pi 5 without a heatsink is still a hell of a lot faster than a Pi 4. And as sibling says, most users appreciate bursty speed while not doing prelonged compute (see also fanless laptops).
(Disclaimer, I work for raspberry pi ltd, not views of employer etc.)
Having the burst performance is helpful.
Most light workloads are very bursty. When you type a command or click on something you want latency to be low. Having the overhead to get it done quickly at the full clockspeed is good if you are latency sensitive.
Throttling has become a bad word. Some feel compelled to avoid it at all costs, doing things like buying big coolers and running synthetic benchmarks to avoid it. Unless you're doing sustained workloads where you need all of the performance, allowing a little throttling is fine.
If you're consistently doing bursty loads that throttle, the thermal cycles are going to do some damage to your chip and it'll fail sooner than if you ran it cooler.
I've been doing some heavy SDR lifting with a couple of my Pi 5s, and my own experience is that the active cooler works extremely well, and more often than not the fan can be shut off and it will work well as a passive cooler.
> It was a "microcontroller" you could program in Python with a friendly Linux environment and is now an expensive, power hungry, hot computer with a microcontroller hanging off of it
The Pi project was never originally a microcontroller - it was always a full-blown SBC you could program any way you want with some GPIO pins attached. People literally used them as (slow) home computers.
The company didn't sell its first microcontroller until years later in 2021 with the Pico, by which point we already had Pi 4. I do though think its a real shame prices for the SBCs have risen as they have.
At one point, the raspberry pi was a decent option if you wanted something hobbyist friendly that could toggle GPIOs and connect to the Internet (and later Bluetooth).
I suspect Espressif has mostly taken over that market now
Yeah every "hardware hacker" I know has pile of ESP32 at home now instead of pile of raspberry pi's.
It's still true that people, out of convenience and familiarity, used Raspberry Pi for tasks where a microcontrollers would have been perfectly adequate
There was definitely usecase overlap due to the presence of the GPIO, but huge numbers of Pis ended up doing things a microcontroller can't - stuff like the PiHole and Retropie projects, and never used their GPIO pins at all.
Thinking of any of the early Pis as microcontrollers ignores a huge amount of the ways in which actual end users interacted with the thing, and even the way it was sold and marketed. Upton was trying to replace early hacker-friendly home computers like the BBC Micro/Apple II, for a new generation.
This just clarified something for me. I've always been annoyed when I see a Pi with nothing connected to its GPIO header; why not just use a cheap thin client? Or an old laptop, for that matter? But that's missing the point. Here's the point:
Pre-Beagleboard-and-Pi, if you wanted an programmable thing to work with GPIO, you used an Arduino or a BASIC Stamp, or just a plain old PIC. But they wouldn't run a real OS.
Pre-Beagleboard-and-Pi, if you wanted an embedded Linux box, you used a WRT54G or a Soekris or an old laptop. But getting GPIO out of them was a PITA. (And often involved lashing an Arduino to the side.)
The Beagleboard (released in 2008), could finally do both. It had gobs of I/O and first-class support for it under Linux. It was pretty affordable. Then the Raspberry Pi came out in 2012, with a similar amount of GPIO, but demolished the price point to where it made sense to use it in place of a microcontroller.
That's really the magic of the Pi. You can keep one cheap gizmo around, and use it to solve (a large fraction of) two classes of problems. It doesn't fully replace everything a PIC or a PC can do, but it replaces an awful, awful lot of them.
Then the esp8266 came along, offering wifi and gpios and an arduino-like programming experience at a price point of literally $1
On the other hand, the RP2350 actually is a microcontroller, and IMO a nice one for many purposes. PIO, high-quality datasheet, nice ecosystem, etc. And the Pi Zero 2(W) can do most things the Pi/Pi 2 could, with a smaller footprint and less power consumption. Variety is nice.
And to the GP's point - the Pi Pico can be programmed in Micro Python.
Something can't be a microcontroller if it runs Linux.
Performance per Watt still outranks any other (quasi-)mainline linux device
Do you have a reference for this? Looking around, I see it being beaten by other ARM SBCs, and even low end Intel devices.
Many years ago, I measured performance per watt of the original Raspberry Pi when they were still relatively new. The performance per watt lagged behind even a beefy Intel box since the original Raspi was so slow that it destroyed any gain it got from using so little power.
EDIT: One set of benchmarks I found as an example: https://bret.dk/raspberry-pi-5-review/#Performance-Per-Watt
They've never been particularly low-power, in a performance-per-Watt sort of way compared to other offerings at whatever present day. In recent times, I've seen completely-believable reports that N150 boxes walk all over it.
They've never been particularly cheap, in a performance-per-dollar sort of way. Used machines from eBay, yard sales, and old broken laptops (that still compute!) have always been better. (They usually come out OK when new is compared to new, though, which is IMHO the only valid comparison.)
Those comparisons were never very favorable.
---
The parts where it shines are: Small; they come in two sizes, and those sizes are small and smaller. That was new, but it's been cloned all over the place.
Built-in GPIO that's meant for people to actually-use and tinker with. That was new, too, but it's also been cloned. (Also: These days, anyone can plug a cheap Pi Pico into any PC with USB and get a fairly intense amount of GPIO to goof with.)
Standardization, and the appliance-like behavior this can enable. Lots of folks, including kids of all ages, just download pre-built images and swap SD cards like they would cartridges in a Nintendo. That's not for me, but it's pretty neat.
Community. They've still got a lot of momentum by being first at these roles. That's good. It helps newbs (who at this point may have never had anything resembling a "real computer" to play with ever before in their lives) to get started.
Low-power enough: It's not ideal when chasing tiny Watts for battery or small solar power, but you probably won't notice it on an electric bill (and despite the bizarro-world cooling rigs people put together, a passive heatsink really is good enough to keep it running in-spec).
raspberry pi has terrible power management as well. turning off a core was impossible on previous generations for example. a few years ago rpi was the worst of all sbcs we measured for battery powered usages. this was for an actual embedded product so it wasn’t going to be used either way (ask broadcom for some chips HAHA)
but there is very little reason to use a rpi over other sbcs if you have a remote idea what you’re doing beyond hobby use
Performance per watt is nice but I’d be more inclined to talk about “problems being solved per dollar.”
If you don’t specifically have a project where you need the GPIO pins built in, I struggle to understand the use case proposition of a raspberry pi compared to a typical x86 mini PC or even just grabbing a think client desktop like a ThinkCentre.
Almost everything that is unique to a Pi compared to an x86 mini PC seems like it makes more sense with an ESP device.
When the Raspberry Pi was $35 and it ran a desktop OS and the cheapest alternative that did that was 5x the price that use case made sense.
They'll do whatever they do.
Maybe a tick-tock release cycle (one with new features and some speed, the next with the ~same features and more speed) is where they're headed, and maybe that makes sense. They wouldn't be the first.
I'd love to see even-lower-RAM versions, though. Most of what I use Raspberry Pis for at home for is not RAM-hungry at all.
My Pi4 network router has 2GB because that was the smallest/cheapest version at release when I got it, but the system itself consistently only uses about 64MB of RAM. It'd do perfectly well and have a ton of breathing room with just 128MB of RAM (which will never happen, but if it did happen...).
I suspect the Pi4 that I use as a set-top box with Kodi would be fine with 512MB.
I've used Zero Ws for all kinds of things over the years and never felt RAM-starved with their little 512MB of RAM.
So I'm learning towards 512MB.
But sure: 1GB options would also be fine even if it does double the price. Our comments serve to demonstrate that there's room in the marketplace for different SKUs with different memory capacities. :)
Ideally each RPi generation should keep the same price (or lower now that it's gotten so high) but with better performance. If they can't do that they just shouldn't create a new generation.
Sure (inflation adjusted), but I see no harm if they also provide higher specced SBCs.
Raspberry Pi isn't in direct competition with N150's.
Their niche is the industrial/embedded space. For that market, power consumption doesn't matter. What matters is that each model is guaranteed to be available till a specific date.
Among the people making things like a DIY NAS, who want fast USB, lots of cores and RAM, small-ish, not-too-bad power consumption, running Linux; and not caring much about GPIOs or passive cooling, it’s in competition with the N150
Among people who want GPIOs and network connectivity, a low price and an open, microcontroller like experience, not caring much about USB speeds and lots of cores and running Linux and suchlike, it’s in competition with the esp8266 & esp32. And the previous generations of RPi.
> Raspberry Pi isn't in direct competition with N150's.
They may not have intended to be in direct competition, but in the current crisis conditions they are priced about the same as equivalent RAM/storage N150s, have even worse supply issues than the N150s, and have worse performance/watt than the N150s.
Its mighty hard to recommend them for new projects at the moment (nor any of the Pi clones, which are also rocketing in cost and dwindling in availability)
IMHO people expected a Pi that offered similar performance as a Mac M1 but with Linux as first class citizen
Is there some serious astroturfing going on with the N100/N150, or am I just jaded?
I have a bunch of old intel atom boards laying around. The Intel Compute Stick (TM) burnt out its flash root drive in a few months. The C2000 board I had burnt out the clock pin to drive the bios. I have a Clover Trail with a PowerVR GPU (I thought I was getting an intel GPU because it was branded Intel Graphics or similar, but nope!) that lost Windows support very quickly after launch, and has no GPU drivers for any other OS.
Instead of being fooled 4 times in a row, I looked into using an N150 for a NAS, but this time I held off a bit until after launch so I could research it first.
Lo-and-behold, they all have crazy PCIe / memory subsystem data corruption issues. I guess there are some chicken bits for the OS developers to set if the kernel can stay up long enough after boot without a panic.
Why would anyone buy this for a NAS / embedded use case?
> I looked into using an N150 for a NAS [...] Lo-and-behold, they all have crazy PCIe / memory subsystem data corruption issues.
Source? I've never had a single problem with PCIE on N100/N150/N200.
I have had a ton of issues with drive corruption on the pi, both via USB3 and PCIE.
https://www.phoronix.com/news/Linux-6.4-Lands-PCID-INVLPG
https://forum.opnsense.org/index.php?topic=48343.0
All CPUs from all vendors have tons of bugs like this, which are mitigated in the operating system kernels, e.g. in the Linux kernel.
I am pretty certain that the Linux kernel must also contain specific code for various quirks of all Arm CPUs that have been used in the various Raspberry Pi models.
Intel had indeed several bugs that were more ugly than usual in their recent CPU models, like also MONITOR not working correctly in Lunar Lake, but even so, Intel still has better documentation for their CPU bugs than most vendors of Arm-based CPUs.
In any case, the bug that you linked was solved in the kernel years ago and it affects a privileged instruction that cannot be used in user programs. It does not have any direct relationship with memory and PCIe corruption.
so a single 3 year old errata for the n100 that was long since patched?
I've been running an N100 for 3 years with a 5 bay external enclosure over USB 3.2 Gen 2 and ZFS, and have not had any issues. It is pretty phenomenal, pulls about the same power, and costs around the same as an RPi 5 but provides substantially more compute and throughput.
I think exactly the opposite: we have no shortage of embedded crap we can buy; what is useful is dismembering intel. It would be better if the pi were risc v but this will do for now.
A Raspberry Pi with sleep and hibernation is like asking Valve to make Half Life 3. They just can't. It doesn't compute.
I have bought an rpi at every generation. And I still have yet to find an actual use for them.
Everything they do from a compute perspective is just better with a mini pc or old laptop with a mobile spec chip.
Everything they do from a programmability perspective is just better with a microcontroller specific to the task.
I just don't see the actual market position for these things. They were supposed to be a cheap board, but you can't actually buy them cheaply because the vendors upcharge so much.
Here are a few places that I've found pi's:
Rapid prototyping: I created a PoC to take a webcam snapshot every half hour and upload it to a server in an afternoon. Freelance project. Could it have been done with a microcontroller? Yes, but not in 4 hours.
Local digital displays in a gym: we built a system with a number of overhead 60" TVs, each with a pi on a VESA mount to show scheduling and workout information in a gym for a client.
HVAC controller: bid on a project where the customer's original concept was a Pi managing a rooftop HVAC system for large buildings. They outgrew the pi and wanted a new solution.
Data aggregator: collecting sensor information via BLE (bluetooth) and uploading to an internet server over Ethernet.
Remember that "cheap" to a consumer doesn't have the same definition for a business. To most profitable businesses, a $100 computer that fits on the back of a TV and consumes a fraction of its power is cheap. In fact, one of the reasons that Pi's were so hard to find for a while is that the Raspberry Pi foundation was prioritizing industrial/commercial customers over hobbyists.
I think I found a use case, but would love to be proven wrong with some faster / cheaper approach:
I've tired of buying S1 compatible sonos speakers on ebay, so I'd like to build a speaker enclosure with a WiFi device that has a high-quality DAC and the ability to use pipewire or similar to do real-time DSP and multi-room audio sync.
A RPI + third party hat should work well for this, or so I am told.
I think at this point the brand reputation and software quality are a big selling point.
If you're trying to build a couple of units of some embedded thing where you need to toggle some GPIOs or serial devices in response to requests over the network, but don't have the expertise or resources to do it with a microcontroller, a Pi is a great option - you know you'll have software support, and you know that the vendor will be making the exact thing you bought for 5-10y.
For hobbyist stuff at home, I agree, though. A mini PC is probably better for homelab stuff, and an RP2350 or ESP32 is probably better for anything embedded or battery powered that you want to do.
Development speed is also one of the forgotten axes. I mentioned upthread a system I built that was done in less than a day. There's no microcontroller solution I know of that would have let me deliver it that fast.
> I have bought an rpi at every generation. And I still have yet to find an actual use for them.
It's amazing how well these fit into the category of products that people feel compelled to buy, play around with, and then forget about.
Flipper Zero is another product that landed in the same space.
What's sad is that Raspberry Pi does have a lot of legitimate use cases and people who want to use them, but the supply has always been swamped by all of the demand.
My use cases:
* Replacement controller for my UFO Catcher - It has WiFi, easy to update, and I can operate the machine remotely with it. It's bolted to the back of small touchscreen that lets me change the machine settings as well.
* Remote printer access - I can monitor from the USB cameras and gather statistics about the prints.(I suspect a lot of 3D printing enthusiasts use them for this purpose.)
Having a small low power computer has been useful for me in those instances.
> I just don't see the actual market position for these things.
Isn't it mainly for learning and hobby-ism?
Obviously the business model is selling a $30-100 SBC to jerks on Hacker News who simultaneously brag about buying literally every single model, but apparently have no use case for them. Even better, these schmucks will never plug the product in, never submit a warrantee, and never harass support. Brilliant model!
I'm a jerk because I've bought them and tried to find a use case for them, but found they weren't as good as other solutions?
You have issues.
The 8GB Pi 5, at $170 [1], is encroaching on Jetson Orin Nano Super's $240 price point [2]. But the Jetson has a faster CPU (newer a78ae cores rather than a76) and, obviously, a whole-ass GPU.
[1] https://www.microcenter.com/product/673711/raspberry-pi-5
[2] https://www.microcenter.com/product/691058/nvidia-jetson-ori...
Nvidia's software platform for the whole Jetson series was, at least in my experience, absolutely awful on the Jetson Nano and Orin boards I worked on. Has that improved at all? I did not appreciate that the only option they provided was a full desktop version of ancient Ubuntu... and even flashing the OS image was a bizarre process.
Edit: looks like they at least have a better headless option now.
Nowadays upstream Linux with UEFI mostly works, with their out of tree drivers. I’ve managed to make it work in NixOS with the stock kernel. Look at the open embedded L4T project, they have some recipes for building that. No need to use nvidia’s kernel anymore!
Also, supposedly on the second half of 2026 they were going to be moving even more stuff out of their Jetson-specific drivers as they already do for their slightly newer chips (so you could use the standard drivers, and standard CUDA builds). Let’s see how that turns out.
Yeah, one of my bigger complaints especially on the Nano was the GPU only had really limited model support (iirc, mostly tflite but maybe I'm misremembering) and it sounds like the newer ones are more normal. That and what seems from the docs to be better headless support would be major improvements. Going further to mainline distro support would make them interesting to me again.
I was always disappointed by the Nano as it was a pretty capable device, but it seemed like not many people picked it up as a platform for cool things which I always attributed to the software.
The 16GB Pi 5 makes the Orin Nano look afforadable.
Pi's refusal to drop a USB-C on Pico due to cost increases is a terrible call IMO.
I seriously cannot fathom being someone doing development who wouldn't pay $0.50 extra to purge the last micro USB from their desktop.
A very small point, but pulling from a feather form factor BOM to compare.
$0.12 for microUSB female connector (rated 1A) $0.26 for a USB-C female (rated 3A). Needs 2 x resistors (< $0.01), 20% larger board area
I think the power capabilities are the biggest item. If you want to pull higher current from a laptop for development or supply from a wall, you have to switch to USB-C.
I don't think either of these prices are that aggressive - pretty sure the cost comes down at volume.
But then you loose market share to ESP32 where people just get USB C on their own.
I wonder if it would be worthwhile for them to produce both. Well, it will be hard to compare because the design cost doesn’t show up in the BOM, haha.
But it seems like it would be useful nowadays, since some laptop have mostly USB-C connectors, and USB-C to USB-C is pretty common. I’ve never seen a C to Micro. Do they even exist?
I have an unfair bias because I design PCBs as a significant part of my job, and switching out to USB on this board appears to be a non-issue.
I have a Pico in front of me, and there's plenty of room there for a USB-C footprint and the two 5.1k resistors. Given that, I cannot reasonably agree that the "design" stage is significant.
In other words, it's a change that I would make to my own board in 2-5 minutes because the stakes are low. My ballpark guess is that such a change at RPi would have to go through a proposal stage, a PCB change review, and then there would be dozens of places to update documentation.
Since backwards compatibility is non-optional, this would result in a separate SKU, which means that the whole distribution chain needs to be updated with a new product.
So, I acknowledge that when you're working at their scale any change like this is the definition of non-trivial. What I don't agree with is the conclusion that it's not still clearly the right thing to do.
> I’ve never seen a C to Micro. Do they even exist?
They do, in spades: https://www.amazon.com/3FT-Micro-Data-Charge-Cable/dp/B0DDWH...
I look forward to the day when they're no longer necessary.
I've always used USB-C to USB-A dongle + USB-A to micro B.
For what it's worth there are third-party rp2350 boards with USB-C connectors if that's important to you. Heck, WaveShare has one with two USB-C connectors: https://www.waveshare.com/rp2350-usb-c.htm
I am aware, thanks.
I do think that you're missing my point, which is that we're significantly past the point in this wretched timeline where they should offer a USB-C version of the reference board for this MCU.
There are broadly available third party RP2350 boards with subc and a variety of additional capabilities if that is important for you.
https://shop.pimoroni.com/en-us/collections/rp2350
https://www.sparkfun.com/sparkfun-pro-micro-rp2350.html
https://www.dfrobot.com/product-2913.html
https://www.seeedstudio.com/Seeed-XIAO-RP2350-p-5944.html
Not the point, though I appreciate folks listing links in an attempt to help.
What I am saying is that we're well into 2026 and there's no good reason for RPi not to offer a USB-C version of the reference board for this MCU.
Would love to see actual security focused hardware/software features, like full OP-TEE, fTPM (or a more ideally a real physical TPM), and similar. For example, so that the OTP isn't the only way to store a disk encryption unlock key.
The existing secure boot mechanisms aren't bad, but allowing for more than one public key hash in OTP would be nice, too.
These kinds of things are expected to be on modern embedded SOCs and SOMs now.
A physical TPM with their overall high-quality software support would be awesome.
I've spent far too much time messing around trying to get TPMs working over SPI or I2C to meet security requirements with 4Bs and 5s over the years.
You do know those are trivially bypassed with a signal processor, right? If physical access is outside your threat model, that's OK, but it makes (for example) the forced Win11 upgrade for DRM^H^H^H boot integrity enforcement seem ridiculous.
https://pulsesecurity.co.nz/articles/TPM-sniffing
The article you link to explains how to defeat the sniffing with TPM 2.0. But also, there’s no reason a physical TPM has to be a separate IC package.
I recently found out about the Radxa Dragon Q6A. A Qualcomm chip with faster CPUs, a good GPU, a DSP and AI accelerator, and a hardware video encoder seems very compelling. It even supports Windows if you want that for some reason.
I run the media lab at one of Europe's must prolific art universities. The variant I tend to use most is the 3B+.
Reasons: - full sized HDMI connector - headphone connector - good bang for the buck
If I had one wish for any new product in the Raspberry line it would be: Do the Raspberry Pi 3++ or something. Same thing. Faster, but with USB-C power connector, 4K Video resolution, 2× USB-C I/O, 2× USB-A peripherals and maybe M.2 support.
Same here. The 2b, 3b and Zero 2 have the best cost-benefit ratio. I recently migrated the Oberon System 3 to those three boards and particulary the Zero is a great host for this bare-bone system: https://github.com/rochus-keller/oberonsystem3native/. I hope the Raspi company postpones the end of life date of the 3b and Zero 2.
The raspberry pi CEO said that some of the boards will be delayed because of RAM prices.
https://www.notebookcheck.net/Raspberry-Pi-discusses-Zero-3-...
As someone who has just had to migrate a project off of radxa due to ongoing supply issues... good luck getting your hands on more than one or two samples of those :/
The only way I'll buy another raspberry pi is if they come with a power supply that's guaranteed to work with them. I got tired of the random reboots in the night and replaced my media center/NAS with an old Nuc.
Official power supply is sold separately. Pis work fine with unofficial USB-PD power supplies but will limit maximum current for USB devices. I have two Pi 5's and one 4 and they all are perfectly stable when powered with an old Lenovo laptop charger.
It's still a huge pain point. I keep a magic "correctly sized, but provides clean power" USB charger with my Pi in a drawer, because I know that, if I lose that charger, I'm buying a replacement. It's unclear to me why they're so borderline. Maybe USB-C would help?