Who would buy a Raspberry Pi for $120?
jeffgeerling.com
jeffgeerling.com
What's so great is that it's all up to you. There are "smaller" or older Pis still being made. More for more is fairly reasonable.
If you have a sort of inbuilt bias against "buying second best" that you might consider this a price increase. I have that feeling often. I don't want the second best camera .... and that ends up leading me into stratospheric prices as I finds better and better things. Fortunately I know this and realise that when I am doing it it's because I don't have a clear use in mind for the item - hence I can't decide what's "enough" so the answer is not to buy it at all.
I migrated most of my self-hosted services from multiple rpi (I've been using several of them for years) to a single cheap Intel N100 NUC that I purchased last year: 16GB RAM/512Gb SSD for 156€, and I've been very pleased with it.
Two things that actually annoy me is that SD cards are way way slower than other storage options and are very prone to corruption, I've lost not critical data, but data that was still useful to me due to SD cards (even I believe if you purchase them from a known manufacturer). So this implies that you will likely want to have another more reliable storage option to keep critical data backed up (more spending). And the other one is the missing power button. I don't know if that has changed but all the models that I own lack of a power switch button.
Lets be honest many people will have old/spare/extra phone/laptop chargers that they can use to power a Pi. SD cards are cheap, a few dollars.
Raspberry is great when one learns or needs a typical, embedded platform. I'm using Raspberry to learn about Kernel, drivers, bootloaders, etc., but I wouldn't use it as my local server/home automation computer. For this, I have a second-hand HP T630 that I bought for $46, and works great.
Also the place where I acquired it no longer exists, hence why I got stuck with it.
Yes I should have known better, but it is as it is, and we learn from our mistakes.
Cheap tiny PCs are better at the high end, and increasingly-capable microcontrollers are better for tinkering. The Pi sits in a weird niche, and its popularity these days is largely based on inertia and network effects.
It's been a slow, strange shift away from a $25 computer meant for kids to learn programming and electronics.
> for some reason they never came back down
That simply is not true if you buy from an authorised distributor. Authorised distributors are required to sell at MSRP.
Secondarily, mini-PCs are much more readily available at better prices now, which makes the Pi look more expensive in comparison.
- multiple cores
- all sorts of IP on the silicon
- usb 3
- gigabit Ethernet
- wifi
- Bluetooth
- camera connector
- dsi
- hell, even a power button
… etc etc. They’re not comparable.
Besides, GP was pointing out that it feels like the price has gone up. I was pointing out that one reason it feels that way is because it has. No-one is arguing you don't get something for that money.
With a lot of other SBC (single board computers) you're lucky to get a very of the Linux Kernal which works in the first place. The likelyhood it will get upstreamed or kept up-to-date with upstream is basically zero in most cases. Raspberry Pi has great software support and the original Pi 1 has only just become end-of-life and unsupported by Raspberry Pi OS at the end of last year.
There is also clarity on when the hardware will be produced until. For the Pi 5 for example they've commited to it still being produced until at least 2036.
Those together mean that if you're using it in industrial, educational, or embeded settings then you can depend on the board lasting a good while longer than other options which might become insecure and useless before the hardware itself fails.
This is slightly less of an issue for hobbyist who are running Homelabs or personal projects on SBCs.
I see a lot of people say this, but I never saw this happen in the UK.
All of the official sellers sell it at MSRP.
> I see a lot of people say this, but I never saw this happen in the UK.
Also UK. There was definitely a time (late 2021, early 2022) lasting some months when most RPi products were only easily available from resellers (I'm using that word charitably) at vastly marked up prices (at least as a home user just wanting a single unit or small number of units, it may have been different as a bigger buyer).
I have a Pi400 where a Pi4 (or 3 or older) would do just as well, because at the time I needed it for some reason the Pi400 hadn't yet been affected by the shortages (and resulting scalping) and was far cheaper to buy new (even with the full starter kit which included PSU & mouse) than any Pi4 despite nominally being more expensive. I think the 400 was even on offer below RRP at the time. The Pi400 became hard to find at RRP soon after I bought mine, I have a friend who was looking a short time later.
The shortage may have lasted longer than I'm remembering, as I only paid particular attention for a bit longer than the time of my own & my friend's needs.
Model As, picos, Zero Ws etc are all cheaper than that original PI price after adjusting for inflation and are typically more flexible in form factor and better CPU too.
2GB Pi5 - $50 16GB Pi5 - $120
I can see the uses of having more RAM, but surely it's cheap enough to make 4GB standard, with options for 8/16GB that are sensibly priced - removing one SKU.
I can't believe that people are still surprised about this kind of pricing, because it's completely normal pricing strategy in almost every market. If this is greed, then I guess greed is simply wanting to be able to run a viable business, and wanting as many customers as possible to be able to afford your products. It's basic price discrimination, and personally I think it's a good thing.
I suspect when you're selling at the quantities that Apple are doing, they'll have to consider the impact their RAM pricing will have on RAM shortages. If they price the highest RAM option much cheaper, everyone will just buy that option, even if they don't need it. At those volumes, that could trigger the need to build new RAM factories, which will drive up RAM prices a lot in the short term, which is rather self-defeating.
They have to make money somehow, it means selling things for more than they pay for. They chose to make higher margins on the 16GB model than on the 8GB or less models, which is not particularly shocking.
https://www.londonstockexchange.com/stock/RPI/raspberry-pi-h...
A charity can't really trade in the same way a company can. For example a charity can't register for VAT (sales tax).
The way it normally works is that a charity exists which takes donations and does charitable stuff, then a seperate company exists which makes money and sells stuff (does trade) and it happens to be run by the same folks, and donate money to the charity (or have some kind of 75% of our profits get donated to charity) promise. That's super trivial to unwind as the charity often has no formal ownership over the trading company, it is really just a beneficiary.
On the grapevine, they had more money than they knew how to get rid of in accordance with their very narrow charitable objectives.
[1] https://www.theguardian.com/uk-news/2024/nov/21/captain-tom-...
Captain tom was just a badly run charity, they didn't trade.
But I think a charity should spend most of its income on charitable activities. According to its accounts the Foundation spends £12m on that out of a £213m income from trading, which seems very small.
The commercial subsidiary is a publicly-traded company and the Foundation (the charity) owns about 49% of it (last year it wasn't publicly traded and they owed ~98% of it iirc). The Foundation received £213m from "other trading activities" last year (which I interpret as income from their share of the commercial subsidiary) out of a total income of £220m. Out of this, the Foundation spent £12m on "charitable activities" and £195m on "raising funds"... [1]
[1] https://register-of-charities.charitycommission.gov.uk/en/ch...
However given the cost of RAM is quite cheap, sales are slower after the initial Pi5 rush and there are a number of competitors would it have been much better for the company to say
4GB is entry level for $50, 8GB for $65, 16GB for $95
Would they have sold more units? Improved the range of software that can run on a RPI, increasing demand further?
Anyone who knows what they're doing is capable of using a cheaper Allwinner or Rockchip SBC or flashing a supported Android TV to Linux because SBCs are for some reason pricier than the same SoC that comes with a plastic box and power brick.. or even (gasp) to run their blinky LED off a $3 Aliexpress ESP32 board.
(I would even line a 32gb skew but realise that is unreasonable for the current soc)
It would cost about $3 to make the 2GB RPI5 a 4GB PI5, and you might save something close to that by removing a SKU. You then get a bunch of new applications (which need more RAM) for your SBC and your product sells more as it looks better value. On the whole that $3 is now looking like a bargin.
I don't think 32GB is needed or sensible, so will disagree there.
32GB is impossible at present. No-one makes a 256 gigabit (32x8) chip to fit in the same footprint, so then you have to do a new board design to support multiple chips, which might not be doable in the board's current footprint.
Micron do a 192 gigabit (=24GB) chip but it's LPDDR5, rather than LPDDR4x, which I imagine the RPi 5's SoC can't drive.
Even if we look at farnell for a qty of 50..
https://uk.farnell.com/micron/mt53e512m32d1zw-046-wt-b/dram-...
50 - £9.29 EX VAT
https://uk.farnell.com/micron/mt53e1g32d2fw-046-wt-a/dram-1g...
50 - £12.04 EX VAT
Those are public prices (not even account price) for buying only 50, and there's ~$3 difference even there.
Looking at Micron's range, they now have a 128Gb (16GB) chip with matching specs, which is the MT53E4G32D8CY-046 WT:C. That chip costs £74.66 per chip from Mouser Electronics (just one site I found selling it) in quantities of 1,360.
In contrast the 2GB chip - which I surmise is a MT53E512M32D1ZW-046 - costs £9.26 each in quantities of 250. There's a price break listed for 2,720 of the things, but I'm not going to ask for a quote.
So that's £65.40 more for the larger RAM chip. Which is about $80 on today's mid-market exchange rates.
Looking at the 4GB it's probably a MT53E1G32D2FW-046, which is £16.72 in bulk quantities of 1,360 from the same wholesaler (for the revision C part). So that's an extra £7.46 or about $9.
Why are other systems' RAM upgrades cheaper? Largely because they use more chips, and therefore each chip has less capacity. That means either the chip is less dense, or it's a smaller die, so fewer defects are likely within the one chip, making yields higher. An SO-DIMM might have two or four chips on it, depending on whether it's populated on one side or both; a full-size DIMM might be 8 or 16 (or 9 or 18 if it has ECC).
Obviously I've made the assumption that this is just a drop-in replacement chip on the existing board. But I assume it would be very hard to redesign the board to fit in the same form factor, with all connectors in the same positions, but add on another RAM chip. If the SoC even supports driving more than one RAM chip, which it may not.
TL;DR that's what Micron charge for the larger chip.
If we use your prices above just as a "scale" then you can already see here that 1X 4GB part is cheaper per GB than 2GB, so as I say it's around $3 in cost to upgrade the base model from 2>4GB.
(EDIT: yes i get that for the mac mini the ram is getting doubled, but here the memory increase is also similar in amount).
i recognize the new memory chip SKU that made it possible, but you need a lot of mental gymnastics before justifying the compromise.
Good article about the same topic https://louwrentius.com/the-raspberry-pi-5-is-no-match-for-a... Even a 10 years old i5-6500 is outperforming the RPI5!
I've been toying the idea of replacing my aging macbook with one + an M4 mac mini - the situations where I need to use the computer on battery are relatively rare, and it'd be substantially cheaper.
I even sometimes use a bigger suitcase with wheels, I can fit 2 screens in that suitcase easily and use a Dell MDS19 stand - the pro for this stand is, you can unscrew the bottom (no screwdriver needed!) and it is just a very thin flat piece. The other part is also basically just a long heavy bar with the monitor mounts, but in total it is easier than to transport the two individual stands.
Previously I have travelled with a 'normal' monitor in a suitcase and had it destroyed by baggage handlers.
I travel with a mouse but no keyboard, had to buy one real cheap.
Another reason why one might want to buy one is as a build server for arm64 packages.
Before that, we were testing Rancher with these ones for ARM testing, too. No miniPC right now offers that, but I think they will be coming soon.
Outside of a real use-case, RPi products are well-polished and fun to play with. There are few other products with an overall presentation - from design to marketing - that are as clean and well done. Personally, I enjoy supporting that.
If you actually use the memory as the input instead of 1, 2, 3, 4 then a line would fit it perfectly (40 + 5x).
BTW, market price for LPDDR4 is around $0.66 per GB:
Impressive that he fitted against 1,2,3,4 and not the memory as input. That's a total blunder.
I should probably stop reading the article at that point, being aware of the the Gell-Mann Amnesia effect.
It is doing everything I want: running https, http, websockets, and monitoring a USB device using python. No GUI. I tried on an esp device but it proved to be too hard to keep those plates spinning. It's the right device for the right job. That's the rule.
* I am willing to corrected about the price!
I've ranted about this before on HN so won't too much now, but suffice to say over the years I've tried to use a Pi for multiple different projects, and it's very rare that anything ever just works, even following apparently clear tutorials. I always run into issues somewhere installing packages from the linux and/or python ecosystems.
If all you need is a base Raspbian install, fair enough; but anything more complicated and you're likely heading for a world of hurt, trying random 'solutions' from years-old Ubuntu forum and StackOverflow posts.
but n100 sbc offer much better performance, and power consumption as well
> I managed to get my N100 mini pc to 1.3-1.5W idle power consumption at wall with active wifi connection (screen turned off), so for headless server idling most of the time the N100 is better option.
https://www.cnx-software.com/2024/04/29/raspberry-pi-5-intel...
I've been a fan of the Pi since the beginning, and in the SBC world, their support is unmatched, but with these improvements, they're speccing themselves more into homelab/server/desktop use than straight tinkering/GPIO stuff, all while the prices are pricing them out of that market.
If you don't want to annihilate the performance of a Pi 5 like the Minisforum mini-PC does, you can buy a complete N100 mini-PC for ~£100 with all of the RAM, storage, case, IO and PSU included that will still destroy a Pi 5 and actually cost you less up-and-running and still get better performance.
[0]https://store.minisforum.com/pages/product (Click UM480, their website design doesn't let me link to it directly)
(I often bring out that old ThinkPad X220 to test software on older hardware, and it's so satisfying how fast that thing still is until you open the browser.)
If I don't need GPIO, PI does not seem to make much sense anymore.
Many of these comments seem to ignore all the advantages of the the rpi ecosystem like footprint, direct IO integration and arm and simply focus on CPU through put O_o
What's considered 'convenient desktop experience' may vary and could be very subjective.
OTOH https://www.qemu.org/docs/master/system/target-arm.html can be very convenient from whichever contemporary X/AMD64-setup.
* Rpi5 16GB - £114.90 (Pi Hut)
* Mini PC N150, 16GB (upgradable), 512GB M.2 SSD (upgradable), Win 11 Pro - £169.95 (Amazon)
But I think paying 50% more for the option of upgrading a low-power machine is daft. Or paying 50% more for an OS I don't want.
Great job making yourself more work...
Maybe now you can see why "and you're good." isn't actually true?
If you want to connect breakouts and do some electronics work etc, a Pi 4 or a lower-RAM/lower-cost Pi 5 is an excellent choice, for those who were using them to host software like I've been doing mini-PCs destroy Pis these days at the price/feature and price/performance level.
If the Pis weren't looking more like these mini PCs in features and cost, they wouldn't be coming up as comparisons.
But I appreciate some of you left your Pis in the sock drawer and are unaware of the seemingly obvious (to you) things that learners trip up on. I spent years maintaining a Pi-based audio player aimed at beginners. I even tried to support some other platforms, should have been "simple" but the little differences matter to those that don't know what's little and what's big.
I should probably give some examples but I guess this is a case of what's obvious to me, isn't to you.
I had a 3x pi cluster and found the 8GB limiting. So made those the controlling nodes and bought orange pis 32gb for the worker nodes. And those were well north of 120.
But that kind of usage case is quite niche. For most people minipcs are probably a better fit in that price class.
I'm unconvinced by the LLM argument...even if the model fits...it's already slow for 8GB class models.
That means about $2 a year where I live. My second best computer - a hp mini computer - draws 11w at idle. Meaning $22/year. At a service time of 5 years a pi does make sense.
Jeffs reasons why you’d need 16G in a Pi feel convoluted.
Even the better low-end N100's usually idle at 5-6W, which is double the Pi 5.
SoC idle temperature depends on environment and cooling solution; the N100 would quickly thermal throttle if you don't have a comparatively large heatsink attached. The Pi 5 will actually give you full performance for a minute or two before throttling (assuming no heatsink).
Other Arm chips are much better, efficiency-wise, but the Pi 5 is still more efficient than any low-end x86 build, especially used.
I’m sure you have many more topics and a ton of other gear to explore to explore and many pi-related videos are fun entertainment (which is fine for it’s own sake) but it always practical.
That 4-SSD nas looks fragile as hell and I think it’s definitely worth exploring alternatives.
But it is not low-end for sure. I'm kind of wasting the computer on this use case. It's just what I had after Pi 1B turned out not being enough.
With 16gb you can have a massive file buffer cache and avoid disk io delay.
Anyway, now that they're available, I'm getting one. I sure have a lot of Pis nowadays.
I bought 4B and found it slow, hot. Thanks to COVID I could resell it for a tidy profit lol. Of course there are use cases I can imagine.
I get that Apple has its own BS reasons for that, but Linux?
Containers make this a ton easier of course but then you’re building in a container and you probably wanna make that container match your target environment, soooo…
Another option is to use Qemu and binfmt_misc - that way you do get the result of "native" compilation, but it's slow.
Because that's the path of least resistance in terms of development, at least unless you need to industrially build for many architectures. If the compiler runs on $ARCH, the output runs on $ARCH, I guess there's some exception but it's that, an exception.
> Unless the build includes tests
Build should indeed include tests, unless you want to ship software that potentially fails in architectures you claim to support.
Cross compiling isn't a first class citizen in many languages, and when it is, cross testing usually isn't, but you can have a toolchain that includes qemu (as a vm or as a dynamic binary translator) to cover those cases.
New 16GB Raspberry Pi 5 on sale now at $120
What makes me sad is the relative failure of the 400/500 concept. I miss the portable all-in-ones from the 80s. But the difficulty in putting a SSD inside and the infuriating mini HDMI were the nail in the coffin.