52
votes
Raspberry Pi Foundation announces details of impending release of the Raspberry Pi 5
Link information
This data is scraped automatically and may be incorrect.
- Title
- The Raspberry Pi 5 is finally here
- Authors
- Emma Roth
- Published
- Sep 28 2023
- Word count
- 485 words
Pretty big performance upgrades across the board- wifi, SD slot, USB, CPU, etc all are getting a significant uplift. Also nice to have a direct PCIe interface now too. Jeff Geerling had a pretty good video covering a lot of bases on it: https://www.youtube.com/watch?v=nBtOEmUqASQ
Rpi continues to be one of those things that I have multiple of that I always "want" but never need or consistently use. I always end up replacing SBCs with more powerful miniPCs
What's your specific use case where miniPC does the trick but SBC doesn't? Gaming?
Comes down to form-factor and price/performance ratio, ease of use of nvme drives and 2.5" SSDs all in an existing chassis.. I don't use the GPIO (yet) anyway. So having nvme slots built in, SATA ports built in, and case/cooling already built in is nice.
They also go for very reasonable prices used, so I can get boxes that absolutely destroy RPi performance for a slight increase in cost (once you factor in power adapter, case, etc for the RPi). They still sip power, but then I can occasionally hit them for media server, file server, and other purposes (using Unraid atm) just to have this multi-purpose, yet still extremely small server that can do a lot for its size and still has form-factor advantages. Also I love using used hardware as a way to have new toys but still minimize e-waste footprint.
Some of what I do, RPi would work for, I just don't prefer it, or SBCs in general, as the form factor to accomplish those.
(Also some of it is I purely sometimes enjoy having stuff that is performance overkill for its purpose- as long as it's still efficient to run generally- I go wild with RAM amounts in my PCs, one of my miniPCs has a small Optane drive in it that apps run off of... etc- so sometimes the fun factor is a big part of it!)
I have two boxes that run Home Assistant, one at my house and another at a family vacation home that I do all the tech & computer shit for.
instead of an RPi or similar, I bought used Lenovo ThinkCentre M600 Tiny boxes from eBay. for $35 each I got a box with 8GB of RAM and a Pentium N3700 CPU - 4 2015-era x86 cores and a TDP of 6 W. this was during the covid-induced supply chain shortages, when it was virtually impossible to find an RPi, much less one at a comparable price point.
something they have over SBCs is native support for dual SSDs - one M.2 and one 2.5". this was a killer feature for me because I'm installing one in a remote location where it's unattended most of the time.
I bought two cheap SSDs for each one (128gb SSDs are so inexpensive now they're practically cereal-box prizes) and set up a ZFS mirror for the root filesystem. a few months ago, one of the drives failed. no problem - it sent me a notification email and kept right on working. the next time I drove up to our family cabin I brought a replacement SSD and swapped it in.
an additional benefit that's somewhat specific to my setup is that I have a dozen or so x86-based NixOS boxes, between personal and work desktops and laptops, my homelab, etc etc, with shared configs between all of them. it's possible to run NixOS on ARM, but it's more convenient for me to stay entirely within the x86 ecosystem and avoid having one-off ARM boxes that need special configs or setup.
I had to setup a print server for my wife's small business and one of the printers drivers only works for x86 machines... and fedora based distros only. I wasted a lot of time on that project...
How old is the printer?
That might be something you could still do with qemu. Print servers don't have to be fast.
Im not sure how old it is, but its just a thermal receipt printer so not fancy.
I'm not familiar with qemu, would your idea be that I run Fedora in an x86 emulated VM? How reliable would that be?
She'll be moving to a new location in a January and that would be the time to swap over hardware. Currently I'm using a thin client for this and I'm pretty sure a Pi Zero2 has more processing power at 1% the size.
That's exactly what I'm proposing. I don't think it's going to give you more performance than the thin client, but I think it should be good enough that you don't have to worry about it. I don't think that I've ever had issues with x86 qemu stability but then again I've never had to use it for long periods of time.
It doesn't happen to be a Star TSP100, does it? That's a really common receipt printer and it looks like CUPS should support it regardless of what system you're running it on.
Its the TSP100iii. IIRC cups didn't have drivers for it unfortunately.
I did some googling and it's possible but the manufacturer makes it a pain due to their licensing. The linux driver they supply is simply a CUPS PPD and graphics filters. It includes the source code for the filters so you can in theory just compile them and manually configure CUPS. It shouldn't be terribly difficult to do if you've ever had to compile a C program before. At least in theory.
I have only compiled stuff manually when following instructions. I did try to compile the driver but I kept getting some error (which I can't remember).
I was kind of done by the time i started trying to compile on the Pi. This was at the end of slowly figuring out the entire project which had a bunch of separate parts to it, just to be able to print a receipt after pushing a physical button so. I had it working on my spare computer, installed that at the shop, and replaced it with the thin-client later on so I could get my computer back. All so I didn't have to buy the more expensive version of the EXACT SAME printer that actually works with Shopify out of the box (I got this printer used for a good price).
It did have some directions in the comments on how to compile them, but those directions were specific to red hat derivitives so it wouldn't work if you followed those steps on Raspberry Pi OS. You did need to make sure you had the CUPS development package as a prerequisite. Fedora also does run on Raspberry Pi, so it might be easier to do it that way.
But like I said, it's a bit of a headache, so the time savings of using that thin client sounds like it was worth it.
I don't know why thermal printers have to be such pains in the rear. I have this bluetooth thermal printer I bought on a whim because it cost practically nothing and I ended up being really disappointed because it has no brand and only works with this annoying app and it doesn't look like there is any other way to communicate with it. I even broke it open hoping to find a bluetooth chip I could circumvent but it's all integrated.
This is so oddly specific, but I've been dealing with nearly the same issue for the last month in my free time at work. I'm working for a point-of-sale company that sells a lot of thermal printers. I've been trying to get all our software working on a rockchip ARM board lately, so I've had to compile CUPS drivers for ARM recently.
We don't really deal with Star TSP100's very often, we mostly sell Epson TM-88's and some rebranded Chinese printers, so I've only got those working so far, but on Tuesday I can take a stab at it. It won't be a raspi or raspian, but another ARM board on Debian, but I should be able to get something working and let @mild_takes know.
I know Epson's language for thermal printers is pretty much a de-facto standard. I haven't looked into it, but there's a decent chance there are some DIP switches on a Star to put it into ESC/POS mode for Epson compatibility. Likewise, your bluetooth printer might just show up as a serial port and might take the same command set. So far this cups driver has worked for every unbranded Chinese printer I've dealt with.
I thought I'd provide a bit of an update to this for you and @Akir
We bought a computer to use at the shop (reasons) running an Ubuntu based distro. I had some success just taking the already compliled filter files and dropping them in the (??)/cups/filters folder as well as downloading the cups development package. It did get it to print but the text was really small, after than I had to leave and do other stuff.
Its not on the RPi yet but i'll try that next.
Geerling also has a couple blog posts with additional details:
Answering some questions about the Raspberry Pi 5
Testing PCIe on the Raspberry Pi 5
This is pretty good news! The price is kind of a bummer, though That $25 price they were aiming for initially certainly wasn't sustainable forever, especially with how bad inflation has been, but $60 is just a tad more than I would like to pay.
In all honesty, it seems like this is much more perfomance than anything I would be using an SBC for. If you want to use it for emulation, media, or high(ish)-performance networking applications, it's great news, but I find even the model 4 is approaching overkill for the stuff that I tend to use them for.
That being said, I just hauled out an old Athlon-based PC that I was planning on turning into a NAS when I got the chance. Now I might just use a Pi instead for some pretty massive power savings.
Yeah, I'm trying to think of some good use for something like this, and I'm drawing a blank. At $60/80 plus power adapter, storage, shipping costs, and maybe case, this is roughly $100 to just to get up and running. Even if you have some of the accessories already, I think the lowest entry price I saw today was $70 for a Pi 5 4GB delivered.
It seems to me that the Pi Zero W at $15 fills the original niche of the Pi pretty well for something that is powerful enough to run a GUI, while compact enough to install somewhere (in the wall, on a robot, etc.). Then if you want something to use as a small computer, home server, or emulation station, there are much faster mini PCs you can get for $100-150 (including power adapter, case, and storage). I feel like the area between these where you need something smaller than a 1L case (but not that much smaller) with as much power as possible while requiring wall power or external battery pack is a really narrow slice.
I'm sure people will find good and suitable uses for it, especially for tinkering, but most of the tasks I've seen suggested are probably better suited for mini PCs.
Yeah, at this price point you start getting into the low end of "real computers," and if you aren't using the GPIO pins it probably makes sense to go in that direction. Especially when you consider how well an old laptop can work for the lightweight home server kind of use case (less e-waste, usually fast enough, usually fairly power efficient, easy to tuck away in some subtle spot, and the battery is practically a built-in UPS)
I think it's sort of the same on the lower end of things too — for little DIY projects I've had great luck with ESP32 boards. They're dirt cheap, have wifi and bluetooth, and are easy to program on. I also kind of prefer ending up with something that's more of an appliance rather than something I need to SSH into and run updates and backup and stuff.
Like you said, I'm sure there's cool things to build with this, and great to have as an option, but I think there's better choices out there now for a lot of projects.
I do a lot of work in the 3D printing space. There's some of the software that is happier on a 4 than a 3, and would probably like a 5 - ones that do motion planning, image analysis for detecting errors, that kind of thing. So I'm possibly more excited than most. Also people who do Pi compute clusters - it's admittedly more of a niche/hobby than practical thing, but a 3x performance boost is still 3x!
That said, yeah, I'm not sure if it will be broadly applicable as a mini PC alternative - it's still probably best suited as something of a compact industrial PC you can tuck in to some corner of your physical device like a robot or such.
I wonder (and have no clue here) if they priced it a bit higher per unit so that they can do bulk unit discounts. If you are only buying one RPi, the $60 price tag isn't outrageous for what you get but being able to discount by $10 and not lose your socks in the deal gives the Foundation the ability to woo larger corporate customers while hopefully addressing some of the stock problems of the previous generation.
Sounds like a solid upgrade. I don’t follow the ins and outs of Pi Foundation drama, if that’s even an appropriate word to use, I vaguely recall some ruffled feathers about the previous generation. I think it was mainly about low supply, making them hard to find anywhere, and they were getting snatched up in bulk by companies before individuals were able to get their hands on them. Is that right?
Hopefully the product is good and the supply issues are all worked out. I have done some hobbying with a Pi 2 and a Pi 4 and love them to death. I’m very interesting in picking up a 5 based on this spec breakdown.
From the announcement on https://www.raspberrypi.com/news/introducing-raspberry-pi-5/
So hopefully that means there won't be the same issues that plagued the 4, at least for a few weeks maybe lol
I love that they did this. But the cynic in me wonders if they picked the end of the year because they know that a lot of their big customers are going to be in the prototype phase during most of that time, and so their orders will be relatively small. Like a situation where they are having their cake and eating it too, since the large corporate orders that made such a mess of the supply last generation will likely be low anyway.
But either way, I like the move.
That's the narrative.
I think reality is a little different from the RPi Foundation's point of view. The way they see it they prioritized businesses but limited their supply as much as they could. They claimed that some of these smaller businesses relied on RPi's and they didn't want to screw them over.
IMO the fact that these are being used in industrial and by hobbyists is what makes it great. Its clear to me a balance needed to be found and some hobbyist were just loud about not being happy.
Also IMO some of the community needed this kick in the pants to start looking at microcontrollers instead of using a computer as a microcontroller.
The Raspberry Pi Foundation is already doing this in the form of the Raspberry Pi Pico. The model 4 was already moderately expensive and over the years microcontroller boards got dramatically cheaper and easier to program because of inexpensive fast MCUs that can run interpreted languages like Micropython or NodeMCU. The Pico helps them fill into that niche.
IMO microcontrollers were already kind of there in terms if ease of use plus cost with Arduino IDE + cheap clones.
Of course I agree these new cheap boards that run micropython lower that bar to entry even lower.
Raspberry Pi 5 is announced an available for pre-order. Notable changes include a significant increase in processor speed (expected to give a 2-3x increase in computing performance) and in-house silicon for the peripheral controller, increasing both the openness of the platform as well as dramatically increasing peripheral bandwidth, with 4x the MIPI video lanes across two connectors, multiple full speed USB 3.0 ports, and for the first time, a single PCI Express 2.0 lane.
Happened to get the reservation email minutes before I finished work, looking forward to getting my hands on it!
I've been thinking of building a LoRaWAN gateway, with the appropriate hat. Nothing extreme, just a few garden sensors, a weather station, and a tag to track my errant indoor/outdoor cat. If I get ambitious, I might open it up for others to use.
I don't need the power of the Raspberry Pi 5 to run that application (unless I want to mine Helium 🙄), but maybe there will be more availability of the 4, used?