Can these boards replace the Raspberry Pi CM4?
ฝัง
- เผยแพร่เมื่อ 18 พ.ค. 2024
- There are two new Rockchip RK3566-based Compute Module 4 competitors, the Radxa CM3, and the Pine64 SOQuartz. Should you go buy one of these instead of a Raspberry Pi?
Maybe, but not yet. I'll explain why.
Mentioned in this video:
- Radxa CM3: wiki.radxa.com/Rock3/CM3
- Pine64 SOQuartz: wiki.pine64.org/wiki/SOQuartz
- GitHub issue for CM3: github.com/geerlingguy/raspbe...
- GitHub issue for SOQuartz: github.com/geerlingguy/raspbe...
- My blog post on the boards: www.jeffgeerling.com/blog/202...
Support me on Patreon: / geerlingguy
Sponsor me on GitHub: github.com/sponsors/geerlingguy
Merch: redshirtjeff.com
#RaspberryPi #ARM #Rockchip
Contents:
00:00 - Move over, Pi
01:02 - RK3566 and special features
02:10 - Expectations vs Reality
04:09 - The problem...
06:44 - There's still hope! - วิทยาศาสตร์และเทคโนโลยี
![[LIVE] : ONE ลุมพินี 63 วันนี้!! คู่เอก "ยอดภูผา vs โซเนอร์"](http://i.ytimg.com/vi/RGVmBsegKEI/mqdefault.jpg)
![[TH] 2024 PMSL SEA W2D4 | Summer | โอกาสสุดท้าย ของสัปดาห์สอง](http://i.ytimg.com/vi/aswciGrYovo/mqdefault.jpg)
I like that Jeff Geerling uses the raspberry pi, and I see that he does it just to have fun, and in fact he says it a lot, for me it is the best, I also have fun, thank you.
Fun == fun!
+1 Jeff... The last rockchip I had in my lab was a Chinese Android tv box with the oddest reset button placement I've ever seen, required a USB A to A cable (yes, I went to the broken phone charger coffin (plastic box)), and it took me about 18 hrs to find the image. Kernel parameters are challenging and fun. Oh and I compiled a new uboot for a Chinese router board on a pine phone via a docker container while on the toilet last weekend! Now I want one of those just for the lolz.
Something that depressingly few developers understand is that _excellent documentation and excellent UX are killer features._
Even more so than just having good hardware. I don't want to have to reverse engineer something I paid money for to make it perform basic tasks, instead of working on my project.
That's why so many SBC's have failed to make a dent in Pi's market share.
Excellent UX is why Slack got so popular; yes, it's just IRC with better UX design, but that's literally it's primary selling feature.
they don't care. they sell a few solely based on marketing hardware specs and then move on to the next. They are not trying to build a brand name. Dump the name, create a new name, make as much as possible, repeat.
@@MrTweetyhack as a society, we oughta consider such behavior as tantamount to murder. (Because it really does kill people.)
nekomakhea: absolutely!
Oh they understand that over at Ubiquiti all right. Even down to the unboxing experience; however, core functionality unit testing is quite the challenge for them. Same with truth in advertising
Amen.
Slack is horrible. Somehow it manages to make a last-gen AM5/DDR5 system with low load lag when it's open.
Yes. My experience with hi-spec SBCs beyond RPi has been a waste of time, frustration, and money. The impressive hardware only equals 1/2 of a consumer product. Tell Jack I said, "Goodbye."
Great PSA to the maker community, Jeff. 👍
In many cases, I wouldn't say it's necessarily a waste-there are still plenty of things to learn from a more jagged experience. But if your goal wasn't pure learning, and you just wanted to get something done, and couldn't get it done... it would be more of a frustrating experience. Lucky for me, I was in this for the learning, and my initial goal was to plug in a PCIe device. I didn't quite get there but I did learn a bit!
@@JeffGeerling "Everybody's somewhere (no matter where I go, there I am.)." My learning experiences are more on the level of just getting motors to run, IoT to communicate, and cameras to work.
That makes me one of your favorite fans, following behind your trail blazing footsteps, but tripping over many pebbles along the way. BR 😎.
Couldn't agree more... You pretty much have to be a low level linux expert to even consider trying to use any of these non-pi ARM boards. Either that or you have to wait for a good amount of time after they've been released for their ecosystems to be fleshed out enough to be really usable.
@@mikekopack6441 Well said. Right before the rPi4 came out, I adopted the Pine Rock64 because it was already out with 4gb of ram and 4k video. I give credit to all those who put in time to build distros for the board, however the builds vary and so does how well the drivers are optimized.
@@mikekopack6441 Or...take the savings of these third party units, freeze them with a "current OS", and dedicate them to their job. My PiZero2 is locked in time. It also won't get tolerated in the RPi forums. Heh.
That's something a lot of people fail to grasp : with Raspberry Pi, you're paying for official support and the community. Yeah, you can find other boards with similar or better specs for cheaper or vastly more powerful for barely more (hell, even x86 if you want), but with Raspberry, you're paying for the official OS, Raspbian (yeah, I'm an old Pi's user) and the hundreds of community images people created in the last decade for every usage (and semi-official images like Ubuntu).
With a lot of other boards, you'll maybe find an official Debian image updated in the last 4 years and that's it, the exception being Pine64 since they have a pretty sizable community and even then, brand new products usually have nothing since they themselves don't spend much time on the software and rely on the community. Having better hardware doesn't mean anything if you can't actually use it, that's why the Raspberry Pi is a really popular emulation box while it's not particularly powerful, people have been developping OS dedicated to it (like RetroPie and Recalbox) for almost a decade for that purpose, you won't find anything similar on other, more powerful, boards.
I will say, although yeah these other SBCs do have generally software and support limitations, I wouldn't say they're all that bad. To use the examples of Retro Pie and Recalbox those offer prebuilt images for a few of the alternate SBCs, while Retro Pie currently doesn't offer prebuilt images for the Pi Zero 2.
You are absolutely correct sir. As an Embedded Engineer I can attest to the documentation woes of cheap Chinese hardware. The bottom line is that they do not care about providing documentation and the support required to establish a platform. Let the buyer beware.
Let's not be xenophobic here; Western engineers are equally bad at documenting their products.
@@TerraMagnus It's not about xenophobia (especially since the context is wrong). It is about commerce and language difference.
If they are more easily available, cheaper and their is a nice clean Debian distro I might be interested.
This is my hope-so far I haven't been able to boot Debian on either :(
Radxa CM3 has fully same price with RPI CM4.
I feel you 100%! I have an ancient Pine A64 "raspberry-like" board which I bought used, and there's pretty much no docs at all about this board, I ended up getting one of those still-kinda-maintained distros just to make it boot and do something "useful"
I agree with the lack of software support and documentation of these non-raspberry pi boards.
I have a FriendlyElec NanoPi Neo 2 and as much as I like it for its size, I have to trust Armbian for continued support and using the GPIOs is a nightmare in comparison to the Raspberry Pis.
Thanks for the great video Jeff!
That was deep. Took awhile to get there, but it really is impressive what RPI has for an ecosystem.
It's so odd to me that there's been no effort (at least, that I've seen) to create an "IBM PC" for ARM based computers. Basically any x86 OS will work on basically any x86 computer because every x86 computer has functionally the same core architecture surrounding the processor. ARM computers are all like the Xbox and PlayStation; they might have the same processor but nothing else about the architecture makes it a PC. Every ARM computer is bespoke and closed off from the rest of the ARM ecosystem. If ARM (and RISC-V) wants to supplant x86, it needs to be at least as good as it.
M1 macs are the only thing running an arm chip and getting better performance per watt on some X86 applications running natively/on bare metal. Apple has had a great many years of developing so its hard to say if open source community will ever get there.
The status quo works fine for most of these silicon vendors: nobody building a phone with one of these SOCs cares about the boot process once they bring up uboot or, let's be honest, Android fastboot, on it.
@@rpavlik1 exactly. The fact that I can't install straight up Linux (or even Windows ARM) on a smartphone sucks.
There've been attempts, but it's never going to take off as long as companies vie for total control of the ecosystem in question.
We look at the PC compatible of the late 80s as a revolutionary explosion that lit our lives on fire, but companies look at it and see IBM suffering massive "lost revenue" in letting other companies release clones and add-ons without taking home the whole pie for themselves. Take it from the nerd who fully immersed themselves in corporate logistics for a year, they have to give up far too much control of the platform they sunk corporate finances into to make it happen.
Are ARM servers (for business, not consumer) capable of just booting any ARM ISO like normal from UEFI without a custom image?
Great exposure for these products, Jeff. Perhaps the issues you've pointed out will make them more widely known, and therefore more of a target for the people working through these issues.
Besides the strengths of the RPi platform that you've already mentioned, just having defined/popularized a set of standard form-factors and interfaces for the hardware (in this case, the interface pinouts being used by all the boards in this video) is a big win for everyone. When I was growing up in the '90s and '00s, The ability to move pieces and parts around to different systems in various ATX systems was huge for my ability to tinker and learn. I hope the next generations will have similar experiences, even though the form factors are changing considerably.
That's part of my goal in posting this video-I could keep toiling on and making a little progress here and there, but maybe a few people who might've not known of the interest in these boards can push them further better than I can so they do become worthy alternatives sooner rather than later!
I am excited about the competition. But you have said it all better than I could.
its among my list of things to do to make A CM4 carrier board and a CM4 alternative but making a SBC seems like a monumental task that is not easily doable by myself.
that said Raspberry Pi documentation is a golden example of good documentation
Fair point Jeff. I hope the guides and images for these boards will mature like the Pi ones did. It's always cool to have an alternative to raspberry pi, but it should be relatively easy to set up and not frustrating like your experience.
This video is the PERFECT explanation of why Raspberry Pi is my go to board, in all of it's form factors. Yes, there are boards with better specs on paper, but good luck getting something to work. If the CM4 had native USB3.0 support on it though another computer, as with the Radxa i'd be in dream land. But it doesn't, so i make do.
Would be nice if you could come back to this, seeing how these boards have developed!
Thank you for what you do Jeff.
Easily the best source for sbc info in plain English plus an awesome sense of humor.
I totally understand that. I'm a software engineer and had a manager switch the language we use. He picked one with no support, documentation, or community and trying to get something working was a nightmare. Turns out the only reason he switched was so he could be the expert and gain more job security. Meanwhile, he lost several brilliant programmers.
"he lost several brilliant programmers" even more job security
I'm assuming the manager has other flaws, since they can't be that brilliant of a programmer if they can't pick up a new language. 😉
I don't care if an SBC is more "powerful", I need the software experience to be on point. Raspberry Pi OS isn't without issues or fault, but is still extremely polished. Hard to take the risc
Whats the problem with oneplus ?
@@prashanthb6521 They've lost the mid range market and their software support isn't as good as similar phones.
Its either the really good Nord N10 at $300 or the OnePlus 9 at $800. At that price, the S21 makes more sense especially the 1 year longer software support for new OS updates.
Google just took the mid ranged crown again at $599 for the Pixel 6. Where the OnePlus 6T was 3 years ago, Google took this crown, especially the support until 2026 with Android updates and even longer security update.
Thank you. This is exactly how I've felt for years.
One of the great things about searching for raspberry Pi information, is that the odds that I'm going to land on Jeff Geerling's page are in the high 90s for any search! A great thing about the raspberry Pi, in other words, in addition to everything else, is you!
I rather expect that you will get these working eventually. But as an old technical writer, I find your explanation most interesting; therefor no support from them equals no support from me.
Thank you so much for giving it to us strait. You must have been tempted to only showcase it after getting it all running without going into the difficulties you had just getting it to do that.
Support! The number 1 requirement for business use cases. This is why my company still likes to buy Dell equipment: Support. It's a pity though. Cuz there are a lot of interesting hardware eco systems out there they don't get the love they deserve.
There's a lot of interesting hardware out there that never gets the ecosystem it deserves. Without an ecosystem, it starves or, worse, festers in designs and systems for years via lock-in.
I fully agree with your observation. I had a time where I also bought several other boards based on the hardware specs. They almost all did end up in a drawer because of software and documentation issues. Overall I came to the conclusion that it is far more likely to get your project working with a raspberry pi then with any of the other boards.
I totally agree with you, I have tried Radxa SBC and went through hell trying to get basic OS booted, managed to get a stable OS build with buildroot but couldn't get the display and many other peripherals running.
Got similar experiences with non-pi boards. The lack of community support makes them harder to use and after a few years the images disappear from the web and you end up with a paper weight. I do not have this issue with my old raspberry Pis. Third party board needs to be really cheaper to be worth it in my opinion.
The reason why other CM isn't popular as Raspberry Pi's CM board is the support you get a ton of support for the Raspberry Pi because the community for the Raspberry Pi is so large
Support is more important than anything else, what do you do if it doesn't boot : )
If it would be easy for newcomers to set them up, and there are some guids and information available for necomers, the communities for those alternatives could grow as well.
It's kind of like the chicken and the egg. Which one came first. Bigger community means more information and more bug fixing. But for a comunity to grow you need at least some information and bugfixing.
Thanks Jeff.
I agree.. there are so many impressive SBCs out there.. but, software/OS support and documentation makes the raspberry pi platform my go-to as well.. just easier to implement and keep stable.. for me
The third 100 pin connector at 1:08 is in a wonderful place that gives ample amounts of room for routing traces.
I am hoping the CM5 continues with the same two 100 pin connectors and adds in a 3rd such connector to form a U shape as well.
The main things I am looking for is another 1-3 PCIe lanes (Gen 2 is ample enough bandwidth.) Maybe have another set of USB, but a SATA port or two would be a god send. (Especially if the SATA controller part of the SoC supports SATA expanders!)
But yes, the Raspberry PI foundation does one thing really well, and that is support and documentation. Since tracking down obscure issues isn't fun, software is a jungle even at the best of times.
I hope if a CM5 is made in the same form factor, that's the approach it takes. It would be amazing to be able to just swap out a CM4 for a CM5 for a speed boost and whatever other cool things it would bring!
I have faith that the Pine unit will get there. I have an early Dev version of the Pinebook Pro and it has been a great device. The only reason I'm not grabbing a SOQuartz is that they went with A55 cores instead of instead of A78s.
Keep it up, nice video clip, thank you for sharing the information :)
lol that toss at the start caught me off guard
No CM4s were harmed in the making of this video ;)
A lot of those boards have amazing specs, but software support lets them down.
I was tempted by the radxa zero, but the reviews all highlighted the software issues.
I've stayed with raspberry pi because they just work for my needs.
Luckily the pi zero 2 W came out and I've picked up a few of them now.
I bought 4 or 5 of Pi Zero W2, because I have some automations to complete. It was like nobrainer to get them and have them ready in the toolbox :)
Yeah, anytime someone mentions Not-a-Pi, I think "the specs are good on paper, with the inevitable letdown on software" and am right >90% of the time. Dang shame too, because some competitors in the space would be nice. And no, 1/2 a product (the hardware and no software to use on it) is not a competitor. The Venn diagram overlap between "people who want to use this" and "people who can (and want to) develop for this" (out of the already small number interested in this at all) is nearly 0. I'd guess most of that pool is the developers of the SBC themselves.
@@IncertusetNescio yup, those raspberry pi killers but can barely do anything useful.
you are correct on supports importance. At first when i started with SBCs 5 or 6 years ago i bought a few orange pis as they were much cheaper, but i've never managed to get them stable enough, had to constantly try to hack them, so eventually i grew tired of it and bough a raspberry pi 4 which is now running home assistant and is very stable as I'm booting it from an SSD.
Woah Man you give me Idea !! TNX A LOT
Absolutely agree with you.
The Pi support, documentation and community is just so much better then any of them.( for now at least an I really hope this will get better soon)
It is so much easier to get started with any kind of project when it comes to the Raspberry hard and software.
Way back when I was trying to decide what board to buy I was somewhat bummed at the Pi's lower specs than brands such as BananaPi. But I was swayed toward the Raspberry Pi because there was so much more community support for it. That decision was the right one then and it's the right one today, as this video points out. I'm constantly stunned at so many new boards and chips released to directly compete with Raspberry but they don't seem to ever get the fact that hardware is only the beginning of a competitive product.
Yeah, those raspberry pi killers. Some of them do work with pi images but riddled with bugs. OK hardware but not much on support and community.
Yeah, this is why I went with a Raspberry Pi as a desktop computer. The support for it is excellent, and dual monitor support off the bat is beautiful. I keep looking at the alternatives, and they never can quite give me the things I need in a computer.
Totally agree
"I'm no wizard."
"Until next time, I'm Jeff Geerling."
If you're not dressed as a wizard in your next video...
It's always nice to have options on the menu, even if you're likely always to order the same meal.
Thanks Jeff for the enlighting and critical analysis. That's what I'm talking about. Facts, just the facts. :) and your insight. Good stuff.
Great review. The lack of software investment and support is a habitual problem with anything non-Pi.
Seems about right for Pine64's way of working. What they often do is make the hardware, make sure it is documented and then provide it to developers in their community to basically get software support up and running. Their phones operate on a similar principle and currently postmarketOS is coming along nicely as far as I know on those, but it has taken a while. So it could be that this board will actually work out over time as a nice (niche?) Pi alternative as there is actually an active community of developers around pine64 hardware.
To some degree it is a more open approach compared to the Raspberry Pi approach to things but it also leaves a lot up to the community to initially fix where you could argue the company itself should provide a better starting base. Anyway, I'd be interested to see how the pine64 board support develops over time.
Ditto!
Indeed. Pine64 doesn't really get involved with the software side of things at all. Software is entirely in the hands of the community of users. As a result, it takes a while for things to really get into shape.
I've been daily-driving my Pinephone for about 15 months, and the improvement over that time has just been tremendous. Sadly, I can't say the same for my Pinecube, which (since it has a much smaller user base) is just beginning to approach usability. The story has been similar for their Pinetime smartwatch, which just recently gained the ability to set the time directly on the device without having to pair it to my phone or laptop.
I like Pine64's way of doing things, but if you're not a dev who wants the challenge of bringing up basic hardware functionality, I'd suggest holding off on their products until they've been out for at least a year (Exception: The Pinecil soldering iron has been excellent since day 1).
Jeff actually points out something really important in general that a lot of starters in development forget. Reliability is key. One can have the most amazing specs and/or have an amazing price/performance ratio. It's close to worthless when it's really hard to work with. Besides all the extra time (= money in most cases) you have to put into.
I am a little confused even what these other companies are aiming for? Is it just hitching the ride with those super small SoC modules, or are just forgetting that proper documentation and accessible hardware is key?
Video started out with my reaching for the credit card for more toys.... ended with let's just wait a spell.... Thanks for the honest review!
for the PINE serial port, there's a couple of options out there that work- use a raspberry pi as your serial interface. Or, if you've got a busblaster floating around, that'll also work for you. The DSD TECH SH-U09C2 will do up to 3mb/s as well, but i don't remember if I tested it or not.
Exactly, the other boards throw out very promising hardware but few software and poor documentation.
As for me, the important consideration was that Raspibian comes up with the license for Wolfram Mathematica. None of the alternatives do.
About this CPU: having hardware USB3 and SATA is nice, but, you know, those are multiplexed in the RK3356: SATA1 with USB3 Host, SATA2 with PCIe 2.0 x1. So you can have one at the expense of the other. Which is not so nice.
Are there generic ARM CPUs which are faster than this Broadcom and still usable for small/personal computers? (I don't count M1, of course.)
I found this to be spot on when working with Orange Pi..a very strong contender with Raspberry Pi...lack of support and documentation yet they work GREAT and are powerful little boards...
Armbian has made leaps and bounds with it's development but like you said...you have to be tech savvy to keep up, however it does work...and works well in almost all cases
Nice video, looking forward to new videos, I just the Raspberry Pi was available, they say by Q4 they should be out everywhere again.... just hate the wait :)
I am excited for these but yeah, can't really beat the Rpi community for documentation and support. It would have taken forever for me to track down sound problems with my Raspberry pi 400 and UBUNTU 64 bit if there were not good documentation.
I have to agree with this video 99%. It was a detailed explanation on how is the current state of those very early rk3566 modules. But things will change quite soon. Rk3566 is a new soc and will take some time to get everything working nicely on mainline out of the box.
It definitely takes some time, and like I said at the end, I'm hopeful that time will make these boards both viable CM4 alternatives-especially for tinkerers who want to build a compute module carrier but might not be able to find a Raspberry Pi CM4 anytime soon!
@@JeffGeerling shortage sucks
Armbian seemed reasonably civilized. I have a super cheap rockchip laptop, and Armbian was pretty easy to setup. I expect hardware projects with the plain boards is more of a challenge?
The Ambian community does seem to be the most helpful in the space, and they often pick up support for the more neglected boards. I wonder if some of the board makers should try to work more directly with Armbian to make it a truly first-class distro solution, and offer ongoing support?
I still have one of the original Pine64 boards that they sold many years ago. I had no end of compatibility issues with it and ended up just abandoning the thought of using it and went back to using Raspberry Pis.
This is exactly how I felt when I bought the OrangePi 4G IoT a while ago and never got it working. The Mediatek eMMC utilities are hard to find and even then the images display in a very weird format making it incompatible with a lot of displays, and even it’s own LCD…
Support is more important than specs. I learned this the hard way with my cubox i4.
Thanks, I must admit it corresponds to my experience. Thank you as usual and until next time, you are...
Wait... who am I until next time!?
The RPI is stable, widely used and a lot of support. Storage on the rpi4 is what we are missing and some I/o peripherals like timers. Hopefully this will be fixed in the next version.
Same impression on my side, BeagleBone Black and Odroid XU4Q are quite the same story: You need to devote a helluva time on getting them to boot to prompt, then to update them, then fix the broken updates, then work some more on them, before even thinking about starting a project on them.
Raspberry Pi: Flash and pop in an SD card, switch on, start project.
It'd be cool to get a SBC with a powerful, high-end SoC including cores like the Cortex-X1 or Exynos M5. Seeing that only the vaporware tier RK3588 promises to bring the aging-but-still-miles-better-than-A72 A76 cores to SBCs is kind of sad.
But I guess it's what you get when the ARM software landscape is a minefield of incompatibilities, implementation quirks and botched support.
Nvidia Jetson...
@@Mueller3D Jetson nano has a laughably puny CPU part and Jetson Xavier has a laughably high price, so that's not really a serious option
I don't know if anyone else commented but when you see that cursor on debian install you can press Alt + F2 to open the command line, some older BIOSes don't like booting debian with a GUI.
I'd love an update on these boards It's be cool
I feel your pain and have a heap of discarded boards to prove it. The one exception for me is odroid which I have found to be quite versatile, sufficiently documented and stable. I'd take a C2 over a PI3 any day for performance especially on ethernet but my use cases don't include wifi. I can't claim I wouldn't love a C4 with 8G of ram like I can get on a PI4 but the networking and emmc of the C4 do make the choice harder despite the ram. Having uboot is certainly a plus as well.
The people who designed the Raspberry Pi really hit the nail on the head providing great software support and documentation from the get-go. Those are vital for getting you started.
I'm willing to cut the other boards some slack for using more open hardware, but I'm also comfortable working with DeviceTree and playing with firmware and bootloaders.
I agree, documents and support are the key for success. I didn't get my first raspberry pi until raspberry pi 3 b+ came out. Because of documents, support for bugs and fixes. Even third-party companies such as (Waveshare, Seeed Studio, Argon and TOFU by Oratek) have invested time developing carrier boards, components etc... for raspberry Pi's for CM3, CM4 and all RPI's Zero to 4.
So true, only sbc I bought other than pi is c.h.i.p, and it got bricked the second time I uploaded the image. I didn't get any help at that time because the company went bust.
Funnily, though I watched this video on another of Pine64's product (a Pinebook Pro laptop) which by now is quite mature and works out of the box. Proof that eventually, they can manage to release mature products.
Which is also why they have a logic of releasing "early developers access" versions of their products as fast as they can to as many developers they can: their community will eventually manage to help to document, to polish software, etc.
Probably, by their upcoming PineReader (and surely in a year or so their Pinebook 2) will be out, support for booting RK35xx hardware using entirely up-steam opensource package will be in much better shape.
I have a plastic bin of Oranges and Bananas (Orange Pi's and Banana Pi's). They were super cheap, but they'd be worthless without Armbian. Though their efforts are greatly appreciated, I grew tired of dealing with random images though. Since official Ubuntu Server images started becoming available for Raspberry Pi's, I've pretty-much stopped using the other fruit boards.
I got excited that hey maybe these alternatives would be good, but I'm again reminded of my own struggle.
I have to admit you got it spot on
yes, Raspberry Pi is very successful and using it is easy enough, because of documentation and support from community
Spot-on! Anyone who had spent time compiling the kernel knows how transient it is and how much effort maintaining and testing a custom branch is. What end user would want a 2 year old kernel with security vulnerabilities?
The recent announcement of dropping 3DNow! instructions highlights this risk; though for older CPUs, we know some of us have them. (*Cough* Busybox on MIPS and Compact Flash!)
depends whose kernel. I'm was very happy running an eight year old kernel until that hardware failed, and a 4 year old kernel today.
It's fair enough.
Usability is king. If you can't get up and running in a jiff, or sometimes at all, what's the point?
Speaking as someone who detests Linux but has to use it a fair bit, it's actually refreshing to see someone who likes it and cares about usability.
Not surprised with this. Am not in love with raspberry but it works. Am still searching for an easy way to load up around 16 CM4 modules with one standard RB Pi 4B controller. So, far no joy. I had hopes for the turning but it was a let down. I understood the design choices but it conflicted with my overall goals. Ideally, likely not feasible, I wanted to build a laptop with one master RPi4B and around 16 CM4's. Where the master could farm out tasks, service as a file server, run the display, mouse and keyboard. Where the CM4s would handle the actual tasks. I know that clustering has took a major turn backwards in the past two decades. So, I have considered writing another OS not based in WIndows, MAC or Linux. Of course finding software would be impossible if I did as i would have to write it all or convert it to another OS. PAINFUL! But at least we would have a working cluster again.
Outtakes! Woohoo!
I still have some random board sitting at my closet, that I gave up on due to lack of any upstream support. IFC6410: I think it has the same SOC as the Nexus 4, and it even uses the Android boot image format. Linaro released a few Ubuntu builds for it, then just stopped. I tried compiling an upstream kernel, and it saw the onboard SATA as having 0 ports. At that point, I gave up, and stuck it in a box in my closet, and haven't touched it since. Come to think of it, I wonder if I should just put it in E-waste?
This is exactly the problem I've had with some Orange Pis few years back.
Most board vendors just design the board (which is not *that* hard, just re-draw what's in datasheet, throw some extras and you're done. Did it, been there), build the prototype, get the "Yes I'm alive" signal from SoC and calls it a day. Some even go to such extremes as compiling a kernel or a full OS for you, but that's it. No documentation on anything, some give you u-boot thingies so you can compile yourself, but that's more rare than often. As someone already said, we need standardized boot environment for ARM devices.
Back in the RPi 3 days, I bought the Rock64 because of its specs. But I never got it to work the way I wanted.
Last week, I tried using the Rock64 as a NAS. There's prebuild images with OMV. But I still couldn't make it work.
Really interesting! I think these boards would be aimed towards integrators that don't want to pay for a pi, or whatever licensing is involved with putting them in products, so usability usually suffers there
Nah. It's greed.
Danke!
great post useful information about the Raspberry Pi competitors. You hit it right on the head, these guys throw their boards out to the market and expect consumers to debug and document their products. I wish more reviewers would warn consumers away from these manufactures who release half baked products, essentially if you can't use it out of the box its a waste of money its not ready for prime time... thanks Jeff for your useful commentary
Hey Jeff, any chance of revisiting these boards? Interested in the P64 in particular.
While some manufacturers provide good information, I generally have an issue with RockChip based SBCs. Mainly because they don't just boot from a filesystem like Raspberry Pis boot partition, but they need the binary blobs to be located at specific memory locations on the SD or eMMC storage. On top of that it's a bit of trial and error to figure out which of the blobs are the correct ones for the board (afaik this depends on things like memory speed, soc version and whether you want to boot from flash or SD).
Even if the manufacturer provides a image, third party support can be quite difficult and even then you might run into issues with the on chip firmware and you have to flash a firmware upgrade via some proprietary software that is poorly documented and apparently some versions of it might select the wrong memory location and brick the board.
5:48 Armbian requires technical know-how? Maybe on boards not yet well-supported, but my experience with Orange Pi Armbian images was pretty easy, just burn it to an SD card and do initial setup, like the Raspberry Pi.
What we really need is proper CL support on the Pi4. Sucks that we can't use the GPU at all on the Pi
Hopefully we will see a RISC-V alternative in the next few years where everything is open all the way down to to HW. As an embedded Linux developer I would be more hesitant to do development on Rockchip that is more subject to change than more established suppliers like Broadcom and Qualcomm/Atheros(even though they both have a lot of proprietary parts). You have to remember its not so much the processor(ARM, RISC-V, Tensilica etc) part you are developing for but rather all of the SoC parts(GPIO, HDMI, WIFI, serial, packet-processors, crypto-engines etc)
The only concern I have with the RISC-V chip approach is that the people paying for the boards may not care to have a standard that even looks like SBC(s). They might want to implement the RISC-V general spec for only 40% and the then use custom extension. This allows for them to produce product for specific function. However it does not mean that we get a usable SBC 'like" board. Well not a price point that we get with the Pi ecosystem.
If you have a Pi Pico, try making it into a UART interface, use the micro usb for console, and its' pins as TX/RX, just make sure you already set the usb baud the same or higher as needed. This workaround was a hackey solution I did ( and didn't document/backup/save someplace useful ) when I had a few ESP32 boards from an eBay auction and didn't know what was on them or their speed. One was at the max speed choosable for the Arduino IDE comport speed... Good luck.
Yeah I've considered this; right now I have everything powered off but the Pi itself might also be able to do 1.5 Mbps serial...
I think they release the hardware before the software and documentation is ready to help crowd source this stuff.
@Jeff: actually the A55 is newer than the A72. A55 is contemporaneous to the A75, and the A53 is contemporaneous to the A72. The true reason the A55 is slower than the A72 despite the A55's higher clock is because A55 is a "LITTLE" in-order core and an A72 is a "big" out-of-order core.
True... that was a bit of a missed statement in my editing. It's faster but not newer.
yes, my wait was worth it!
i really want a snapdragon version , of course the price, but i'm pretty sure can be possible and really nice board
Lol EXCELLENT VIDEO! You had me for a second!
Maybe yocto with something like phoronix-suite (there already is layer for that) would be good way to benchmark these. Not having yocto bsp layer is anyway deal breaker in any professional scenario for which those modules were supposedely developed (afaik there is meta-rockchip).
At this point the selling point is that it is available! One year ago I thought to myself that the PI4 would eventually be back in stock. Boo I was wrong!
Did you manage to get the soquarzt-cm4 working ? I found that manjaro arm images work .
I totally agree with you. Why people kept using RPis for years even when its performance was pretty bad, had no USB3, and limited to 100Mbps ethernet is because there's tons of information everywhere.
I used to have a Beaglebone Black board, and I still remember the lack of information compared to RPis, though it's probably 100x better situation than these RockChip boards.
Yeah. You got it. It’s the support that counts, both official and community. System integrators mostly won’t spend too much time doing vendors job…
I personally am a software guy, I expect the hardware I buy to just work so I can do my stuff. And RPi is just the safe and reliable choice. Yes, other SBCs might have more processing power, but I think if that's what limits your project you're probably doing something wrong. And others might be cheaper, but honestly that difference in price is far smaller than the value of the time I'd waste by troubleshooting some stupid little issues.
I've had similar experiences with orange pi great on paper but nightmare to get running and maintain