Raspberry Pi has the biggest community, and therefore it’s the best choice for anyone looking to get started.
If you’re the type of person who needs very specific features like AES acceleration, then you don’t want to invest in a board aimed at education markets anyway.
The Raspberry Pi 4 is actually very performant due to the Cortex-A72 cores, which are not common on other SBCs at low price points. It’s actually an incredible performance value in the space, even if it doesn’t fit everyone’s exact situational requirements.
I have seen high endurance ones available.
https://www.digikey.com/en/product-highlight/a/atp/advancedm...
I even set up a testing rig to do randomized power shut down thousands of times; they didn't fail:
https://forums.balena.io/t/raspberry-pi-atp-amlc-microsd-car...
Found these via this 2018 HN comment:
There have also been, for a long time now, SD cards with much better durability than the cheapest ones you can buy on Amazon. I personally look for ones labeled "high-endurance" or for use with dashcams because those are much closer to a proper SSD in terms of wear cycles.
I wish the RPi foundation would make it clearer in their documentation and blog posts you're going to have a bad time if you use a generic SD card as if it was a general-purpose hard drive.
eMMCs are made pesirically for this purpose and its a shame there is bo pi model using them
I'm fully aware of that. In fact, and ironically enough, I'm one of those people you are referring to: https://news.ycombinator.com/item?id=26157437
>I wish the RPi foundation would make it clearer in their documentation and blog posts you're going to have a bad time if you use a generic SD card as if it was a general-purpose hard drive.
Conversely, I wish they would tell all their approved sellers not to include those "generic" SD cards (Raspbian/RPi OS preinstalled and all) with Pi's.
I'd bet 80% of the sales are to hobbyists/companies.
It's a weird setup. Have a look at this page: https://www.raspberrypi.org/about/meet-the-team/
There are two groups of people:
* Raspberry Pi Foundation (100+)
* Raspberry Pi Trading (27)
My understanding: The people in "Raspberry Pi Trading" do the hardware/software/documentation/product management/sales work, plus publish the monthly MagPi magazine. That's remarkably few people for this, btw. The profit of this operation goes to support the foundation, per the wikipedia page. There sure seems to be a lot of fluff there.
I think it would be wise to change that balance (100+ vs 27 full times) a little bit more in advantage of Raspberry Pi Trading part.
Hobbyists are part of the educational target. Raspberry Pi has introduced more people to embedded Linux than any other project.
We didn’t always live in an era of cheap SBCs. Before Raspberry Pi, the barriers to entry were much higher and much more expensive.
As much as people like to criticize the Pi for not having every possible feature or not being completely open source at every level, it doesn’t actually matter for 99% of projects. It gets the job done cheaply, quickly, and with a huge support community behind it.
The Raspberry Pi CM4 is now available for anyone who needs more granular access to IO or extra features like eMMC.
I bought myself a rk3399 based compute module a while ago, the support was... sparse.
* The small amount of documentation that does exist is vague and incomplete. Many of the links were actually broken, so I had to fall back to using the docs for older models.
* The key expired on their apt repo and they never bothered to renew it.
* The selection of prebuilt Linux distributions is small and several generations behind.
* The community is small, so you can’t rely on things like stackoverflow as much as you might in PI land.
Overall, I had a good time with it, but getting started on a PI would have been so much easier.
Now if this were the rx3588 with the A76s it would be a different story.
Even so, it seems the 1.5Ghz is more about running the system without a heatsink. Give one a heatsink and they run at over 2Ghz. The perf bump from 1.5 to 2.2Ghz is not insignificant.
https://www.jeffgeerling.com/blog/2020/raspberry-pi-400-can-...
https://hothardware.com/reviews/hot-clocked-pi-raspberry-pi-...
Also: Arm itself says the A55 is ~18% faster than the a53. So, its still quite slow.
[0] https://d-i.debian.org/daily-images/arm64/daily/netboot/SD-c...
There are plenty of things that an RK3399 (or the new Rockchip being used in the OP) can handle just fine.
I was definitely interested to see that this new chip supports up to 8GB of RAM. I think there are a lot of interesting applications for RAM-rich, Gig-E bearing nodes that can run off a battery. (Examples: Redis at the edge, “Field” PostgreSQL replicated to the cloud when possible, nested VMs protecting trusted “firmware” from remotely-loaded containers and machine images, etc.)
Among other things, this means the PCIe aperture is squeezed down to a very measly 2×16MB — which is not even remotely enough for a lot of PCIe cards.
Also, if you install 4GB RAM, you only get to use something like 3.5G.
Oh and if I remember correctly, PCI MSI are broken, which again limits the PCIe card selection even more...
http://opensource.rock-chips.com/images/e/ee/Rockchip_RK3399...
Page 13 ("1.1 Address Mapping") - same thing, everything crammed into 32bit.
PCIe Aperture is documented in TRM Part 2:
https://www.t-firefly.com/download/Firefly-RK3399/docs/TRM/R...
Page 711 - and I misremembered, it's 64MB total in 1×32MB + 32×1MB blocks. For comparison, my 6 year old server has 64GB PCI aperture space on each of its 2 sockets, my Laptop has 16GB. (Obviously both >32bit address spaces... which the RK3399 really should be too, but isn't.)
MSI is blacklisted in the Linux kernel somewhere if I remember correctly; it just randomly loses interrupts.
P.S.: Address space doesn't need to make a 32->64bit jump, my laptop CPU has 39bit, my server has 46bit; even 36bit (like old Intel 32bit PAE) would've been sufficient on the RK3399. But 32bit isn't.
P.P.S.: the PCI aperture size is the most obviously a problem when you try to plug in a GPU. E.g.:
0c:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Baffin [Radeon RX 550 640SP / RX 560/560X] [1002:67ff] (rev cf) (prog-if 00 [VGA controller])
...
Region 0: Memory at 900000000 (64-bit, prefetchable) [size=4G]
Region 2: Memory at 880000000 (64-bit, prefetchable) [size=2M]
Region 4: I/O ports at f000 [size=256]
Region 5: Memory at f7e00000 (32-bit, non-prefetchable) [size=256K]
...
Capabilities: [200 v1] Physical Resizable BAR
BAR 0: current size: 4GB, supported: 256MB 512MB 1GB 2GB 4GB
This GPU actually allows you to reduce the size it needs, but the absolute minimum you can set it to is 256MB...The real problem is that PCI address space is not normally a scarce resource, and some RAID controllers & network cards (particularly high-performance/server ones) use a random bunch of it (>= 128MB) , and that's not even starting on SR-IOV. And these requirements aren't well documented because PCI memory space is not normally a problem. If you're lucky you can google "<card name> lspci", if not you only notice after you bought the card...
https://www.hardkernel.com/shop/odroid-n2-with-4gbyte-ram-2/