Re: Future of armv7

yes, this is similar to what i want to do

except i don't have products sold yet. so no loss i guess

i saw armv7 allwinner h3 being most mature juding from wiki so i got 13 of nanopi neo cores without headers soldered on. this was during covid chip crisis, had no other option

the soc seems to be in lts stage and it's still sold, i didn't expect it to be restocked. but it was

so i went on testing and then was like wait 32bit arm is going away now?

i get why 32bit is hassle but those are still made, don't have replacement support and are powerful enough to be used. i was surprised what this postage stamp can do. compares to bbb

unsure how, but we need like couple of platforms for each purpose to like form full embedded infra set

no idea if not enough companies use fbsd on embedded or what

we could have a set of

headless low power io
headless server
headless network high throughput
headless network low throughput
video send
video receive
wifi ap / sta
lte wan uplinks

network could support switch chips, sfp modules for fiber, single pair ethernet variants

now, me, even if i'm able to sell something and then hire a dev, are unable to fully realize that dream alone

you could gobble up above with existing hw but it leaves ton to be deserved, especially if you want hw to be (relatively) open too

so unsure how to like achive it. who's going to write that. surely not me. and if me, surely not fast or in quality required. simply not possible. need a redundant array of indepent brains for all this

in case of myself, i have helped (embedded) fbsd to go forward a bit (goood!) but that's all



On December 17, 2025 8:53:35 PM GMT+02:00, Oskar Holmlund <[email protected]> wrote:
>Hi,
>
>In our(*) products we use FreeBSD current and it runs on an ARMv7 SoC.
>We have sold ~3k units on the current hardware generation and will be on
>the market until 2032:ish and then be in maintenance for another 10 years.
>
>Our products are used in the paper industry and are therefore used
>around the clock. We have not received any reports of instability
>originating from the operating system itself so far. All defects have
>been due to the application running on top of the operating system.
>
>In our case we will solve it even if ARMv7 support will be removed, we
>dont rely on the ports-tree neither the binary releases of FreeBSD. Just
>a couple more local patches that needs to be applied and maintained.
>
>Of course its sad to see it go away, it will be a little bit tricky to
>explain for people who care about the content of the SBOM that we
>actually still use FreeBSD current code even if the ARMv7 support is not
>present. Probably we will rename it to something else to avoid questions.
>
> From my point of view, ARMv7 still have a place in the world,
>especially in "embedded systems". Its fast enough to have a GUI up in
>less than 10s, do the thing it supposed to do and be rock solid. Of
>course ARMv7 is worthless to have as a build machine to build world or
>browse the web, but who will do something like that? :)
>
>The SoC vendors seems to have a similar view of their products for
>example NXP i.MX 6 is planned to be manufactured until December 2035 and
>AMD Zynq 7k until 2040.
>
>Maybe we can start cleaning up the ARM subtree, make sure we have
>developers interested in maintaining EVKs that are actually available in
>the market for new developers to build upon. We dont need to provide
>binaries, its enough that the code base compiles and is able to boot
>relevant EVKs to be interesting for developers to take it to the next level.
>
>(* I'm a long-time freelancer for the company that makes the machines,
>so technically it's not "ours", but having worked with them for the last
>15 years, it's a bit "ours")
>
>--
>Bästa Hälsningar
>Oskar Holmlund
>Tel 070-3220292
>