Re: What's the plan for powerpc64 in FreeBSD 16

On Sun, 30 Nov 2025 at 10:39, Minsoo Choo <minsoochoo0122@proton.me> wrote:

>
> This argument doesn't make sense. Using less memory doesn't necessarily mean going back to 32-bit architecture. Everyone wants to consume less memory, but then why 32-bit architectures came out after 16-bit and 64-bit architecture came out after 32-bit? Sometimes the advance of technology doesn't allow us to consume the amount of memory we used to.

Hi! I work on wearables tech at a large company! 32 vs 64 bit
footprint actually absolutely matters
for memory and power consumption!


> And people learned lessons from the fragmentation era and that's why everyone is working hard to defragment the ecosystem.
>
> RISC-V has profiles without enforcement, but still, companies like SiFive and Tenstorrent stick with them voluntarily. Why do you think they do that?
>

> Arm ISA is strictly protected by Arm and companies can choose an ISA version (like armv8.5) to ensure that their chips work without much effort on software side.

Having done like, half a dozen chipset bring-ups and debugging across
cortex-A, cortex-M and Risc-V 32 bit
cores at my current job I would just like to add that outside of the
CPU bring-up errata, it's 2025 and almost
no-one is doing just a CPU architecture bring-up anymore.

Like, I get it. We migrated to LLVM which didn't have all the platform
support that we really wanted.
We're trying to build package sets including things like
firefox/chrome on target platforms like armv7
where people aren't using it for that. Our build files, the CPU
architecture support, the pmap stuff
needed for older platforms that may be seen as hindering VM work, I
get it. It's a lot of work.
Almost noone is being paid to work on that stuff upstream.

But like, I also don't buy the "kill it" argument for some stuff when
we actually have users and interest,
and maybe if we didn't treat things as a "crappy platform" or "I hate
working on this" - see how
the FreeBSD developer community has treated the Raspberry Pi platforms
as an example - we wouldn't
keep hitting these issues.

FreeBSD "Runs" on armv8, and "runs" on risc-v, but we absolutely do
not "run" run on them. Maybe
it runs enough on your current platform, but we're missing like 90% of
what would make it run really
well outside of a couple of armv8 platforms.

Noone is shipping a 32 bit or 64 bit CPU with a data/address bus and
off the shelf peripherals that
"almost already work". Once you've sunk the cost into the CPU
architecture, the rest of it is errata
and then all the peripherals, all of the clock/power control, all of
the firmware interfacing, etc.

We "boot" on an armv8 "based" qualcomm snapdragon X1. The ACPI table
is enough to bring up
basic hardware. If we want anything else, we're like a hundred tiny
drivers away from really "supporting"
that as a platform.

This discussion about CPU architectures is pretty hilarious. I get it,
I really do, and there is a cost, I also
get that. But it's really hard to give these discussions actual weight
when they may be technically
correct, but not actually "usable" correct.

> OpenPower states that companies cannot promote their chip as Power if tbeir chip does not implement all the required instructions in the profile.
>
> We are not living in pre-1990s anymore.

See comment above. Noone's buying a straight CPU with a data/address
bus and external
peripherals anymore, and again the cost isn't in the CPU architecture,
it's in all of the stuff
around it.

I'd also like to call out that like, my experiences with 64 bit risc-v
hardware hasn't been all
that great, and I do wonder if we have more armv7 users than risc-v users. :-)



-adrian