Good evening,
my interest in Power architecture got started when I read in February 2020
about the new open license model on a well-known German news ticker [1]. I
attended the OpenPOWER Virtual Summit 2020 North America [2] in September
and heard there the first time about Libre-SOC [3], previously also known
as Libre-RISCV SoC [4] before they switched from RISC-V to Power. During
FOSDEM 2021, I noticed their LibreSOC Project stand [5] and asked multiple
beginner questions, which Luke Kenneth Casson Leighton (on Cc) nicely took
care of (thank you!).
Given an idea is a Power-based SBC [6], I wondered whether I could just run
a Fedora for ppc64le (which seems to currently target POWER8 LE and later),
on it. Unfortunately Luke's answer was no, because LibreSOC won't support
VSX [7], which are IBM's SIMD (six hundred?) instructions, but the LibreSOC
Project plans to have Simple-V Vectorisation according to their FOSDEM talk
[8]. Note, the A2O POWER processor core [9] doesn't support VSX, too (same
seems to apply to A2I and Microwatt, but they're either POWER7 or likely not
powerful enough for typical PC usage anyway -> outside of Fedora's target?).
Trying to follow the path on how to get Fedora for ppc64le to a "POWER-Pi",
I had to learn that glibc developers seem to treat IBM POWER9 strictly equal
to ABI rules rather handling the different features more fine granulated. My
understanding here might be imprecise or wrong, this is what I understood
and I'm no expert at this topic. Nevertheless I'm trying to precise it a bit
more:
The IBM Power ISA Version 3.1 [10] specifies on page 8 compliancy subsets,
where the Linux Compliancy Subset (LCS) is explained on page 10 ff. Page 11
refers specifically to the "Scalar Float Optional Features" and mentions in
the first row of the table VSX. However what seems to be implemented e.g.
in glibc is something like "#ifdef POWER9" rather "#ifdef HARDWAREFEATURE",
which effectively makes VSX support mandatory rather optional.
Given the only publicly known modern implementor of VSX right now, not on
an EOL schedule, seems to be IBM's POWER9, it would from my point of view
be helpful that VSX support in at least glibc is made optional.
When asking whether the new glibc-hwcaps [11] could be a solution, Luke's
answer read for me like a conditional "yes", because at least VSX should be
covered by glibc-hwcaps then (again, same as above...my understanding, I'm
still no expert here). But this is why I'm Cc-ing here Florian Weimer, who
implemented [12] this new glibc feature. I'm not sure if Florian is the
right person for this topic at all, but maybe he knows the right people? :)
Finally, to close the circle to the subject of my mail: (how) can we reach
Power support for Fedora (and other mainstream Linux distributions) beyond
IBM POWER (e.g. Libre-SOC)?
[1] https://www.heise.de/newsticker/meldung/Prozessor-ISA-Power-OpenPower-Found…
[2] https://openpowerfoundation.org/events/openpower-summit-2020-north-america/
[3] https://libre-soc.org/
[4] http://lists.libre-riscv.org/pipermail/libre-riscv-dev/2019-October/thread.…
[5] https://stands.fosdem.org/stands/the_libresoc_project__a_hybrid_3d_cpu_vpu_…
[6] https://libre-soc.org/22nm_PowerPI/
[7] https://en.wikipedia.org/wiki/AltiVec#VSX_.28Vector_Scalar_Extension.29
[8] https://fosdem.org/2021/schedule/event/the_libresoc_project_simple_v_vector…
[9] https://openpowerfoundation.org/a-deep-dive-into-a2i-and-a2o/
[10] https://ibm.ent.box.com/s/hhjfw0x0lrbtyzmiaffnbxh2fuo0fog0
[11] https://sourceware.org/pipermail/libc-announce/2021/000030.html
[12] https://sourceware.org/pipermail/libc-alpha/2020-June/115250.html
Regards,
Robert
