unless I misunderstand something (which is entirely possible), the Suse "solution" is only a partial solution... they will compile shared objects for each of the instruction set possibilities and load the one that is applicable on the current machine.

but doesn't that mean that all the base apps need to be compiled for the lowest common denominator? In the examples people (in this thread) are worrying about, that would be "v2".

Fred


On Mon, Jan 8, 2024 at 7:45 AM aleksander.baranowski via CentOS-devel <centos-devel@centos.org> wrote:
As far as the idea (glibc-hwcaps) is concerned, both are pretty much the
same.

I think
https://gitlab.com/x86-psABIs/x86-64-ABI/-/jobs/artifacts/master/raw/x86-64-ABI/abi.pdf?job=build
is a good read. It's fresh and it might be appropriate to read it first
before entering the deeper discussion.

If Fedora manages to maintain baselines, the downstream might follow
(CentOS Stream, RHEL 10). I don't think it will be mainstream the other
way around. Also if production started it might be hard to push such a
huge change ¯\_(ツ)_/¯.

Best,
Alex

On 1/8/24 12:49, František Šumšal wrote:
> On 1/8/24 12:46, Trevor Hemsley via CentOS-devel wrote:
>> I do think the SuSE method of dealing with this is significantly
>> better and I wish RH would adopt it rather than just deprecating
>> otherwise working hardware:
>>
>> https://www.theregister.com/2023/03/09/opensuse_finds_x86_64_solution/
>
> FWIW Fedora is currently considering this as well:
> https://fedoraproject.org/wiki/Changes/Optimized_Binaries_for_the_AMD64_Architecture
> _______________________________________________
> CentOS-devel mailing list
> CentOS-devel@centos.org
> https://lists.centos.org/mailman/listinfo/centos-devel
_______________________________________________
CentOS-devel mailing list
CentOS-devel@centos.org
https://lists.centos.org/mailman/listinfo/centos-devel