[PATCH] x86: Remove AVX10.2 256 bit rounding support

[Jiang, Haochen]

> From: Jan Beulich <jbeulich@suse.com>
> Sent: Wednesday, March 19, 2025 5:32 PM
> 
> On 19.03.2025 10:17, Haochen Jiang wrote:
> > AVX10 whitepaper just got a major change that all the platforms will
> > support 512 bit vector width (previously, E-core is up to 256 bit,
> > leading to hybrid clients and Atom Server 256 bit only).
> >
> > Ref: https://cdrdv2.intel.com/v1/dl/getContent/784343
> 
> Before I look at the patch in any detail: That's only the Technical Paper,
> not the specification. Are you sure you also mean to remove that capability
> from the spec? I.e. not allowing re-consideration in the future, or others
> to implement only a 256-bit variant? Even if so, should we really rip out
> everything, rather than just making that feature default-disabled?
> 

Yes, this means removing that capability from SPEC.

Making that feature default-disabled might be a choice, with a new Binutils
only CPUID.

But I doubt if we really need that since they will not exist on any CPU anymore.

Thx,
Haochen