RE: [PATCH v3] Support AVX10_V2_AUX instructions

> From: Jan Beulich <[email protected]>
> Sent: Friday, August 28, 2026 2:32 PM
> 
> On 28.08.2026 08:17, Jiang, Haochen wrote:
> >> From: Jan Beulich <[email protected]>
> >> Sent: Friday, August 28, 2026 1:55 PM
> >>
> >> On 27.08.2026 21:48, Christian Ludloff wrote:
> >>>> In fact doc version 1.15
> >>>
> >>> latest ACE spec is 1.16.2
> >>
> >
> > I don't know that has been updated either. It seems to me
> > that some of my comments are not addressed. (Here I mean
> > the exception type) I will mention that.
> >
> > It seems to me that FP8 to FP4 convert modrm is fixed, it is a clear
> > bug in documentation, while FP8 to FP6 still remaining a
> > questionmark to me.
> 
> That is (has been) slightly odd, yes, but at least all pieces say the
> same now. I'm therefore inclined to take the patch as is, accepting that
> a fix may be needed if the doc (and implementation?) was changed.
> 

Let's go with that and hopefully everything will reach an agreement in
the future. At least Intel and AMD should not ship HW with different
behavior on the same encoding. It then will become a nightmare.

And both modrm.r/m in operand 1 or modrm.reg in operand 1 could
be ok since they both have reasons. modrm.r/m in operand 1 is the
same way we did in FP8 to FP4. Just because it is an odd type so there
is no memory operand allowed for that operand. But functionality-wise,
they are similar, where I slightly lean to this encoding. modrm.reg in
operand 1 is also ok since no memory is involved. It is not a wrong
assignment.

Thus, let's do not block the patch since we still need ACE patch based
upon that.

Thx,
Haochen