RE: [PATCH v3] Support AVX10_V2_AUX instructions
"Jiang, Haochen" <[email protected]>
| Newsgroups | gmane.comp.gnu.binutils |
|---|---|
| Message-ID | <SJ5PPF77D28E3C2C5E72FE66F2460F732ADECAC2@SJ5PPF77D28E3C2.namprd11.prod.outlook.com> |
> 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. > Interesting, especially when at the top of > https://x86ecosystem.org/resource/ai-compute-extensions-ace- > specification/ > is says "June 15, 2026". > > > also, instead of the AVX10_Vn_AUX insanity, has > > anyone considered AVX10_AUX_Vn for naming? > > you know, move the Vn further away from AVX10, > > and over to the AUX side, where it belongs? > > Now that it's clear that the Vn here isn't intended to have relationship > with the AVX10 version itself, this may indeed be a little better. Yet > then, with AVX10_V1_AUX not being the baseline of AVX10_V2_AUX (as per > what the patch still does, despite my earlier question), the numbering > here is questionable altogether. Exactly, AVX10_V1_AUX and AVX10_V2_AUX will be independent bits, just like most of those AVX512 stuffs. They might be included into a AVX10 version in future. AVX10_V1_AUX already did that for AVX10.2. And as I mentioned before, I am not a big fan of that naming since it carries no actual meaning to me and might cause confusion on numbers, just like this scenario. Thx, Haochen