Re: freeze for sbcl-2.6.8
Stas Boukarev <[email protected]>
| Newsgroups | gmane.lisp.steel-bank.devel |
|---|---|
| Message-ID | <CAF63=12dKra0oDzEFu7KgM4rT88P02KkvsazNnKpkn6K4D5DhA@mail.gmail.com> |
Well, avx2 is older than xsavec, so can't use it to enable avx2 support. On Sun, Aug 23, 2026 at 12:42 AM arthur miller <[email protected]> wrote: > > Of course it is not, and that wasn't what I said either. I just asked a simple question, > so I know if I should spend time on something or not. Xsavec has been out for 10 > years, so I am sure you and other people here are aware of it. > > Just asking if it is not in SBCL because nobody had time or interest > to add it or if there is some technical reason why it is not in use, that is all. Sorry, if I > bother with a regression, was just a curious question that popped on my mind. > ________________________________ > Från: Stas Boukarev <[email protected]> > Skickat: den 22 augusti 2026 22:41 > Till: arthur miller <[email protected]> > Kopia: [email protected] <[email protected]> > Ämne: Re: [Sbcl-devel] freeze for sbcl-2.6.8 > > xsavec is not synonymous to avx2. > > > > On Sat, Aug 22, 2026 at 11:35 PM arthur miller <[email protected]> wrote: > > > > Yeah of course, I understand; the question was out of context and not framed well. > > > > If I frame it like this: is there a technical reason to not use xsavec/xsaves/xrestors > > in SBCL instead of xstore/xrstor? Or is it not in there just because it is not done yet. > > > > Not for this particular case, it is a regression, but wonder if it is something for a future > > patch or not? Should have ask back whan I was working on zmm support, but than it > > was a bit too much other stuff. > > ________________________________ > > Från: Stas Boukarev <[email protected]> > > Skickat: den 22 augusti 2026 22:19 > > Till: arthur miller <[email protected]> > > Kopia: Douglas Katzman <[email protected]>; Christophe Rhodes <[email protected]>; [email protected] <[email protected]> > > Ämne: Re: [Sbcl-devel] freeze for sbcl-2.6.8 > > > > xsave (for avx512) isn't invoked unless avx512 registers are used > > within a function. So, xsavec isn't going to improve anything. > > > > On Sat, Aug 22, 2026 at 11:02 PM arthur miller <[email protected]> wrote: > > > > > > Forgive me for chimming in, but a question, since you are already > > > talking about xstore/xrstor and want more performance. Can we switch to > > > "compact" versions of those (xsavec & co)? These, potentially > > > write/restore less data, in cases where saving/restoring xmm/ymm/zmm is > > > enabled. > > > > > > I learned about it when I was patching for zmm. I didn't dare to try > > > to change that, but instead patched in support to what was already in. > > > > > > As I understand it, when full store is needed (zmm registers in use), > > > than the performance is roughly the same, but if zmm/ymm/xmm regs do not > > > need to be saved becuase they weren't written since the last clear, they > > > won't be saved either. > > > > > > Compacted stores/restores are supported on Linux since kernel 4.4, year > > > 2016, and on Windows since version 10 some release number in 2015. > > > ________________________________ > > > Från: Douglas Katzman via Sbcl-devel <[email protected]> > > > Skickat: den 22 augusti 2026 18:07 > > > Till: Stas Boukarev <[email protected]> > > > Kopia: Christophe Rhodes <[email protected]>; [email protected] <[email protected]> > > > Ämne: Re: [Sbcl-devel] freeze for sbcl-2.6.8 > > > > > > I have the fix for that crash. I mistakenly thought that FPR-SAVE/FPR-RESTORE no longer need to preserve RDX when using XRSTOR but they still do - but only for #+win32, so I'll add that. > > > However, Gemini claims the following theoretical performance gain by simply avoiding XSAVE/XRSTOR entirely in the YMM case (not the avx512 of course) which is a very interesting point - > > > > > > ~4x Faster: 16 * VMOVAPS runs in ~15–20 cycles vs 80+ cycles for XSAVE/XRSTOR. > > > > > > Do you think I can commit this register changing patch + its fix for now, and we'll leave the above suggestion for next release (despite also being a valid fix for my crash) ? _______________________________________________ Sbcl-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/sbcl-devel