Andrea Pinski [Tuesday, 18 August 2026, 10:10:37 CEST]:
> On Tue, Aug 18, 2026 at 1:06 AM Matthias Kretz <[email protected]>
> wrote:
> > diff --git a/libstdc++-v3/testsuite/experimental/simd/pr114958.cc
> > b/libstdc++-
v3/testsuite/experimental/simd/pr114958.cc
> > index 94c9e0a2d18..263a2d4bab3 100644
> > --- a/libstdc++-v3/testsuite/experimental/simd/pr114958.cc
> > +++ b/libstdc++-v3/testsuite/experimental/simd/pr114958.cc
> > @@ -1,7 +1,7 @@
> >
> > // { dg-options "-std=c++17" }
> > // { dg-do compile { target x86_64-*-* } }
Changing this to 'dg-do compile { target { x86_64-*-* i?86-*-* } }' has
consequences for dg-additional-options, no?
> > // { dg-require-effective-target c++17 }
> > -// { dg-additional-options "-march=x86-64-v3" { target x86_64-*-* } }
> > +// { dg-additional-options "-march=x86-64-v3 -mno-avx512f" { target
> > x86_64-*-* } }
IIUC, an i686 compiler then would *not use AVX2* instructions, failing the
test. Do I simply extend the target pattern to { x86_64-*-* i?86-*-* } as
well? I guess it would be cleaner to use -mavx2 instead of -march=x86-64-v3 in
that case?
> > // { dg-require-cmath "" }
> > // { dg-final { scan-assembler-times "vperm(q|pd)\[\\t \]+\\\$144" 1 } }
--
──────────────────────────────────────────────────────────────────────────
Dr. Matthias Kretz https://mattkretz.github.io
GSI Helmholtz Center for Heavy Ion Research https://gsi.de
std::simd
──────────────────────────────────────────────────────────────────────────
