x86-64: Use YMM registers in memcmpeq-evex.S
Checks
Context |
Check |
Description |
redhat-pt-bot/TryBot-apply_patch |
success
|
Patch applied to master at the time it was sent
|
redhat-pt-bot/TryBot-32bit |
success
|
Build for i686
|
linaro-tcwg-bot/tcwg_glibc_build--master-arm |
success
|
Testing passed
|
linaro-tcwg-bot/tcwg_glibc_check--master-aarch64 |
pending
|
Patch applied
|
linaro-tcwg-bot/tcwg_glibc_build--master-aarch64 |
success
|
Testing passed
|
linaro-tcwg-bot/tcwg_glibc_check--master-arm |
success
|
Testing passed
|
Commit Message
Include x86-evex256-vecs.h by default to use YMM registers in
memcmpeq-evex.S.
---
sysdeps/x86_64/multiarch/memcmpeq-evex.S | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Comments
* H. J. Lu via Libc-alpha:
> Include x86-evex256-vecs.h by default to use YMM registers in
> memcmpeq-evex.S.
> ---
> sysdeps/x86_64/multiarch/memcmpeq-evex.S | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/sysdeps/x86_64/multiarch/memcmpeq-evex.S b/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> index 3666f649cd..7eefff77f1 100644
> --- a/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> +++ b/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> @@ -42,7 +42,7 @@
> # endif
>
> # ifndef VEC_SIZE
> -# include "x86-evex512-vecs.h"
> +# include "x86-evex256-vecs.h"
> # endif
> # include "reg-macros.h"
It would be nice if the commit message could describe the why, not just
the what. 8-)
Thanks,
Florian
On Thu, Jun 1, 2023 at 12:41 AM Florian Weimer <fweimer@redhat.com> wrote:
>
> * H. J. Lu via Libc-alpha:
>
> > Include x86-evex256-vecs.h by default to use YMM registers in
> > memcmpeq-evex.S.
Should we create an evex512 impl?
> > ---
> > sysdeps/x86_64/multiarch/memcmpeq-evex.S | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/sysdeps/x86_64/multiarch/memcmpeq-evex.S b/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> > index 3666f649cd..7eefff77f1 100644
> > --- a/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> > +++ b/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> > @@ -42,7 +42,7 @@
> > # endif
> >
> > # ifndef VEC_SIZE
> > -# include "x86-evex512-vecs.h"
> > +# include "x86-evex256-vecs.h"
> > # endif
> > # include "reg-macros.h"
>
> It would be nice if the commit message could describe the why, not just
> the what. 8-)
>
We use -evex.S postfix for 256-bit evex. This was mistakenly implemented as
evex512.
> Thanks,
> Florian
>
On Thu, Jun 1, 2023 at 8:44 AM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>
> On Thu, Jun 1, 2023 at 12:41 AM Florian Weimer <fweimer@redhat.com> wrote:
> >
> > * H. J. Lu via Libc-alpha:
> >
> > > Include x86-evex256-vecs.h by default to use YMM registers in
> > > memcmpeq-evex.S.
>
> Should we create an evex512 impl?
YMM is good enough for now. There is no need for ZMM.
> > > ---
> > > sysdeps/x86_64/multiarch/memcmpeq-evex.S | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/sysdeps/x86_64/multiarch/memcmpeq-evex.S b/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> > > index 3666f649cd..7eefff77f1 100644
> > > --- a/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> > > +++ b/sysdeps/x86_64/multiarch/memcmpeq-evex.S
> > > @@ -42,7 +42,7 @@
> > > # endif
> > >
> > > # ifndef VEC_SIZE
> > > -# include "x86-evex512-vecs.h"
> > > +# include "x86-evex256-vecs.h"
> > > # endif
> > > # include "reg-macros.h"
> >
> > It would be nice if the commit message could describe the why, not just
> > the what. 8-)
> >
>
> We use -evex.S postfix for 256-bit evex. This was mistakenly implemented as
> evex512.
>
>
> > Thanks,
> > Florian
> >
@@ -42,7 +42,7 @@
# endif
# ifndef VEC_SIZE
-# include "x86-evex512-vecs.h"
+# include "x86-evex256-vecs.h"
# endif
# include "reg-macros.h"