This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH, rs6000] Don't use vec_all_eq for vector double in Altivec tests
- From: David Edelsohn <dje dot gcc at gmail dot com>
- To: Bill Schmidt <wschmidt at linux dot vnet dot ibm dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Mon, 24 Feb 2014 23:27:10 -0500
- Subject: Re: [PATCH, rs6000] Don't use vec_all_eq for vector double in Altivec tests
- Authentication-results: sourceware.org; auth=none
- References: <1393295445 dot 20991 dot 94 dot camel at gnopaine>
On Mon, Feb 24, 2014 at 9:30 PM, Bill Schmidt
<wschmidt@linux.vnet.ibm.com> wrote:
> Hi,
>
> Some of the recently introduced Altivec tests pass for -mcpu=power8, but
> fail to compile for older processor models. This is because the
> vec_all_eq interface, when used with vector double arguments, maps to a
> built-in that is only available on Power8 and later hardware. This
> patch modifies the tests to avoid using vec_all_eq for vector double
> arguments.
>
> Bootstrapped and tested on powerpc64-unknown-linux-gnu with
> -mcpu=power7, and on powerpc64le-unknown-linux-gnu with -mcpu=power8,
> with no regressions. Is this ok for trunk?
>
> David, this should clear up the recent failures on AIX.
>
> Thanks,
> Bill
>
>
> gcc/testsuite:
>
> 2014-02-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
>
> * gcc.dg/vmx/ld-vsx.c: Don't use vec_all_eq.
> * gcc.dg/vmx/ld-vsx-be-order.c: Likewise.
> * gcc.dg/vmx/ldl-vsx.c: Likewise.
> * gcc.dg/vmx/ldl-vsx-be-order.c: Likewise.
> * gcc.dg/vmx/merge-vsx.c: Likewise.
> * gcc.dg/vmx/merge-vsx-be-order.c: Likewise.
Okay.
Thanks, David