This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: altivec predicates
- From: Aldy Hernandez <aldyh at redhat dot com>
- To: David Edelsohn <dje at watson dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Sat, 2 Feb 2002 10:20:34 +1100
- Subject: Re: altivec predicates
>
> Yes, I understood why you took that approach, although I still
> don't like it. You can't just generate the name based on the comparison
> type and modes? It's always *vcmp{type}{mode}.
i guess we could do:
__builti_altivec_predicate_v4si (__type_eq, __CR6_LT_EQ, v1, v2)
where we would define:
#define __type_eq 0
#define __type_ge 1
etc
but that's not much cleaner (though safer).
how about keeping the current interface, but having a table of
valid opcodes, and having the builtin expander verify that it is
valid?
> Aldy> what do you suggest? i'm using predicate because that is what the
> Aldy> motorola altivec specs call them.
>
> Yes, I inferred that the naming came from Motorola Altivec
> documentation.
>
how about __builtin_altivec_compare_MODE ?
--
Aldy Hernandez E-mail: aldyh@redhat.com
Professional Gypsy Lost in Australia
Red Hat, Inc.