This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFC/RFT] Tree-level lowering of generic vectors, part 2


On Wed, Jul 21, 2004 at 12:46:07PM +0200, Paolo Bonzini wrote:
> >Using minus instead of negate gets the wrong answer for floating point.
> 
> This is done unconditionally by optabs for MODE_FLOAT (not for 
> MODE_VECTOR_FLOAT because control has already been passed to 
> expand_vector_unop):

Which does not change the face that it gets the wrong answer.

> I will do it as a follow-up.

Ok.

> +/* Build a constant of type TYPE, made of VALUE's bits replicated
> +   to fit TYPE's size.  */
> +static tree
> +build_replicated_const (tree type, tree inner_type, HOST_WIDE_INT value)

You failed to update the comment.

> +  /* All powers of two <= 32 give a different result modulo 37.  */

Still outdated comment.

> +  NEXT_PASS (pass_lower_vector_ssa);

Don't add this until we've actually merged the vectorizer.
Even then, we'll want to have it controled by the vectorizer 
actually doing something.

Ok with those changes.


r~


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]