Re: [PATCH] Optimize VEC_PERM_EXPR with same permutation index and operation [PR98167]

[Jakub Jelinek <jakub@redhat.com>]

On Wed, Nov 16, 2022 at 04:30:06PM +0100, Richard Biener via Gcc-patches wrote:
> On Wed, Nov 16, 2022 at 4:29 PM Richard Biener
> <richard.guenther@gmail.com> wrote:
> >
> > On Wed, Nov 16, 2022 at 4:25 PM Tamar Christina <Tamar.Christina@arm.com> wrote:
> > >
> > > Hi,
> > >
> > > This patch is causing several ICEs because it changes the permutes from a single register
> > > permute to a multi register due to the lowering of the expressions to different SSA names.
> > >
> > > See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107717
> > >
> > > I have a prototype fix which adds a new rule to CSE this back to a single register permute,
> > > but would this be the right solution? It seems hard to later on during expand realize that
> > > the two operands are the same.
> > >
> > > It's probably also ok to just block this from happening after vec_lower, however I'm worried that
> > > If it wasn't CSE'd before vec_lower it'll lower it so something much less efficient.
> >
> > You can use
> >
> >  (vec_perm (op@7 @0 @1) @3)
> 
> Err, (vec_perm (op@7 @0 @1) @7) obviously.

Even:

--- gcc/match.pd.jj	2022-11-15 07:56:05.240348804 +0100
+++ gcc/match.pd	2022-11-16 16:35:34.854080956 +0100
@@ -8259,7 +8259,7 @@ and,
  (simplify
   (op (vec_perm @0 @0 @2) (vec_perm @1 @1 @2))
    (if (VECTOR_INTEGER_TYPE_P (type))
-    (vec_perm (op @0 @1) (op @0 @1) @2))))
+    (vec_perm (op@3 @0 @1) @3 @2))))
 
 /* Similar for float arithmetic when permutation constant covers
    all vector elements.  */
@@ -8298,4 +8298,4 @@ and,
 	 }
       }
       (if (full_perm_p)
-	(vec_perm (op @0 @1) (op @0 @1) @2))))))
+	(vec_perm (op@3 @0 @1) @3 @2))))))

From quick look at the dumps, it seems to work fine.

	Jakub