Hi!
On Tue, Aug 07, 2018 at 02:25:06PM -0500, Will Schmidt wrote:
> Some testcases to exercise the vec_splat() built-in.
> In building and updating these tests I consciously violated the
> 80 char per line rule, as I was doing experimentation with the
> assorted values and making lots of local updates. I've left them as-is
> just because, but can update that if so desired.
That is fine in testcases (and even practically unavoidable in many
dg statements).
> +// vec_splat() using variable vectors should generate the vspltb instruction.
> +/* { dg-final { scan-assembler-times "vspltb" 24 } } */
> +// vec_splat() using a constant vector should generate a load.
> +/* { dg-final { scan-assembler-times {\mlvx\M|\mlxvw4x\M} 3 } } */
You may want to use {} and \m\M everywhere, for symmetry and for future-
proofing things a bit. It's not necessary of course.
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/powerpc/fold-vec-splat-floatdouble.c
> @@ -0,0 +1,54 @@
> +/* Verify that overloaded built-ins for vec_splat with float and
> + double inputs produce the right code. */
> +
> +/* { dg-do compile } */
> +/* { dg-require-effective-target powerpc_vsx_ok } */
> +/* { dg-options "-maltivec -mvsx -O1" } */
-mvsx implies -maltivec.
Is there any reason some of these tests use -O2 and some use -O1?
> new file mode 100644
> index 0000000..729b0cc
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/powerpc/fold-vec-splat-longlong.c
> @@ -0,0 +1,62 @@
> +/* Verify that overloaded built-ins for vec_splat with long long
> + inputs produce the right code. */
> +
> +/* { dg-do compile } */
> +/* { dg-require-effective-target powerpc_p8vector_ok } */
> +/* { dg-options "-mvsx -O2" } */
What is that p8vector_ok for? You don't force p8 vector support on.
I think you can remove this.
Okay for trunk with that last one resolved. Thanks!
Segher
