Re: [PATCH, rs6000 2/2] Add compatible implementations of x86 SSSE3 intrinsics <tmmintrin.h>

[Segher Boessenkool <segher@kernel.crashing.org>]

On Tue, Oct 23, 2018 at 10:29:35AM -0500, Paul Clarke wrote:
> On 10/22/2018 06:38 PM, Segher Boessenkool wrote:
> > On Mon, Oct 22, 2018 at 01:26:11PM -0500, Paul Clarke wrote:
> >> Target tests for the intrinsics defined in pmmintrin.h, copied from
> >> gcc.target/i386.
> >>
> >> Tested on POWER8 ppc64le and ppc64 (-m64 and -m32, the latter only reporting
> >> 16 new unsupported tests), and also by forcing -mcpu=power7 on ppc64.
> > 
> > Why are they unsupported?  lp64?  Why do many of those tests require
> > lp64 anyway?  It's not obvious to me.
> 
> None of the x86 intrinsics compatibility implementation code has thus far supported -m32, and I'd venture that it's not interesting to anyone.  Or at least not worth the effort.

You can use it with -m32 just fine.  Disabling the tests isn't a good idea.
Either disable the *feature*, or enable the tests.

> > You tested on a >=p8 (with a compiler defaulting to that, too) I hope ;-)
> 
> As stated, "Tested on POWER8 ppc64le" (which defaults to -mcpu=power8), or did you mean something else?

The big-endian part.  It does not default to p8 unless you arrange for
that specially.  The tests require p8 hardware, but that means if you
tested BE without --with-cpu=power8 or similar you really didn't test
anything.


Segher