This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: RFC: SIMD pragma independent of Cilk Plus / OpenMPv4
- From: Torvald Riegel <triegel at redhat dot com>
- To: Tobias Burnus <tobias dot burnus at physik dot fu-berlin dot de>
- Cc: gcc at gcc dot gnu dot org
- Date: Wed, 11 Sep 2013 09:01:27 +0200
- Subject: Re: RFC: SIMD pragma independent of Cilk Plus / OpenMPv4
- Authentication-results: sourceware.org; auth=none
- References: <20130909133737 dot GA1628 at physik dot fu-berlin dot de>
On Mon, 2013-09-09 at 15:37 +0200, Tobias Burnus wrote:
> Dear all,
>
> sometimes it can be useful to annotate loops for better vectorization,
> which is rather independent from parallelization.
>
> For vectorization, GCC has [0]:
> a) Cilk Plus's #pragma simd [1]
> b) OpenMP 4.0's #pragma omp simd [2]
>
> Those require -fcilkplus and -fopenmp, respectively, and activate much
> more. The question is whether it makes sense to provide a means to ask
> the compiler for SIMD vectorization without enabling all the other things
> of Cilk Plus/OpenMP. What's your opinion?
>
> [If one provides it, the question is whether it is always on or not,
> which syntax/semantics it uses [e.g. just the one of Cilk or OpenMP]
> and what to do with conflicting pragmas which can occur in this case.]
SIMD extensions are discussed in both ISO C and C++ study groups (CPLEX
and SG1, respectively). The exact form that this will take is still
under discussion (it won't be pragmas, but there's no consensus yet on
whether and how much language constructs are needed), but I would
believe that C/C++ will eventually get SIMD features. Thus, we should
need it at some time anyway (although the languages might adopt
something language-based for fork/join parallelism too).
Torvald