[petsc-dev] OpenMP/Vec

[Lawrence Mitchell]

On 28/02/2012 20:18, Jed Brown wrote:

...

>     I prefer inlining the loop bounds into macro/function calls at point of
>     use, rather than relying on magic names from a macro definition, but I'm
>     not completely wedded to the idea.
> 
> 
> Not a big deal, we could optimize either variant.

Moving forward, I've done a bunch of this macroisation, which makes the code
more PETSc-like.

As you point out it's probably unwise to call omp_get_num_threads() on every
loop iteration, so I went for the magic names in the end.  A typical simple
loop now looks like:

VecOMPParallelBegin(xin, private(i) shared(x));
for (i=__start; i<__end;i++) x[i] = work(x[i]);
VecOMPParallelEnd();

In the --without-openmp case, this transforms into:

do {
   PetscInt __start = 0, __end = xin->map->n;
   for ...;
} while (0);

Rather than adding an nthreads slot to vectors, I put it in the generic
PetscObject struct -- we're threading matrices too.

I don't know how to write the correct tests for configure to probe for the
existance of _Pragma, so that's still missing.

Cheers,

Lawrence

-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.