Generate"scop" compatible file

[Alexandre Pilkiewicz]

Dear List.

I'm currently trying to use Cloog for polyhedral optimization, but I
would like to check it's output. For that, I need an output that is
"scop-compatible", but this is not the case with cloog, due to
intensive usage of modulo and divisions. Is it possible to ask cloog,
with a clever set of options, to produce this kind of code ?

Thanks in advance!

Alexandre Pilkiewicz

[Sven Verdoolaege]

On Wed, Oct 20, 2010 at 10:45:35AM +0200, Alexandre Pilkiewicz wrote:
> Dear List.
>
> I'm currently trying to use Cloog for polyhedral optimization, but I
> would like to check it's output.

What exactly would you like to check?

> For that, I need an output that is
> "scop-compatible", but this is not the case with cloog, due to
> intensive usage of modulo and divisions.

I suppose that depends on your definition of "scop". If you allow
quasi-affine constraints, then it shouldn't be a problem.
Also, I don't think CLooG introduces and modulos if they aren't
already present in the input. For floord and ceild, you
can just multiply the expressions by the denominator. That is,

i <= floord(N,2)

is the same as

2 * i <= N

> Is it possible to ask cloog,
> with a clever set of options, to produce this kind of code ?

No. For upper bounds of a loop, we could in principle just
write out something like "2 * i <= N", but for the lower bound
of a loop, you need to assign a particular value to the iterator.

skimo