RFC: Enable vectorization of call instructions in the loop vectorizer

[Hal Finkel]

----- Original Message -----
> From: "Renato Golin" <renato.golin at linaro.org>
> To: "Arnold Schwaighofer" <aschwaighofer at apple.com>
> Cc: "Hal Finkel" <hfinkel at anl.gov>, "llvm-commits" <llvm-commits at cs.uiuc.edu>
> Sent: Monday, December 16, 2013 3:48:30 PM
> Subject: Re: RFC: Enable vectorization of call instructions in the loop vectorizer
> 
> 
> 
> 
> On 16 December 2013 21:08, Arnold Schwaighofer <
> aschwaighofer at apple.com > wrote:
> 
> 
> 
> 
> Yes and we can easily model this in the cost model by asking what is
> the cost of a (library) function call (vectorized or not) and have
> this return a reasonably high value.
> 
> 
> I'm wondering, in the case Nadav said about #pragma vectorize, how
> are we going to tell the compiler what is the cost of this or that
> vector-variant of foo?
> 
> 
> When the cost tables can change independently of user-code, this
> looks like a tom&jerry exercise.

I think we're okay here for the time being ;) -- Arnold suggested that we use some large value to represent the cost of the scalar function. The metadata (or whatever) just need to specify some multiplicative factor by which we scale that cost. In general, I think that should be fairly easy to determine.

 -Hal

> 
> 
> cheers,
> --renato

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory