Re: [Gomp-discuss] Plan ... coments wanted !

[Diego Novillo]

On Thu, 30 Jan 2003, Steven Bosscher wrote:

> > They probably don't have concurrency-aware optimizers.  We
> > should.
> 
> Isn't that maybe a bit too much of a good thing?  How much benefit do
> you expect from that, compared to the amount of extra work involved?
> 
You don't *have* to do it.  The state of the art in optimizing
explicitly parallel programs is very primitive.  You do have to
leave yourself open for enhanchements, though.

> I looked at the OdinMP (http://vvv.it.kth.se/labs/cs/odinmp/) results. 
> That is a "C to C-with-pthreads" compiler that takes OpenMP C source
> code and parallelizes it.  It's a really simple compiler, no
> optimizations at all.  So it's really just a preprocessor.
> 
> And the results look very good!
> 
Of course they do.  If you know what you are doing, no compiler
will beat hand crafted parallel code.  Particularly if the
algorithm is embarrassingly parallel.

> Such results would suggest that parallelizing *before* we lower GENERIC
> to GIMPLE might give us good results too.
> 
Parallelizing?  We are not going to be parallelizing anything.
We are only going to be implementing the OpenMP #pragmas.

What we need to do is fairly straightforward.  Take the OpenMP
directives, create new tree nodes to represent them in GENERIC or
GIMPLE and write the transformation to libgomp.


Diego.