This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] un-ssa - part 1
On Mon, 2003-03-31 at 23:31, Diego Novillo wrote:
> On Mon, 31 Mar 2003, Andrew Macleod wrote:
>
> > This patch takes a first crack at the un-ssa pass. It implements a basic
> > partitioner which doesn't do anything different than we have today (ie, it
> > puts all versions of a variable into the same partition, so we still
> > aren't taking care of overlapping live ranges.).
> >
> Great! However, this version imposes a rather severe performance
> regression. Compile times have gone through the roof. These are
> the timings for compiling all the files in the C front end
> without the SSA->normal patch:
>
[...]
> >
> Ouch. A 1337% performance degradation. And it still doesn't do
> the partitioning. On the bright side, there's plenty of room for
> improvement ;)
>
> I'm wondering if it's not something trivial. It's late, so I
> still haven't looked at the patch real close. I'll check it out
> tomorrow.
Im sure it is. I have done no performance yet, and calling the algorithm
all the time is likely overkill. I just havent had time to go look. Its
certainly easy enough to turn off until I do have time.
The partition builder is a hog at the moment since I just brute force it
through evey use and def, and it makes a full pass through the IL just
like the coalescer will... It probably wont get much slower even when it
does build a conflict graph and resolve it.
In any case, I'll change it so that none of it is called for now. I want
to get this code out of my directory. I can disable in the compiler for
the moment.
Andrew