This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Normalizing the bitmap APIs.

On 10/12/12, Richard Biener <> wrote:
> On Oct 12, 2012 Diego Novillo <> wrote:
> > On 2012-10-12 04:26 , Richard Biener wrote:
> > > What's the issue with always returning the changed status?
> > > bitmap operations (even more so sbitmap operations) are
> > > memory-bound, accumulating one more register isn't speed
> > > critial.
> >
> > Not a big issue, but it was going to be a behaviour change,
> > which we are trying to minimize.
> >
> > I have no problem in always returning a status change, if you
> > are OK with that.
> I am ok with that.

There is some rationale for being concerned about performance,
as the checking routines need to read memory locations that they
otherwise would only write.  As a consequence, we have reason to
believe that there would be both an new latency waiting on the
the additional read and increased pressure on the read bandwidth.

I haven't done any performance measures, but I take as evidence
that someone thought there was enough of a performance difference
to make it worth the extra work of writing a second set of routines.

I'm okay with always returning the status change, but we might get
a significant performance change, and I don't want the blame if it
shows up.  :-)

Lawrence Crowl

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]