Re: Performance of Integer Multiplication on PIII

[Kevin Atkinson <kevin at atkinson dot dhs dot org>]

On Sat, 3 Nov 2001, Jan Hubicka wrote:

> > When running these same tests on on Mobile Pentium MMX (using -march=i586)
> > Gcc code does out perform mine.  I do not have anything in between to run
> > these tests on so I would appreciate it if someone with a Pentium Pro and
> > PII (or is that the same thing as a Pentium Pro?) could run them and post
> > the results.
> >
> > So I guess the lesson here is that on PIII integer multiplication is fast
> DId the timming in PIII of imul instruction changed?
> Gcc believes its latency to be 4 cycles for PPro to PIII.  THe core is same,
> but there may be small differences.  Is this the case?

After reading the optimization guide from http://www.agner.org/assem/ I
discovered that the design of the PIII is very similar to PPro, and the
latency is still the same.  However the PPro and PIII has a very
complicated pipeline and are as a result are able to execute up to 4
multiplications in parallel which is what is going on here as none of the
multiplications depend on each other in the dependency chain due to
register renaming.

> Honza
> > enough that doing special tricks to avoid integer multiplication will hurt
> > performs in stead of helping it.
> >
> > Is this a known issue?  And if so does anyone plan on addressing it.  I
> > seams to me that the only fix would me to introduce more options for
> > -march and -mcpu for the faster Pentiums.