[rescue] Perverse Question
Francisco Javier Mesa-Martinez
lefa at ucsc.edu
Sun Jun 15 06:21:02 CDT 2003
> Actually, do you have benchmarks of that? In 3 years of using linux  I've
> become sceptical about every sort of fud (no offense here).
RTFM, you have been using linux on a X86 machine I suppose. GCC is mostly
X86 optimized, RISC versions of GCC are notoriously bad when compared to
vendor specific offerings. Then again, this is just another proof that you
get what you pay for.
On the average I have seen mips tools (Which are the ones I use most)
producing code that is about 20% smaller in footprint (which is
significant for a RISC machine, instruction bandwith is a major
bottleneck). And the autoparallelization framework that the compiler has
is just fantastic. Most SGIs I have dealt with are large SMP boxes, so
this is fundamentals. Still some people still use gcc+mpi on these boxes
too, but it is just because they do not want to spend the extra time to
learn the tools (specially since the makefiles, compiler flags, etc, will
be non portable), which I find really annoying. One should take advantage
of the underlying architecture, so you should learn the damned specific
tools :). BUt then again, most people assume that their code will
eventually be migrated to a large linux cluster so they are already
building their code from the ground up to be using message passign, even
though it is going to be run on a large SMP-ish machine (Cc-NUMA is not
really SMP but close enough).
Under sun for example, we have run some interesting benchmarks. For
example a graph coloring algorithm runs well over 40% faster with the
sun tools than with gcc 2.8.ish (this was over a year ago or so, so the gcc
was not current... but neither were the sun workshop compilers). This
meant that our ultra 60 with a 400 (maybe 450) USII was actually keeping
up with a 1GHz Athlon linux box running the same code!!! So I saw that as
a fantastic achievement... of course this may be application specific
Thanks for the insult BTW....
> (Not that it would astonish me much, after all who knows more about Sun cpu's
> then Sun?)
Also the ratio of x86 gcc developers to other architectures is like 100:1,
so of course the x86 will have better cisc intel support.
To give credit where credit is due, it is a fantastic achievement to have
a free compiler set that supports so many architectures. So even if it
ain't optimal, I salute them for offering tools for free to the masses :).
Also I asusme that the tradeoffs between designing a portable framework
and trying to also optimize it must be a complete nightmare...
More information about the rescue