Re: [OT] GCC vs Intel C++ compiler benchmark

[Tim Prince <tprinceusa at mindspring dot com>]

On Sunday 27 January 2002 04:05, Andreas Jaeger wrote:
> Claus Fischer <claus.fischer@clausfischer.com> writes:
> > The rather biased article on http://www.open-mag.com/754088105111.htm
> > compares Intel's C++ favourably to GCC, without disclosing many
> > enlightening details.
> >
> > I'm not concerned with that but with the notable OS difference
> > between Linux (SuSE 7.3) and Windows (XP Pro). IMHO a CPU bound
> > benchmark should see less than 1 % influence from OS and C library.
> > The graphics shows roughly 7% better performance on Windows for the
> > same (Intel) compiler.
>
> I've done some measurements myself on Linux and the Intel compiler on
> Linux is really superior.
I think you mean superior to gcc on certain benchmarks, when the best options 
are selected.
>
> > Are there any ABI differences which would justify such a difference?
>
> The Intel compiler uses different calling conventions for local
> functions, can handle whole program optimizations, ...  I don't have a
> full list of optimizations that the Intel compiler handles better but
> it would be interesting to know what kind of optimizations really help
> most for which CPUs.
The whole program optimizations remain problematical for large projects.
The peer review system of gcc has performed well in sorting out which 
optimizations are generally applicable and reliable.
>
> GCC 3.1 is in some areas already superior to GCC 3.0 and I expect
> further improvements for the future with the ongoing projects like the
> AST-Optimizer Branch, the new register allocator, midlevel RTL and the
> work on the cfg-branch.
As I understand it, Honza and Andreas have improved gcc-3.1 by 5% over 
gcc-3.0 on 6 of the SPEC CPU2000 benchmarks.  This is HUGE, particularly 
since the improvements they made apply generally, not just to these specific 
benchmarks, and not simply to the specific CPU model they tested.
>