Good News. Intel's ICC 8.0 Beta looks promising, now.

[Paul A. Renard]

Back in February, 2003, I reported that Intel's icc 7.0
compiler was producing code using Pooma constructs that was 2X-4X slower
than KCC.  Since then, the folks at Intel have worked hard, and for
my little test (reproduced at the end of this message), the icc 8.0 Beta
compiler (l_cc_b_8.0.023) is now producing  code slightly faster
(maybe 5-10%) than KCC, and certainly comparable to hand-written
loops.

The only optimization items for compiling were:
            -O3
-DNOPAssert -DNOCTAssert -tpp7 -xW
but the last two are particular to Pentium 4 vectorization, which plays a
very small part in the tests I did, and which probably caused the
"slightly faster", rather than "just about the same
speed".

So, icc 8.0 seems to be a useful choice in compilers (for Linux and
Windows).

-Paul

(What follows is my original posting)

> Given the following:
>
>  const int
> N=128;
>  Array<2,complex<float> > u(N,N);
>  Iota<2>::Iota_t ij(iota(u.domain());
>  Iota<2>::Index_t I(ij.comp(0));
>  Iota<2>::Index_t J(ij.comp(1));
>  Array<1,complex<float> > cx(N), cy(N);
>
>  // Values for u, cx, cy are filled elsewhere.
>
>  // Then the following is called:
>  void compute(){
>   u *= cx(I)*cy(J);    // runs 4X slower with icc
> than KCC
>  }
>
>  When I time this routine, I find that it runs about 4X
> slower when compiled with Intel's icc (Version 7, -O3 -DNOPAssert
> -DNOCTASSERT) than with KCC (version 4.0f, +K3 -DNOPAssert,
> -DNOCTAssert).  As expected, the KCC version runs as fast as
> hand-written loops.