Parallels for Mac 7 having problem with ICC AVX optimized code

test.c

/*
Now let's try compile and benchmark it:

$ icc -v
icc version 13.0.0 (gcc version 4.6.0 compatibility)

Let's try without AVX set (just up to SSE4.2):

$ icc -march=corei7 test.c -lm && time ./a.out
r=171828183.102

real	0m1.031s
user	0m1.024s
sys	0m0.004s

Now let's try with AVX... ooops exp is 7x slower!

$ icc -march=corei7-avx test.c -lm && time ./a.out
r=171828183.102

real	0m7.019s
user	0m7.012s
sys	0m0.000s

Now let's try with older Intel compiler, so we are sure it isn't new:

$ /opt/intel/composer_xe_2011_sp1/bin/icc -v
icc version 12.1.3 (gcc version 4.6.0 compatibility)

$ /opt/intel/composer_xe_2011_sp1/bin/icc -march=corei7 test.c -lm && time ./a.out
r=171828183.102

real	0m1.050s
user	0m1.048s
sys	0m0.000s

$ /opt/intel/composer_xe_2011_sp1/bin/icc -march=corei7-avx test.c -lm && time ./a.out
r=171828183.102

real	0m7.067s
user	0m7.064s
sys	0m0.000s
*/

#include <math.h>
#include <stdio.h>
int main (int argc, char const *argv[])
{
  double c;
  double r = 1.0;
  for(c = 0.0; c < 1.0; c += 0.00000001) {
    r += exp(c);
  }
  printf("r=%.12g\n", r);
  return 0;
}