This are the results of EuroBen V5.0 on a Cray X2 (1 core) with the
following testing circumstances:
--------------------------------------------------------------------------
 Computer                  Cray X2, 1.6 GHz, 4 proc. units/node
 Memory size               30 GB/node -- 7.5 GB/proc. unit
 Compiler version          Cray Fortran V6.0.0.5
 Compiler options          -O3
 Operating System version  Unicos/lc V2.0.62
 Working precision          64-bits precision                           
--------------------------------------------------------------------------
Remarks:
1) Program 'mod1d': Timing routine was not good enough for the bank read
   operations --> results are nonsense.
2) Program 'mod2a': Timing was somewhat unstable. Pick results that are
   best.
3) Program 'mod2as': Persistent floating-point exception for vector length
   N > 200.
4) Programs 'mod2b' and 'mod2d' were also run with the BLAS/LAPACK routines
   from 'libsci'. In both cases the results were much faster.
5) Program 'mod2cr': A floating-point exception CAN occur. But when run several
   times the program terminates normally most of the time. Seems the runtime
   is not stable.
6) For programs 'mod2f' and 'mod2g' cache effect are noticeable for N > 10^18.
