===========================================================================
		"DARWIN"			"ZORBA"
		Pentium II 333 MHz		UltraSPARC-II 295 MHz
		128M RAM, Linux 2.0.32		124M RAM, SunOS 5.6
---------------------------------------------------------------------------

   MFLOPS(1)		101.3090			85.7290
   MFLOPS(2)		 62.0210			61.9243
   MFLOPS(3)		 82.2173			76.8800
   MFLOPS(4)		 99.2164			89.5174

===========================================================================


  FLOPS C Program (Double Precision), V2.0 18 Dec 1992
          Iterations      =  128000000
---------------------------------------------------------------------------
MFLOPS(1) Weightings: FADD 40.4%, FSUB 23.1%, FMUL 26.9%, FDIV 9.6%.      -
          The code for MFLOPS(1) is, essentially, scalar. That is,        -
          vector processor performance will reflect scalar performance    -
          which will lie far below expected vector performance. Also,     -
          the percentage of FDIV's is somewhat 'high' at 9.6%.            -

MFLOPS(2) Weightings: FADD 38.2%, FSUB 9.2%, FMUL 43.4%, FDIV 9.2%.       -
          The code for MFLOPS(2) is fully vectorizable. The percentage    -
          of FDIV's however is somewhat on the 'high' side.               -

MFLOPS(3) Weightings: FADD 42.9%, FSUB 3.4%, FMUL 50.7%, FDIV 3.4%.       -
          The code for MFLOPS(3) is fully vectorizable. The percentage    -
          of FDIV's is however 'moderate' at 3.4%.                        -

MFLOPS(4) Weightings: FADD 42.9%, FSUB 2.2%, FMUL 54.9%, FDIV 0.0%.       -
          The code for MFLOPS(4) is fully vectorizable. The percentage    -
          of FDIV's is however zero.                                      -
---------------------------------------------------------------------------