FPE benchmarks

[glyn@cs.UAlberta.CA (Reade Glyn Devin)]

Ok folks, here are three of the Orca/C floating point benchmarks done under
four different conditions.  Where the FPE was used, it was done using the
Orca FPE sysfloat library.

Note that I have the FPE in slot 3 without overriding slot 3's functions
(the control panel is set to the default).

Presumably, the FPU would have similar performance to the FPE.

I hope this helps.

Test		SANE		SANE & Zip 9/32	    FPE		FPE & Zip 9/32
===============================================================================

Float (1000 iterations)
	Time: 	50 sec		20 sec		8 sec		5 sec
	C:	1.783903e+04	1.783903e+04	1.765462e+04	1.765462e+04

Gamm (15000 units)
	Time:	33 sec		14 sec		8 sec		5 sec
	n:	50		50		50		50
	acc:	1.3448051e+01	1.3448051e+01	1.3448051e+01	1.3448051e+01
	acc1:	2.6896116e-01	2.6896116e-01	2.6896116e-01	2.6896116e-01

Savage (250 iterations)
	Time:	73 sec		30 sec		5 sec		1 sec
	Sum:	2.500000e+02	2.500000e+02	3.200000e+01	3.200000e+01



--
        G. Devin Reade          glyn@cs.ualberta.ca