The representative performance specifications in the table below are results averaged over many repetitions of each function call. Data is in cache. Results were measured using a 233 MHz Pentium^® II processor and a 512K second level cache. The tests were run under Windows* 95.

Selected Functions	Execution Times Single Precision Floating Point	Execution Times 16 Bit Integer	Units
1024 dim. Add (bAdd2)	2.79	0.85	clock/element
1024 dim. Multiply (bMpy2)	3.03	0.85	clock/element
1024 dim. Minimum (Min)	2.34	0.78	clock/element
1024 dim. Square Root (bSqrt1)	28.07	10.59	clock/element
1024 dim. Dot Product (DotProd)	3.48	0.74	cpMAC
256 Pt Complex FFT* (Fft)	54.42	23.71,43.11,89.37**	microseconds
1024 dim Hamming Window (MinHamming)	24.27	17.52	microseconds
32 tap x 1024 dim. FIR Filter (bFir)	571	140	microseconds
32 tap x 1024 dim. Convolution (Conv)	404.8	388.7	microseconds

* Integer FFT results were obtained using DoFastMMX, DoIntCore, and DoFloatCore, respectively