FracNEON V0.2 and FracVFP V0.2 by Michael Kbel
------------------------------------------------
These applications demonstrate the use of the VFP and the NEON-Unit
of the Cortex-A8 CPU of the BeagleBoard.

FracVFP calculates a Mandelbrot fractal two times, one time with single
precision, one time with double precision floats and displays the results
(Time taken/Iterations done/Iterations per second).

FracNEON calculates the same Mandelbrot fractal with single precision,
but due to the possibilities of the NEON SIMD instructions, 4 pixels at
one time are calculated. Also it can be seen that the NEON unit itself
is much faster for single precision compared to the VFP, otherwise the
speed gain wouldn't be that high.

Both applications use the current desktop screen mode. It should be at
a minimum of 800x600x16MCol. I tested it at 800x600/1024x768/1280x1024
successfully.

The source code can be assembled with the latest version of !extASM by
Terje Slettebo. Both apps use the frame of his !CubeDemo. So very special
thanks to him !

If you got any questions,comments or problems you can contact me at ->

E-mail: michael.kuebel@googlemail.com
Web...: http://www.mikusite.de (Check RiscOS section for latest versions)

History:
--------
(27/Jan/2011)
FracNEON V0.2 - Speed up of about 25%. Now all the logic is implemented to
                replace instantly diverged or maximum iterated pixels as
                the event occurs.
FracVFP  V0.2 - No changes

(14/Jan/2011)
FracNEON V0.1 - First release
FracVFP  V0.2 - Small speed up (+5%) by using 4 times loop unrolling
                instead of 2 times. Some code clean up and colour adjustment
(05/Jan/2011)
FracVFP  V0.1 - First release







