🚀 go-pugleaf

In article <MOSS.92Jul25222337@ibis.cs.umass.edu> moss@cs.umass.edu writes:
>The "find first set bit" operation was recently discussed at length in
>comp.arch, and could be relevant, especially if the one bits are sparse.  If
  An interesting application is to locate point sources on a CCD scaned image.
In a virtual reality case here the sources of IR light are 4 LED emitters
mounted on the back of the helmet that holds 2 cameras.  The LEDs are not in
one plane so location of full spatial postion is possible.  By using a run
length encoding the 300 * 400 CCD map can be "compressed" and processed so
that when the data is read by the program most of the background 0's have
been compressed into a # representing the count.
  The result is that in real time (60 Hz) the computer (with a very limited
input bandwidth) can take in the compressed image, find the sources, refine
their location to about 1/3000 (center extraction) and do it in real time.
This is the only real world application I know of where data compression 
actually aided in processing of the data.
  And getting back around to square one.  All that this thing is, is a 
glorified set bit searcher  :-)
  As for S/W solutions, the best would be an array (assuming its a smallish
8 or maby 10 bit #) but that can vary depending upon architecture.  If its
a bigger number then just divide it into byte sized pieces and work your way
up thru the bytes.  This may or may not be faster than bit shift/test carry.
  - Eric