| ??? 02/14/01 09:10 Read: times |
#9292 - RE: use of multiply in large data arrays |
I'm writing a column for May '01 about using binary trees to quickly access tables where you are mapping an inifinite set of inputs to a finite set of outputs. Its part 3 of a series on table-lookup method. I use assembly language only.
An example would be where I compute a floating point number and need to compute a final sinusoidal function before displaying it in decimal form. Using tables, I can pre-compute the sinusoidal function and the decimal base conversion as a direct table-lookup mapping. Neither calculation therefore has to be done by the microcontroller. :) It uses the binary tree structure with implied addressing instead of pointers. The tree holds the set of maximum values for each display class. Traversing the tree quickly makes the high/low decisions converging upon its associated answer. At the last step of the tree traversal, the pointer offset into the tree is the same as the pointer offset into the final answer table. You can see why I need until May to make the topic understandable. :) aka j http://www.geocities.com/mingpzong |



