Fixed Point Math

[R0B]

[RwGast]

[R0B]

[Dan East]

Those routines would have been used for a global fixed point replacement, if it were feasible. They are not used currently. <br><br>All the fixed point conversions that are in use now are hand-coded, and were meticulously written to work exactly with the precision / range encountered in that routine. They are integrated directly in the source in place of the original code.<br><br>The current FixedPointMath source is very inefficient. The only criteria during the conversion phase was that the fixed point code be accurate. When I discovered that a fixed point number of a predefined, static precision could not store all the various floating point values Quake encounters, I dropped that part of the project. What is needed is FixedPointMath code that dynamically adjusts the precision based on the value that needs to be stored. Of course then you also need to store what the precision is, in addition to the value. So each fixed point "number" would have to be a handle or pointer to a structure that actually contains the data.<br><br>Dan East

[R0B]

I'm not planing on undertaking this anytime soon, myself, but just out of curiosity, for dynamic percision, does that contantly change, or is there an exact percision for each var, and it's just a matter of finding out what that is?