Floating point accuracy and large game worlds
Author
I''m looking for a good way to handle LARGE areas of space for my game, but I''m worried about floating point accuracy.
I was thinking of using a pair of long integers together to represent distance along a single axis, effectively dividing space up into cubes, each representing the range of a single long integer.
Is this a decent way to approach this or is there a simpler more effective way?
Will I incurr some kind of performance penalty passing integers to D3D''s matrix structures that are designed for float values?
1) int -> float... not sure bout the performance loss. but do it the other way around and u''re definitely talking bout performance issues. how bad? get a profiler 
2) if floats dont cut it, doubles are still viable. if precision issues still crop up, the long data type may still compensate. however, using long also limits u to discrete steps in coordinate specifications. as long as ur world space does not span too much, it shouldnt be a problem. but a simpler solution if that occurs.. use MORE longs
2) if floats dont cut it, doubles are still viable. if precision issues still crop up, the long data type may still compensate. however, using long also limits u to discrete steps in coordinate specifications. as long as ur world space does not span too much, it shouldnt be a problem. but a simpler solution if that occurs.. use MORE longs
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
