> Most things can be accurately scaled to a 32bit integer without
> distorting human perception (a latitude, for example, is accurate to
> centimeters when expressed in 32bits). The devil is in the details, and
> fixed point calculations are exceedingly sensitive to programming
> errors, and using 64bit arithmetic for intermediate results can
> actually be slower than floating point on the i386 architecture.

Don't forget that the world is moving in two directions: faster
floating-point operation, simply because most of today's CPU-intensive
applications require it, and more stuff offloaded to the graphics
card. This all really makes any fixed-point programming in the FDM
more than questionable.

64bit arithmetic is slower than double even on i386, but on most
cleaner platforms like PowerPC and Alpha, even 32bit arithmetic is.

  Andras

===========================================================================
Major Andras
    e-mail: [EMAIL PROTECTED]
    www:    http://andras.webhop.org/
===========================================================================

_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Reply via email to