Re: [Algorithms] Matching Cross platform FP
Brought to you by:
vexxed72
From: Tony C. <to...@mi...> - 2006-07-18 20:49:21
|
That would be called "fixed-point arithmetic", no? Exactly as suggested (and ironically what old school games used to do back in the days when we didn't use floating-point because the performance sucked.)=20 - Tony -----Original Message----- From: gda...@li... [mailto:gda...@li...] On Behalf Of Don Gunter Sent: Tuesday, July 18, 2006 3:12 PM To: 'Game Development Algorithms' Subject: Re: [Algorithms] Matching Cross platform FP I see what ya'll meant about Game Engineers usually being at odds or frustrated with the hardware folks<grin>. NO WONDER. We're still killing ya'll with that floating point arithmetic. =20 Are there any options you have available that split the float into a fixed point integer and a fixed point fraction/real component? I.E high word the integer and the low word the fraction? It limits you to a fixed resolution. and calcs would probably be expensive for what ya'll do right? You could shift the real part left to bring the real component into integer space... sorry. Blabbing. Probably only work when converting the dataset from one platform to the other because of the time complexity. =20 |