General Discussion > Math and Science

A faster Newton's Method Square Root

<< < (2/2)

Xeda112358:
Yup, I see no good reason for BCD floats. There is some usefulness to decimal floats, though-- they have the one advantage of BCD (don't have to be careful where base-10 representation is a must), but are nearly as efficient as binary floats in terms of storage. A 64-bit decimal64 has 16 digits of precision.

As for emulators, I use TIlem, but it does have glitches of it's own. One I found was it sometimes doesn't emulate the carry flag properly with add/adc/sbc on IX/IY, which was hell when testing some math routines :P I believe the problem is the emulator stores them internally as int32 or int64 and doesn't handle 16-bit overflow.

Navigation

[0] Message Index

[*] Previous page