[Neo.cmg]

The topic says it all...I was wondering if anyone knew of an article describing how one might go about performing double precision mathematical calculations using 32 bit processors. I was under the naive impression that double precision calculations required 64 bit processors.

Thanks,

Neo.cmg

[alien]

When speaking of 32-bit computers, that usually means the native address size of the computer, or the length of an integer/general purpose register. IOW, 32 bit integer and load/store units.

But all modern processors have separate floating point unit(s), which may well be 64 bits, even if the rest of the processor isn't. This goes for the 32 bit G4 and many other processors.

[BLAZE_MkIV]

If i remember from back in my college days float are storred as a number and an exponent. a double precision float just has a more bits devoted to the exponent. I may be wrong its been quite a while.

[Richard Edgar]

If i remember from back in my college days float are storred as a number and an exponent. a double precision float just has a more bits devoted to the exponent. I may be wrong its been quite a while

Well, given that increasing the exponent size wouldn't increase the available precision.....

Possessing a hardware FPU has nothing to do with it (although alien's comment is correct for most current processors). One could do quad precision FP arithmetic on an 8-bit CPU with no floating point hardware - it would just take a while. A register machine is equivalent to a Turing machine, and hence modern CPUs are all capable of performing the same operations. It's just that some software trickery might be necessary.