> Date: Fri, 27 Aug 2004 02:02:35 -0400 > From: Results <[EMAIL PROTECTED]> > > Virtually every system in the world is designed around the > idea that CPUs do integer math faster (much faster) than > floating point. Despite the increases in chip speed, a > calculate-in-integer-set-decimal-on-display approach is > still advantageous. Beyond that, F-Correlatives and > I-Descriptors work better with integers, so ad-hoc reporting > is easier.
Good summary, Chuck! I would add that it is the best way to accurately store numbers, especially financial data, if you are concerned about precision. <number theory lesson for those interested> If you recall, the "decimal" portion of a number can be represented as a sum of negative powers of 10 (e.g., .58 = (5 * 10^-1) + (8 * 10^-2)). In a binary storage system, this must be converted to negative powers of 2. A close approximate value using 10 bits is 2^-1 + 2^-4 + 2^-6 + 2^-9 = (0.5 + 0.0625 + 0.015625 + 0.001953125) = 0.58078125 Even if I increase the binary precision, I would still have that trailing 25 at the far end of my representation. While this might not seem significant, when you total hundreds, thousands or millions of transactions, you will eventually be off in a significant digit. Accountants don't like that... </number theory lesson for those interested> --Tom Pellitieri Century Equipment ------- u2-users mailing list [EMAIL PROTECTED] To unsubscribe please visit http://listserver.u2ug.org/
