Jan 19th 2013, 6:51:44
What exactly is an absolute number?
All numbers in computers are represented either as an integer (as above), or as a floating point value.
A 4-byte floating point value has about 6 significant digits accuracy, with a mantissa. Example: 1.234567+08 is a floating point number which means 1.234567 * 10^8 or the number 123456700. The "+08" is the mantissa, and the range is -128 to 127 (8 bits) and the 1.234567 is the significand (24 bits).
A 8-byte (64-bit) floating point value however has roughly twice the significant digits (15 digits, 52 bits) and a mantissa of 12 bits. Such a number can represent integers accurate up to about 1,000,000,000,000,000 before losing precision. Performance wise, most modern FPUs can do floating point maths as fast as integer maths, but converting a floating point number into an integer number (usually for display purposes) is very costly in CPU time, but is likely negligible for EE.
All numbers in computers are represented either as an integer (as above), or as a floating point value.
A 4-byte floating point value has about 6 significant digits accuracy, with a mantissa. Example: 1.234567+08 is a floating point number which means 1.234567 * 10^8 or the number 123456700. The "+08" is the mantissa, and the range is -128 to 127 (8 bits) and the 1.234567 is the significand (24 bits).
A 8-byte (64-bit) floating point value however has roughly twice the significant digits (15 digits, 52 bits) and a mantissa of 12 bits. Such a number can represent integers accurate up to about 1,000,000,000,000,000 before losing precision. Performance wise, most modern FPUs can do floating point maths as fast as integer maths, but converting a floating point number into an integer number (usually for display purposes) is very costly in CPU time, but is likely negligible for EE.