CMPE12cGabriel Hugh Elkaim 1 What do floating-point numbers represent? Rational numbers with non-repeating expansions in the given base within the specified exponent range. They do not represent repeating rational or irrational numbers, or any number too small or too large. Floating Point Format

CMPE12cGabriel Hugh Elkaim 2 IEEE Double Precision FP IEEE Double Precision is similar to SP –52-bit M 53 bits of precision with hidden bit –11-bit E, excess 1023, representing – –One sign bit Always use DP unless memory/file size is important –SP ~ … –DP ~ … Be very careful of these ranges in numeric computation

CMPE12cGabriel Hugh Elkaim 3 Floating Point Arithmetic Floating Point operations include Addition Subtraction Multiplication Division They are complicated because…

CMPE12cGabriel Hugh Elkaim 4 Floating Point Addition 1.Align decimal points 2.Add 3.Normalize the result Often already normalized Otherwise move one digit x Round result x x x x x x 10 2 Decimal Review How do we do this?

CMPE12cGabriel Hugh Elkaim 5 Floating Point Addition First step: get into SP FP if not already.25 = = Or with hidden bit.25 = = Example: in SP FP Hidden Bit

CMPE12cGabriel Hugh Elkaim 6 Second step: Align radix points –Shifting F left by 1 bit, decreasing e by 1 –Shifting F right by 1 bit, increasing e by 1 –Shift F right so least significant bits fall off –Which of the two numbers should we shift? Floating Point Addition

CMPE12cGabriel Hugh Elkaim 7 Floating Point Addition Shift the.25 to increase its exponent so it matches that of ’s e: – (127) = 100’s e: – (127) = Shift.25 by 8 then. Easier method: Bias cancels with subtraction, so Second step: Align radix points cont ’s E 0.25’s E

CMPE12cGabriel Hugh Elkaim 8 Carefully shifting the 0.25’s fraction S E HB F (original value) (shifted by 1) (shifted by 2) (shifted by 3) (shifted by 4) (shifted by 5) (shifted by 6) (shifted by 7) (shifted by 8) Floating Point Addition

CMPE12cGabriel Hugh Elkaim 9 Floating Point Addition Third Step: Add fractions with hidden bit (100) (.25) Fourth Step: Normalize the result Get a ‘1’ back in hidden bit Already normalized most of the time Remove hidden bit and finished

CMPE12cGabriel Hugh Elkaim 10 Normalization example S E HB F Need to shift so that only a 1 in HB spot  discarded Floating Point Addition

CMPE12cGabriel Hugh Elkaim 11 Floating Point Example 0xD4F x56B00000

CMPE12cGabriel Hugh Elkaim 12

CMPE12cGabriel Hugh Elkaim 13 Another SP FP Example 0xD5D x

CMPE12cGabriel Hugh Elkaim 14

CMPE12cGabriel Hugh Elkaim 15 Floating Point Subtraction Mantissa’s are sign-magnitude Watch out when the numbers are close x x 10 2 A many-digit normalization is possible This is why FP addition is in many ways more difficult than FP multiplication

CMPE12cGabriel Hugh Elkaim 16 Floating Point Subtraction 1.Align radix points 2.Perform sign-magnitude operand swap if needed Compare magnitudes (with hidden bit) Change sign bit if order of operands is changed. 3.Subtract 4.Normalize 5.Round Steps to do subtraction

CMPE12cGabriel Hugh Elkaim 17 S E HB F smaller bigger switch order and make result negative bigger smaller switched sign Floating Point Subtraction Simple Example:

CMPE12cGabriel Hugh Elkaim 18 Floating Point Multiplication 1.Multiply mantissas 3.0 x Add exponents = 3 3. Combine x Normalize if needed 1.50 x 10 4 Decimal example: 3.0 x 10 1 x5.0 x 10 2 How do we do this?

CMPE12cGabriel Hugh Elkaim 19 Floating Point Multiplication Multiplication in binary (4-bit F) x Step 1: Multiply mantissas (put hidden bit back first!!) x

CMPE12cGabriel Hugh Elkaim 20 Floating Point Multiplication Second step: Add exponents, subtract extra bias Third step: Renormalize, correcting exponent Becomes Fourth step: Drop the hidden bit (127)

CMPE12cGabriel Hugh Elkaim 21 Multiply these SP FP numbers together 0x49FC0000 x0x4BE00000 Floating Point Multiplication

CMPE12cGabriel Hugh Elkaim 22

CMPE12cGabriel Hugh Elkaim 23

CMPE12cGabriel Hugh Elkaim 24 Another SP FP Example 0xC9F4 × 0x484F

CMPE12cGabriel Hugh Elkaim 25

CMPE12cGabriel Hugh Elkaim 26 Floating Point Division True division Unsigned, full-precision division on mantissas This is much more costly (e.g. 4x) than mult. Subtract exponents Faster division Newton’s method to find reciprocal Multiply dividend by reciprocal of divisor May not yield exact result without some work Similar speed as multiplication

CMPE12cGabriel Hugh Elkaim 27 Questions?