1. Chapter 8 Computer Arithmetic

2. 8.1 Unsigned Notation Non-negative notation  It treats every number as either zero or a positive value  Range: 0 to 2 n -1 Unsigned Two ’ s Complement  Negative numbers have a 1 as the most significant bit and positive number (or zero) have 0 as the leading bit.  Range: -2 n-1 -1 to 2 n-1 -1 Refer to Table 8.1

3. 8.1 Unsigned Notation(continued) Addition and Subtraction(Figure 8.1)

4. Figure 8.1

5. 8.1 Unsigned Notation(continued) Arithmetic overflow in addition  In non-negative notation, carry bit can set an overflow flag.  In two ’ s complement notation an overflow can occur at either end of the numeric range. (Figure 8.2)

6. Figure 8.2

7. 8.1 Unsigned Notation(continued) Arithmetic overflow in subtraction  X-Y is implemented as X+(-Y)  Y is converted to -Y by taking it ’ s two ’ s complement.  Implementation of micro-operation SUB: X  X-Y (Figure 8.3)

8. Figure 8.3

9. 8.1 Unsigned Notation(continued) Overflow generation in unsigned two ’ s complement subtraction (Fig. 8.4)  The result should be larger than or equal to zero.

10. Figure 8.4

11. 8.1 Unsigned Notation(continued) Multiplication  A simple algorithm for x  y Z =0 For I=1 to y do { z = z+x }  Shift-add multiplication  Partial products

12. 8.1 Unsigned Notation(continued) Multiplication  another simple algorithm for x  y  The result in two n-bits registers, V.U U =0 For I = 1 to n do { IF Y0= 1 then CU = U +X; linear shift right CUV; circular shift right Y }  Refer to table 8.2  RTL code for realize the algorithm  Refer to Table 8.3  Hardware implementation (Refer to Figure 8.5)

13. Figure 8.5

14. 8.1 Unsigned Notation(continued) Multiplication  Booth ’ s algorithm (Figure 8.6)  This algorithm works directly on two ’ s complete numbers.  UV  X*Y U =0; Y -1 = 0; For I=1 To n DO {IF start of a string of 1 ’ s in Y THEN U = U-X; IF end of a string of 1 ’ s in Y THEN U =U+X; Arithmetic shift right UV; Circular shift right Y AND copy Y 0 to Y -1 }

15. Figure 8.6

16. Division Z = X  Y  Z = 0; WHILE x  y DO { z = z + 1, x = x-y }

17. Division(continued) Shift-subtract division of two binary values.  Dividend: UV, divisor: X, quotient: Y, remainder: U  IF U  X THEN exit with overflow; Y = 0; C = 0; FOR i = 1 TO n DO { linear shift left CUV; linear shift left Y; IF CU  X THEN {Y 0 = 1, U = CU -X } }

18. Division(continued) Non-restoring division algorithm Re-storing division algorithm

19. Figure 8.7

20. How to compare U and X Figure 8.8

22. Figure 8.9 Hardware implementa tion of restoring algorithm

23. 8.2 Signed Notations Signed-magnitude Signed-two ’ s complement  To add and subtract two signed-two ’ s complementation values, we simply treat the sign bit as the most significant bit of magnitude.  Multiplication can be accomplished by using Booth ’ s algorithm

24. 8.3 Binary Coded Decimal BCD Format: 4 bits can represent one decimal digit. Addition and subtraction (Figure 8.13)  We adjust the hardware that adds numbers to account for the BCD representation.  When the sum of two digits is more than 9, adding 6 to the result generated by a binary adders produces the correct result.(Figure 8.11)  Nine ’ s complement (Figure 8.12) or ten ’ s complement can be used for subtraction.

25. Figure 8.11

26. Figure 8.12

27. Figure 8.13

28. 8.4 Special Arithmetic Hardware Pipelining  Arithmetic pipeline (Figure 8.14)  To increase the throughput, the numbers of results generated per time unit.  Speed up: a metric used to measure the performance of a pipeline.

29. 8.4 Special Arithmetic Hardware(continued) Pipelining The maximum speedup Lookup table (Figure 8.15 and 8.16)

30. Figure 8.14

31. Figure 8.15

32. Figure 8.16

33. 8.4 Special Arithmetic Hardware(continued) Carry-save adder(Figure 8.17) Wallice Tree: a combinatorial circuit used to multiply two numbers.(Figure 8.18)

34. Figure 8.17

35. Figure 8.18

36. Example X * Y X = 111 Y = 110 ----- 000  PP0 111  PP1 111  PP2 ------- 101010  Final sum calculated

37. Figure 8.19

38. Example X * Y (6-bit) X = 1011 Y = 1110

39. Figure 8.20

40. Figure 8.21 An 8X8 Wallice Tree Multiplier

41. 8.5 Floating Point Numbers Number format  Normalized  NoN Not a number  Biasing Numeric characteristic  Precision  Gap  Range  Rounding  Guard bit