1 / 74

# ELEC 5200-001/6200-001 Computer Architecture and Design Fall 2009 Computer Arithmetic (Chapter 3)

ELEC 5200-001/6200-001 Computer Architecture and Design Fall 2009 Computer Arithmetic (Chapter 3). Vishwani D. Agrawal James J. Danaher Professor Department of Electrical and Computer Engineering Auburn University, Auburn, AL 36849 http://www.eng.auburn.edu/~vagrawal vagrawal@eng.auburn.edu.

## ELEC 5200-001/6200-001 Computer Architecture and Design Fall 2009 Computer Arithmetic (Chapter 3)

E N D

### Presentation Transcript

1. ELEC 5200-001/6200-001Computer Architecture and DesignFall 2009 Computer Arithmetic (Chapter 3) Vishwani D. Agrawal James J. Danaher Professor Department of Electrical and Computer Engineering Auburn University, Auburn, AL 36849 http://www.eng.auburn.edu/~vagrawal vagrawal@eng.auburn.edu ELEC 5200-001/6200-001 Lecture 13

2. What Goes on Inside ALU? • Machine instr.: add \$t1, \$s1, \$s2 • What it means to computer: 000000 10001 10010 01000 00000 100000 Arithmetic Logic Unit (ALU) Control unit Flags Registers Registers ELEC 5200-001/6200-001 Lecture 13

3. Basic Idea • Hardware can only deal with binary digits, 0 and 1. • Must represent all numbers, integers or floating point, positive or negative, by binary digits, called bits. • Devise electronic circuits to perform arithmetic operations: add, subtract, multiply and divide, on binary numbers. ELEC 5200-001/6200-001 Lecture 13

4. Positive Integers • Decimal system: made of 10 digits, {0,1,2, . . . , 9} 41 = 4×101 + 1×100 255 = 2×102 + 5×101 + 5×100 • Binary system: made of two digits, {0,1} 00101001 = 0×27 + 0×26 + 1×25 + 0×24 +1×23 +0×22 + 0×21 + 1×20 =32 + 8 +1 = 41 11111111 = 255, largest number with 8 binary digits, 28-1 ELEC 5200-001/6200-001 Lecture 13

5. Base or Radix • For decimal system, 10 is called the base or radix. • Decimal 41 is also written as 4110 or 41ten • Base (radix) for binary system is 2. • Thus, 41ten = 1010012 or 101001two • Also, 111ten = 1101111two and 111two = 7ten ELEC 5200-001/6200-001 Lecture 13

6. Signed Integers – What Not to Do • Use fixed length binary representation • Use left-most bit (called most significant bit or MSB) for sign: 0 for positive 1 for negative • Example: +18ten = 00010010two –18ten = 10010010two ELEC 5200-001/6200-001 Lecture 13

7. Why Not Use Sign Bit • Sign and magnitude bits should be differently treated in arithmetic operations. • Addition and subtraction require different logic circuits. • Overflow is difficult to detect. • “Zero” has two representations: + 0ten = 00000000two – 0ten = 10000000two • Signed-integers are not used in modern computers. ELEC 5200-001/6200-001 Lecture 13

8. Integers With Sign – Other Ways • Use fixed-length representation, but no sign bit • 1’s complement: To form a negative number, complement each bit in the given number. • 2’s complement: To form a negative number, start with the given number, subtract one, and then complement each bit, or first complement each bit, and then add 1. • 2’s complement is the preferred representation. ELEC 5200-001/6200-001 Lecture 13

9. 1’s-Complement • To change the sign of a binary integer simply complement (invert) each bit. • Example: 3 = 0011, – 3 = 1100 • n-bit representation: Negation is equivalent to subtraction from 2n – 1 Infinite universe -1 0 3 6 9 12 15 0 3 6 9 12 15 Modulo-16 4-bit universe 0 3 6 -6 -3 -0 0000 0011 0110 1001 1100 1111 ELEC 5200-001/6200-001 Lecture 13

10. 2’s-Complement • Add 1 to 1’s-complement representation. • Some properties: • Only one representation for 0 • Exactly as many positive numbers as negative numbers • Slight asymmetry – there is one negative number with no positive counterpart ELEC 5200-001/6200-001 Lecture 13

11. 2’s-Complement Infinite universe -1 0 3 6 9 12 15 0 3 6 9 12 15 1’s complement 0 3 6 -6 -3 -0 0000 0011 0110 1001 1100 1111 0 3 6 9 12 15 2’s complement 0 3 6 -5 -2 -1 0000 0011 0110 1001 1100 1111 ELEC 5200-001/6200-001 Lecture 13

12. Three Representations Sign-magnitude 000 = +0 001 = +1 010 = +2 011 = +3 100 = - 0 101 = - 1 110 = - 2 111 = - 3 1’s complement 000 = +0 001 = +1 010 = +2 011 = +3 100 = - 3 101 = - 2 110 = - 1 111 = - 0 2’s complement 000 = +0 001 = +1 010 = +2 011 = +3 100 = - 4 101 = - 3 110 = - 2 111 = - 1 (Preferred) ELEC 5200-001/6200-001 Lecture 13

13. 2’s Complement Numbers 000 -1 +1 0 -1 111 001 +1 Positive numbers 010 +2 Negative numbers -2 110 011 +3 -3 101 - 4 100 Overflow Negation ELEC 5200-001/6200-001 Lecture 13

14. 2’s Complement n-bit Numbers • Range: –2n –1 through 2n –1 – 1 • Unique zero: 00000000 . . . . . 0 • Negation rule: see slide 9. • Expansion of bit length: stretch the left-most bit all the way, e.g., 11111101 is still – 3. • Overflow rule: If two numbers with the same sign bit (both positive or both negative) are added, the overflow occurs if and only if the result has the opposite sign. • Subtraction rule: for A – B, add – B to A. ELEC 5200-001/6200-001 Lecture 13

15. Converting 2’s Compliment to Decimal n-2 an-1an-2 . . . a1a0 = -2n-1an-1 + Σ 2i ai i=0 8-bit conversion box -128 64 32 16 8 4 2 1 -128 64 32 16 8 4 2 1 1 1 1 1 1 1 0 1 Example -128+64+32+16+8+4+1 = -128 + 125 = -3 ELEC 5200-001/6200-001 Lecture 13

16. For More on 2’s Complement • Chapter 2 in D. E. Knuth, The Art of Computer Programming: Seminumerical Algorithms, Volume II, Second Edition, Addison-Wesley, 1981. • A. al’Khwarizmi, Hisab al-jabr w’al-muqabala, 830. Abu Abd-Allah ibn Musa al’Khwarizmi (~780 – 850) Donald E. Knuth (1938 - ) ELEC 5200-001/6200-001 Lecture 13

17. MIPS • MIPS architecture uses 32-bit numbers. What is the range of integers (positive and negative) that can be represented? Positive integers: 0 to 2,147,483,647 Negative integers: - 1 to - 2,147,483,648 • What are the binary representations of the extreme positive and negative integers? 0111 1111 1111 1111 1111 1111 1111 1111 = 231 - 1= 2,147,483,647 1000 0000 0000 0000 0000 0000 0000 0000 = - 231 = - 2,147,483,648 • What is the binary representation of zero? 0000 0000 0000 0000 0000 0000 0000 0000 ELEC 5200-001/6200-001 Lecture 13

18. Addition • Adding bits: • 0 + 0 = 0 • 0 + 1 = 1 • 1 + 0 = 1 • 1 + 1 = (1) 0 • Adding integers: carry 1 1 0 0 0 0 . . . . . . 0 1 1 1 two = 7ten + 0 0 0 . . . . . . 0 1 1 0 two = 6ten = 0 0 0 . . . . . . 1 (1)1 (1)0 (0)1 two = 13ten ELEC 5200-001/6200-001 Lecture 13

19. Subtraction • Direct subtraction • Two’s complement subtraction 0 0 0 . . . . . . 0 1 1 1 two = 7ten - 0 0 0 . . . . . . 0 1 1 0 two = 6ten = 0 0 0 . . . . . . 0 0 0 1two = 1ten 1 1 1 . . . . . . 1 1 0 0 0 0 . . . . . . 0 1 1 1 two = 7ten + 1 1 1 . . . . . . 1 0 1 0 two = - 6ten = 0 0 0 . . . . . . 0 (1) 0 (1) 0 (0)1 two = 1ten ELEC 5200-001/6200-001 Lecture 13

20. Overflow: An Error • Examples: Addition of 3-bit integers (range - 4 to +3) • -2-3 = -5110 = -2 + 101 = -3 = 1011 = 3 (error) • 3+2 = 5011 = 3 010 = 2 = 101 = -3 (error) • Overflow rule: If two numbers with the same sign bit (both positive or both negative) are added, the overflow occurs if and only if the result has the opposite sign. 000 111 0 001 1 -1 010 – + 2 110 -2 3 -3 011 - 4 101 100 Overflow ELEC 5200-001/6200-001 Lecture 13

21. Design Hardware Bit by Bit • Adding two bits: a b half_sum carry_out 0 0 0 0 0 1 1 0 1 0 1 0 1 1 0 1 • Half-adder circuit a half_sum XOR b carry_out AND ELEC 5200-001/6200-001 Lecture 13

22. One-bit Full-Adder • One-bit full-adder truth table a b ci half_sum carry_out sum co 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 1 0 1 0 1 0 0 1 1 1 0 0 1 1 0 0 1 0 1 0 1 0 1 1 0 0 1 1 1 0 0 1 0 1 1 1 1 0 1 1 1 ELEC 5200-001/6200-001 Lecture 13

23. One-bit Full-Adder Circuit ci FAi XOR sumi ai XOR AND bi AND OR Ci+1 ELEC 5200-001/6200-001 Lecture 13

24. 32-bit Ripple-Carry Adder c0 a0 b0 sum0 FA0 sum1 a1 b1 FA1 sum2 a2 b2 FA2 sum31 FA31 a31 b31 ELEC 5200-001/6200-001 Lecture 13

25. How Fast is Ripple-Carry Adder? • Longest delay path (critical path) runs from cin to sum31. • Suppose delay of full-adder is 100ps. • Critical path delay = 3,200ps • Clock rate cannot be higher than 1/(3,200×1012) Hz = 312MHz. • Must use more efficient ways to handle carry. ELEC 5200-001/6200-001 Lecture 13

26. Speeding Up the Adder a0-a15 16-bit ripple carry adder sum0-sum15 b0-b15 cin a16-a31 16-bit ripple carry adder 0 b16-b31 0 sum16-sum31 Multiplexer a16-a31 16-bit ripple carry adder 1 b16-b31 This is a carry-select adder 1 ELEC 5200-001/6200-001 Lecture 13

27. Fast Adders • In general, any output of a 32-bit adder can be evaluated as a logic expression in terms of all 65 inputs. • Number of levels of logic can be reduced to log2N for N-bit adder. Ripple-carry has N levels. • More gates are needed, about log2N times that of ripple-carry design. • Fastest design is known as carry lookahead adder. ELEC 5200-001/6200-001 Lecture 13

28. N-bit Adder Design Options Reference: J. L. Hennessy and D. A. Patterson, Computer Architecture: A Quantitative Approach, Second Edition, San Francisco, California, 1990, page A-46. ELEC 5200-001/6200-001 Lecture 13

29. MIPS Instructions (see p. 175) • Arithmetic: add, sub, addi, addu, subu, addiu, mfc0 • Data transfer: lw, sw, lhu, sh, lbu, sb, lui • Logical: and, or, nor, andi, ori, sll, srl • Conditional branch: beq, bne, slt, slti, sltu, sltiu • Unconditional jump: j, jr, jal ELEC 5200-001/6200-001 Lecture 13

30. Exception or Interrupt • If an overflow is detected while executing add, addi or sub, then the address of that instruction is placed in a register called exception program counter (EPC). • Instruction mfc0 can copy \$epc to any other register, e.g., mfc0 \$s1, \$epc • Unsigned operations, addu, addiu and subu do not cause an exception or interrupt. ELEC 5200-001/6200-001 Lecture 13

31. Multifunction ALU operation c0 a0 b0 result0 ALU0 result1 a1 b1 ALU1 result2 operation a2 b2 ALU2 ci ai bi FAi 3 NOR 2 resulti mux OR 1 result31 ALU31 a31 b31 AND 0 ELEC 5200-001/6200-001 Lecture 13

32. Binary Multiplication (Unsigned) 1 0 0 0 two = 8ten multiplicand 1 0 0 1 two = 9ten multiplier ____________ 1 0 0 0 0 0 0 0 partial products 0 0 0 0 1 0 0 0 ____________ 1 0 0 1 0 0 0two = 72ten Basic algorithm: n = 1, 32 If nth bit of multiplier is 1, add multiplicand × 2 n –1 to product ELEC 5200-001/6200-001 Lecture 13

33. Multiplication Flowchart Start Initialize product register to 0 Partial product number, n = 1 LSB of multiplier ? Add multiplicand to product and place result in product register 1 0 Left shift multiplicand register 1 bit Right shift multiplier register 1 bit n = 32 n < 32 i = ? Done n = n + 1 ELEC 5200-001/6200-001 Lecture 13

34. Serial Multiplication shift left shift right Multiplicand (expanded 64-bits) 32-bit multiplier 64 64 shift Test LSB 32 times add 64-bit ALU LSB = 1 LSB = 0 64 3 operations per bit: shift right shift left add Need 64-bit ALU 64-bit product register write ELEC 5200-001/6200-001 Lecture 13

35. Serial Multiplication (Improved) 2 operations per bit: shift right add 32-bit ALU Multiplicand 32 32 add 1 Test LSB 32 times 32-bit ALU LSB 1 32 write 64-bit product register shift right 00000 . . . 00000 32-bit multiplier Initialized product register ELEC 5200-001/6200-001 Lecture 13

36. Example: 0010two× 0011two 0010two× 0011two = 0110two, i.e., 2ten×3ten = 6ten ELEC 5200-001/6200-001 Lecture 13

37. Signed Multiplication • Convert numbers to magnitudes. • Multiply the two magnitudes through 32 iterations. • Negate the result if the signs of the multiplicand and multiplier differed. • Alternatively, the previous algorithm will work with some modifications, listed next. ELEC 5200-001/6200-001 Lecture 13

38. Multiplying 2’s Complement • Use one extra bit for multiplicand addition. • Extend sign bit in right shift (Examples 1 and 2 in next slides). • If multiplier is negative, then the last addition is replaced by subtraction (Example 2). Why? See slide 15. • See B. Parhami, Computer Architecture, New York: Oxford University Press, 2005, pp. 199-200. ELEC 5200-001/6200-001 Lecture 13

39. Example 1: 1010two× 0011two 1010two× 0011two = 101110two, i.e., -6ten×3ten = -18ten ELEC 5200-001/6200-001 Lecture 13

40. Example 2: 1010two× 1011two 1010two× 1011two = 011110two, i.e., -6ten×(-5ten) = 30ten *Last iteration with a negative multiplier in 2’s complement. ELEC 5200-001/6200-001 Lecture 13

41. Booth Multiplier Algorithm • A. D. Booth, “A Signed Binary Multiplication Technique,” Quarterly Journal of Mechanics and Applied Math., vol. 4, pt. 2, pp. 236-240, 1951. • Direct multiplication of positive and negative integers using two’s complement addition. ELEC 5200-001/6200-001 Lecture 13

42. A Multiplication Trick • Consider decimal multiplication: 4 5 7 9 9 9 0 1 4 5 7 4 1 1 3 4 1 1 3 4 1 1 3 4 5 6 5 4 7 5 7 Three additions • Operations for each digit of multiplier: • Do nothing if the digit is 0 • Shift left, i.e., multiply by some power of 10 • Multiply by the digit, i.e., by a number between 1 and 9 ELEC 5200-001/6200-001 Lecture 13

43. What is the Trick? • Examine multiplier: 99901 = 100000 – 100 + 1 • Multiply as follows: 457 × 100000 = 45700000 457 × ( - 100) = - 45700subtraction 45654300 457 × 1 = 457addition 45654757 Reduced from three to two operations. ELEC 5200-001/6200-001 Lecture 13

44. Booth Algorithm: Basic Idea • Consider a multiplier, 00011110 (30) • We can write, 30 = 32 – 2, or 00100000 (32) = 25 +11111110 (–2) = –21 00011110(0) 30 • Interpret multiplier (scan right to left), check bit-pairs: • kth bit is 1, (k-1)th bit is 0, multiplier contains -2k term • kth bit is 0, (k-1)th bit is 1, multiplier contains 2kterm • kth bit is 1, (k-1)th bit is 1, -2k is absent in multiplier • kth bit is 0, (k-1)th bit is 0, 2k is absent in multiplier • Product, M×30 = M×25 - M×21 M: multiplicand • Multiplication by 2k means a k-bit left shift ELEC 5200-001/6200-001 Lecture 13

45. Booth Algorithm: Example 1 • 7 × 3 = 21 0111 multiplicand = 7 ×0011(0) multiplier = 3 11111001 bit-pair 10, add -7 in two’s com. bit-pair 11, do nothing 000111 bit-pair 01, add 7 bit-pair 00, do nothing 00010101 21 ELEC 5200-001/6200-001 Lecture 13

46. Booth Algorithm: Example 2 • 7 × (-3) = -21 0111 multiplicand = 7 ×1101(0) multiplier = -3 11111001 bit-pair 10, add -7 in two’s com. 0000111 bit-pair 01, add 7 111001 bit-pair 10, add -7 in two’s com. bit-pair 11, do nothing 11101011 - 21 ELEC 5200-001/6200-001 Lecture 13

47. Booth Algorithm: Example 3 • -7 × 3 = -21 1001 multiplicand = -7 in two’s com. ×0011(0)multiplier = 3 00000111 bit-pair 10, add 7 bit-pair 11, do nothing 111001 bit-pair 01, add -7 bit-pair 00, do nothing 11101011 - 21 ELEC 5200-001/6200-001 Lecture 13

48. Booth Algorithm: Example 4 • -7 × (-3) = 21 1001 multiplicand = -7 in two’s com. ×1101(0)multiplier = -3 in two’s com. 00000111 bit-pair 10, add 7 1111001 bit-pair 01, add -7 in two’s com. 000111 bit-pair 10, add 7 bit-pair 11, do nothing 00010101 21 ELEC 5200-001/6200-001 Lecture 13

49. Booth Advantage Serial multiplication Booth algorithm 00010100 20 ×00011110 30 00000000 00010100 00010100 00010100 00010100 00000000 00000000 00000000________ 000001001011000 600 00010100 20 ×000111100 30 111111111101100 00000010100 __________________ 0000001001011000 600 Four partial product additions Two partial product additions ELEC 5200-001/6200-001 Lecture 13

50. Faster Multiplication • Using repeated additions, we need as many clocks as there are bits, say n, in multiplier. • Multiplication can be done in one clock. Of course, the period of clock will have to be longer; but may not be as long as n times. ELEC 5200-001/6200-001 Lecture 13

More Related