* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Arithmetic & Logic Part 2
Survey
Document related concepts
Transcript
Arithmetic and Logical Operations Part II 1 Unsigned Numbers Addition in unsigned numbers is the same regardless of the base. Given a pair of bit sequences X and Y, where X = xnxn-1...x1x0 Y = ynyn-1...y1y0 we need to add three terms: xi, yi, and ci for each bit position i = 0...n. ci is the carry bit in the ith bit position. Assume c0 = 0. Example 15.1 +) c1=1 011 (310) 001 (110) 100 (410) Note: ci+1 =(xi+yi+ci)div 2 zi =(xi+yi+ci)mod 2 1+1=10 z0 Example 15.2 +) 010010 011000 101010 000111 011001 100000 Overflow in Unsigned Numbers Given a pair of n bit sequences X and Y, we apply the following check for overflow: Overflow occurs if (X + Y) mod 2 X + Y An overflow occurs if a sum is no longer representable within a fixed number of bits. An incorrect sum is produced when there is a carry out from the most significant position. Example 15.3 Let the number of bits, n, be 4 1111 Carry out of 1 from the most significant 0001 position 10000 also we can check using (X + Y) mod 2n X + Y which gives us 0000 10000, thus overflow occurs. Sign Magnitude Integers If two integers in sign magnitude have the same sign, the two integers are added. The sign of the result is the same as that of the addends. The sign bit is not included in the process. Any carry out from the most significant bit of the magnitude is thrown away. Example 15.4 1 00100 (-4) +) 1 00101 (-5) 1 01001 (-9) 0 11111 (31) 0 1 (1) 0100000 (overflow) 0 00010 (2) 0 00111 (7) 0 01001 (9) A carry out from the most significant bit has occurred If two integers have different signs, a subtraction is performed and the sign of the result is decided in advance. The sign will be that of the integer with the larger magnitude. The subtraction of unsigned or sign magnitude integers is the same as the longhand subtraction of decimal numbers. For unsigned numbers, there are results which are not representable. Example 15.5 +) +) 0 00101 (5) 1 00011 (-3) 0 00010 (2) 0 00100 (4) 1 00010 (-2) 0 00010 (2) +) 1 00010 (-2) 0 00101 (5) 0 00011 (3) becomes 0 is now 10 borrow Overflow in Signed Magnitude If both addends are of the same sign, and there is a carry out from the most significant bit of the magnitude then overflow has occurred. If the addends have different signs, overflow will not occur. Two’s Complement In two’s complement, the same algorithm is applied to the operands regardless of the sign. Adding two numbers is done by simply applying the same algorithm used for unsigned numbers. Subtraction can be performed simply by adding the additive inverse of the subtrahend. Overflow in Two’s complement In two’s complement, overflow does not necessarily happen when there is a carry out of the most significant bit. An overflow occurs when both addends are of the same sign and the result is of the opposite sign If the carry into the most significant bit is not the same as the carry out from the most significant bit an overflow has occurred. Example 15.6 0101 (5) + 0010 (2) 0111 (7) 0011 (3) + 1100 (-4) 1111 (-1) 0101 (5) + 1101 (-3) 10010 (2) 1001 (-7) + 0111 (7) 10000 (0) carry out discarded. no overflow Example 15.7 1111 1000 (-8) 1111 1000 (-8) 1111 0000 (-16) 0000 0101 (5) 0100 0000 (64) 0100 0101 (69) 0111 1110 (126) 0110 0000 (96) 1101 1110 (-34) 1000 0010 (-126) 1111 1101 (-3) 0111 1111 (127) overflow Observations In Example 12.7, an overflow occurs when the numbers have the same signs but the result has a different sign. If you take both the carry-in of the msb and carry-out from the msb as inputs to an XOR boolean expression, the following holds: if the result is 1 there is an overflow if the result is 0 there is no overflow One’s Complement In one’s complement, addition is performed similar to the two’s complement with a slight modification: The carry out from the most significant bit is added to the partial sum. Example 15.8 0011 (+3) 1100 (-3) 1111 (0) 1101 (-2) 1011 (-4) carry-out is 11000 added to the 1 partial sum 1001 (-6) 0001 (+1) 1001 (-6) 1010 (+5) 0111 (+7) 1100 (-3) 10011 1 0100 (+4) Multiplication It may take as many as 2n bits to represent the product of two n-bit numbers Multiplication in unsigned numbers uses the longhand method we are already familiar with. By sign extending both the multiplier and the multiplicand to the size needed for the result, the algorithm for multiplying 2’s complement is the same as that of unsigned numbers Multiplication Algorithm while (count < no. of integer bits){ check if the multiplier’s last bit is 1 if (multiplier bit = 1){ add the multiplicand to the product} shift the multiplicand left 1 bit count = count + 1 shift the multiplier right 1 bit } Example 15.9 x 0010 0011 multiplicand multiplier 0000 + 0000 + 0000 initial product 0000 0010 0010 0100 0110 multiplicand shifted left final product If we have n-bit two’s complement integers, we both sign extend the multiplier and the multiplicand to the size needed for the result, i.e., 2n. By making both the multiplier and the multiplicand 2n the result of the operation will have 4n bits. The correct product is contained in the least 2n bits of the 4n-bit wide result Example 15.10 Let the number of bits for integers be 4 bits, i.e. n = 4. We sign extend the numbers using 2n bits. -3 x 6 = ? -3 = 1101 6 = 0110 11111101 00000110 showed only relevant addends. 1111101 note: there are only two 111101 1’s in the multiplier xxxxxxxx11101110 -18 discard Example 15.11 2 x -2 = -4 0000 0010 x 1111 1110 2 = 0010 0000 0000 0000 0001 +) 0010 0100 1000 0000 1111 -2 = 1110 0000 0100 Note: each addend 1000 is the multiplicand shifted left 1 bit 0000 in each step 0000 0000 0000 0000 1100 (-4) Example 15.12 -1 x -2 = ? -1 = 1111 11111111 (-1) 11111110 (-2) 11111110 11111100 11111000 11110000 11100000 11000000 10000000 00000010 (+2) -2 = 1110 Division Division of unsigned binary numbers is the performed similar to the longhand division of decimal numbers No convenient algorithm exists for division of signed magnitude and complement integers An exception that must be handled is division by zero Division Algorithm Align Most Significant Bits remainder = dividend while count < # bits shifted + 1{ remainder = remainder - divisor if (remainder < 0){ remainder = remainder + divisor shift left quotient & set lsb to 0} else shift left quotient & set lsb to 1 shift the divisor right count = count + 1 } Example 15.13 What is 31/4? (00011111 / 0100) C 0 1 2 Q 0001 0011 0111 D 0001 0000 0000 1000 0000 0100 R 0001 1111 0000 1111 0000 0111 R-D 1111 0111 0011 Notice the alignment of the bits for the D. 0100 became 00010000. The D was shifted left twice to be aligned. Count is 2+1=3, so the loop ends here. Example 12.14 What is the result of 99/3? (01100011 / 0011) C 0 1 2 3 4 5 Q 0000 0001 0000 0010 0000 0100 0000 1000 0001 0000 0010 0001 D 0110 0000 0011 0000 0001 1000 0000 1100 0000 0110 0000 0011 R 0110 0011 0000 0011 0000 0011 0000 0011 0000 0011 0000 0011 R-D 0011 Neg Neg Neg Neg 0 Notice we get the right answer just as the loop ends from the count. Dividing Signed Numbers We can use the algorithm to perform division on the absolute values of the numbers. Compare the signs of the dividend and the divisor. If they are the same, the quotient is positive. If they are not the same, the quotient is negative. The sign of the remainder is always the same as the sign of the dividend