This preview shows pages 1–4. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Problem 1. Using the diagrams on slides 20 and 23 from lecture 4, what is the maximal latency
to drive out the ﬁnal CarryOut and all Sum bits for a 16bit hierarchical carry lookahead adder?
Assume that for any logic gate (and, or, xor) with a fanin of k (i.e. number of inputs), the delay
of that gate is k/2 times T (rounded down to the nearest whole number). So a gate with fanin 2
or 3 would have delay T, and fan—in of 4 or 5 would have delay 2T, etc. Print out those two slides and show your work rather than just giving the ﬁnal answer (which should be in terms of T). G = A and B ( = T )
P = A xor B ( H
H From Slide 20, we get that G (alpha, beta, gamma)
= GO.P1.P2.P3 + Gl.PO.P1 + G2.PO.P1 + G3 { {ﬁx w,
T T T T T T T T T T T
T + 2T(.) T + T(.) T + T(.) T
3T + 2T(+)
G(alpha, beta, gamma) = 5T xT(.l+) —> indicates that there is delay ’x’ for that operation Similarly for P (alpha, beta, gamma) = PO.P1.P2.P3 we get, T +
2T()
P(alpha, beta, gamma) = 3T From slide 23,
// C12 = Gy + GB.Py + Goc.PB.Py + CO.Poc.PB.Py 5T 5T T 5T 3T 3T 0 3T 3T 3T 5T 5T+T(.) 5T+T(.) T+2T(.) 6T + 2T(+) C12 = 8T
Using the formula for C3 from Slide 20 with Cin = C12,
C15 = G2 + G1.P2 + GO.P1.P2 + Cin.PO.P1.P2 T T T T T T 8T T T T T T+T(.) T+T(.) 8T+2T(.)
lOT + 2T(+)
C15 12T 815 A15 xor B15 xor C15
OT OT 12T
815 = 12T + T(Xor) = l3T Extend formula from slide 23 to get, C16 = G8 + G728 + GB.P7.P6 + Goc.PB.Py.P8 + CO.Poc.PB.Py.P8
We get C16 = 9T Maximal Latency = 13T (Sum) Problem 2. Use Booth's Algorithm to multiply 1010 by 0110. Show your work as was done on
slide 12 of lecture 5. Operation Multiplicand Product / Next
0. Initial 1010 0000;0110 0 00 —> NOOP
1a. 0000 0p11 0 ‘ Shift
1b. + 0110 0000 10 —> Sub
\ 0110 4011
2a 0011 0001 1 Shift
2b 0011 OOLl 1 11 —> NOOP
3a 0001 100% 1 Shift
3b + 1010 000p 01 —> Add
gﬁawiXxﬁﬂ£m%~ 1011 100:0[ .
4a I j>1101 1100i0 Shift
4b 1101 1100 p Done Problem 3. In the worst case, Booth's algorithm can actually perform more operations (adds and
shifts) than the multiply algorithm of slide 5 of lecture 5. What is that worst case? Worst case for Booth’s algorithm is a sequence 1010 Problem 4. To avoid the worst case in problem 3, we will try a new version of Booth's algorithm
 one that looks at three bits of the multiplier at a time rather than just two. Can you formulate a
table like the one in slide 10 that handles a three bit Booth‘s algorithm? HINT  consider the
current bit, the bit to the right, and then the bit to the right of that. ——
nothing Mddle of a run of Os.
do nothin I
{)6 :noth§ng
add 1) 10 : sub,
03 : add {3*}:
> —b + 2b = ’0
add 2b 11 : nothing,
{)1 :add 13*?
> +2b suhb Dlzadd
10:5ub2 * b
>b—2b=b Sub b 7 f}
1
sub 2b 90 : nothing,
If) : sub b*2
> 2b 1 I nothing Middle of a run of 15.
Do nothing
After each operation product is Shifted twice. An adding or subuactmg b after a shift to right is similar to shifting b to left and doing add or
subtraction. So we can use 2*!) (shifted b) when it is the second operation. Rf '03: 5 : 0
was :7 it"? + Hum A44 hawmeme
A , a ’ 1,. W
(“00 O $M.Hr13 U“
+093 0 MA ‘0‘: we 0’. j: (3} i? a , L,
.r Del») Vi}: (“Ad/W) +
00! o co: 2 A3 a=010101 =21
b=000110 > 223:001100 b=111010 2b=1}0100 b=6 000000 01010} O (010: addb)
00030 000110 010101 0 (shift mithmeﬁc twice) 000001 100101 0 (010: addb}
000110
000131100101 1 (Shiﬁtwica) 000001 111001 0 (010: add b)
000110
00013.1 11100} 0 (Shift mice) 000001 H1110 0 product : 0130001 111110 =126
There are 3 operaﬁens, where there are 6 operations in {moth 2 algorithm (for each 01 and 10
there is one add or subtract) ...
View
Full
Document
This note was uploaded on 04/18/2010 for the course CS 151B taught by Professor N/a during the Spring '10 term at UCLA.
 Spring '10
 N/A

Click to edit the document details