COMP541 Arithmetic Circuits

1
COMP541Arithmetic Circuits

• Montek Singh
• Mar 20, 2007

2
Topics

• Adder circuits
• How to subtract
• Why complemented representation works out so well
• Overflow

3
Iterative Circuit

• Like a hierachy, except functional blocks per bit

4
Adders

• Great example of this type of design
• Design 1-bit circuit, then expand
• Lets look at
• Half adder 2-bit adder, no carry in
• Inputs are bits to be added
• Outputs result and possible carry
• Full adder includes carry in, really a 3-bit
  adder

5
Half Adder

• S X ? Y
• C XY

6
Full Adder

• Three inputs. Third is Cin
• Two outputs sum and carry

7
Two Half Adders (and an OR)
8
Ripple-Carry Adder

• Straightforward connect full adders
• Carry-out to carry-in chain
• C0 in case this is part of larger chain, or just
  0

9
Hierarchical 4-Bit Adder

• We can easily use hierarchy here
• Design half adder
• Use in full adder
• Use full adder in 4-bit adder
• Verilog code in textbook

10
Behavioral Verilog

• // 4-bit Adder Behavioral Verilog
• module adder_4_b_v(A, B, C0, S, C4)
• input30 A, B
• input C0
• output30 S
• output C4
• assign C4, S A B C0
• endmodule

Addition (unsigned)
Concatenation operation
11
Whats Problem with Design?

• Delay
• Approx how much?
• Imagine a 64-bit adder
• Look at carry chain

12
Carry Lookahead Adder

• Note that add itself just 2 level
• Idea is to separate carry from adder function
• Then make carry approx 2-level all way across
  larger adder

13
Four-bit Ripple Carry
Reference
Adder function separated from carry
Notice adder has A, B, C in and S out, as well as
G,P out.
14
Propagate

• The P signal is called propagate
• P A ? B
• Means to propagate incoming carry

15
What Does This Mean?

• No Carry Here
• So the propagate signal indicates that condition
  of incoming should pass on

16
Generate

• The G is generate
• Its G AB, so new carry created
• So its ORed with incoming carry

17
Said Differently

• If A ? B and theres incoming carry, carry will
  be propagated
• And S will be 0, of course
• If AB, then will create carry
• Incoming will determine whether S is 0 or 1

18
Ripple Carry Delay 8 Gates
19
Turn Into Two Gate Delays

• Design changed from deep (in delay) to wide

20
C1 Just Like Ripple Carry
21
C2 Circuit Two Levels
G from before and P to pass on
This checks two propagates and a carry in
22
C3 Circuit Two Levels
Generate from level 0 and two propagates
G from before and P to pass on
This checks three propagates and a carry in
23
What Happens as Scale Up?

• Can I realistically make 64-bit adder like this?
• Have to AND 63 propagates and Cin!
• Compromise
• Hierarchical design
• More levels of gates

24
Making 4-Bit Adder Module

• Create propagate and generate signals for whole
  module

25
Group Propagate

• Make propagate of whole 4-bit block
• P0-3 P3P2P1P0

26
Group Generate

• Does G created upstream pass on because of string
  of Ps (also G3)?
• Indicates carry generated in block

27
Hierarchical Carry
Left lookahead block is exercise for you
28
Practical Matters

• FPGAs like ours have limited inputs per block
• Instead they have special circuits to make adders
• So dont expect to see same results as theory
  would suggest

29
On to Subtraction

• First, look at unsigned numbers
• Motivates why we typically use complemented
  representation
• Then look at 2s complement
• Imagine a subtractor circuit (next)

30
One-bit Subtractor

• Inputs Borrow in, minuend and subtrahend
• Review subtrahend is subtracted from minuend
• Outputs Difference, borrow out
• Could use like adder
• One per bit

31
Example
32
Correcting Result

• What, mathematically, does it mean to borrow?
• If borrowing at digit i-1 you are adding 2i
• Next Slide

33
Correcting Result 2

• If M is minuend and N subtrahend of numbers
  length n, difference was
• 2n M N
• What we want is magnitude of N-M (with minus sign
  in front)
• Can get by subtracting previous result from 2n
• N - M 2n (M N 2n)

This is called 2s complement
34
Put Another Way

• This is equivalent to how we do subtraction in
  our heads
• Decide which is greater
• Swap if necessary
• Subtract
• Could build a circuit this way
• Or just look at borrow bit

35
Algorithm

• Subtract N from M
• If no borrow, then M ? N and result is OK
• Otherwise, N gt M so result must be subtracted
  from 2n (and minus sign prepended)

36
Pretty Expensive Hardware!
37
That Complex Design not Used

• Thats why people use complemented interpretation
  for numbers
• 2s complement
• 1s complement

38
1s Complement

• Given binary number N with n digits
• 1s complement defined as
• (2n 1) - N

39
2s Complement

• Given binary number N with n digits
• 2s complement defined as
• 2n N for N ? 0
• 0 for N 0
• Exception is so result will always have n bits
• 2s complement is just a 1 added to 1s complement

40
Important Property

• Complement of a complement generates original
  number
• NOTE We havent talked about negative numbers
  yet. Still looking at unsigned
• Lets look at new design for subtractor

41
New Algorithm for M-N

• Add 2s complement of N to M
• This is M (2n N) M N 2n
• If M ? N, will generate carry (why?)
• Discard carry
• Result is positive M - N
• If M lt N, no end carry (why?)
• Take 2s complement of result
• Place minus sign in front

42
Example

• X 101_0100 minus Y 100_0011

43
Example 2

• Y 100_0011 minus X 101_0100
• No end carry
• Answer - (2s complement of Sum)
• - 0010001

44
Adder-Subtractor

• Need only adder and complementer for input to
  subtract
• Need selective complementer to make negative
  output back from 2s complement
• Or go through adder again. See next slide

45
Design of Adder/Subtractor
S low for add, high for subtract
Inverts each bit of B if S is 1
Adds 1 to make 2s complement

• Output is 2s complement if B gt A

46
Signed Arithmetic

• Review characteristics of signed representations
• Signed magnitude
• Left bit is sign, 0 positive, 1 negative
• Other bits are number
• 2s complement
• 1s complement

47
Observations

• 1s C and Signed Mag have two zeros
• 2s C has more negative than positive
• All negative numbers have 1 in high-order

48
Advantages/Disadvantages

• Signed magnitude has problem that we need to
  correct after subtraction
• Ones complement has a positive and negative zero
• Twos complement is most popular
• Arithmetic operations easy

49
Conclusion 2s Complement

• Addition easy on any combination of positive and
  negative numbers
• To subtract
• Take 2s complement of subtrahend
• Add
• This performs A ( -B), same as A B

50
Overflow

• Two cases of overflow for addition of signed
  numbers
• Two large positive numbers overflow into sign bit
• Not enough room for result
• Two large negative numbers added
• Same not enough bits
• Carry out can be OK

51
Examples

• 4-bit signed numbers
• 7 7
• 7 7
• Generates carry but result OK
• -7 -7
• 4 4
• Generates no Cout, but overflowed

52
Overflow Detection

• Condition is that either Cn-1 or Cn is high, but
  not both