CS 3843 Spring 2011
Computer Organization
Lecture 2

Digital Logic Design I

Topics for today:

Digital signals

Electronic computers use Boolean logic for all computations. We need something to electrically represent 0 and 1. We use different levels of voltage, e.g. 0V and +5V. Boolean, i.e. binary, logic is used because it's easiest to have two levels instead of, say, 10.

This graph shows an electric signal representing a change from 0 to 1 and back to 0.

Wires

Wires implement communication of the digital signals in the computer. Wires transport signals from one place to another. They are made of some conductive metal. Although they have their limits and can be tricky, it's OK for us to think of them as being arbitrarily long, narrow, inexpensive, and fast as we build our mental model of what's going on inside the machine.

Transistors

In a digital computer, transistors are used as the most basic element of computation. They are connected by wires. Transistors are tiny switches. They have three terminals: When the gate terminal is triggered, current is allowed to flow from the source to the drain. Otherwise, no current flows from the source to the drain. Again, we can think of them for now as being arbitrarily small, cheap, and fast. Transistors are connected to one another and to power and ground by wires. Power and ground are concepts we don't really need for our purposes; they're just necessary for keeping the machine on. Just think of wires as pipes, like for water, that move information around, and transistors as faucets that turn the water on and off (1 and 0).

There are two types of MOS transistors used in the CMOS technology that dominates current computer architecture:

We'll get back to transistors in a little bit.

Truth Tables

A Boolean function is a function whose domain is a vector of bits and whose range is one bit. A Boolean function can be completely described by a truth table giving the values of the function for each combination of bits in the bit vectors.

Here's a truth table: 

a b c  f(a,b,c)
0 0 0     0
0 0 1     0
0 1 0     0
0 1 1     1
1 0 0     0
1 0 1     1
1 1 0     1
1 1 1     1

Logic Gates

Logic gates are physical components that implement logical functions like AND and OR. They are connected by wires to inputs, other gates, and outputs (as well as other things we will find out about later). Some important gates are: Logic gates can have more than two inputs; what would those truth tables look like?

Symbols for gates

When designing a digital logic circuit, we draw a circuit diagram showing the various gates and wires connecting them. The following symbols represent logic gates:

Logic gates from transistors

How do we compute with transistors? We'd like to be able to work with AND, OR, NOT, etc., not with "turn on" and "bad at passing 0."

For example, consider the NOT function: 

in  out
--  ---
0    1
1    0
Here is an example of a NOT gate (or inverter) implemented with transistors: 
              1                               
             _|                               
        |-o||_                                
___in___|     |__out___                       
        |    _|                               
        |--||_                             
              |                               
              0
How about a more complex function like NOR? Recall that the truth table for NOR is: 
a  b  a NOR b
0  0    1
0  1    0
1  0    0
1  1    0
How would we implement this in transistors?

Combinational Circuits

Combinational circuits are acyclic directed graphs whose vertices are inputs, outputs, or gates and whose edges are wires. They are drawn as circuit diagrams using the symbols we've seen. For instance, here is a combinational circuit:

The squares represent inputs to the circuit, while the unattached outputs represent outputs from the circuit. This circuit happens to be a 2-bit adder.

Implementing Boolean functions with digital circuits

Let's look at an interesting function we might want to compute using digital logic: 
a b c  f(a,b,c)
0 0 0     0
0 0 1     0
0 1 0     0
0 1 1     1
1 0 0     0
1 0 1     1
1 1 0     1
1 1 1     1
How would we implement this function with digital logic? First, we have to figure out a formula that computes this function. How about this: 
(a & b) | (b & c) | (a & c)
Yeah, that'll work. Now how do we implement this with logic gates? Like this:
Another way would be to use a 3-input OR gate:

Simulating other gates with NAND

It turns out that NAND is a very useful gate. A 2-input NAND consumes only 4 transistors, as opposed to AND and OR which each consume 6. Also, NAND can be used to compute any other logic function, i.e., it is complete for logic. How could we prove that?

The reasons why there are alternate symbols for NAND (as well as NOR) is because those bubbles represent NOTs that can be placed in pairs on wires to make normal ANDs and ORs into NAND (and NORs). For instance, the circuit we just saw can be drawn with just NANDs like this:


(Note that the 3-input gate is a NAND, just as an OR symbol with two bubbles on the inputs is a 2-input NAND.)

Computing with logic gates

We can use logic gates to compute with binary numbers; that's how computers do arithmetic. Imagine how you would design a circuit to do a bitwise OR, or a bitwise AND.

Majority function

Another useful function to compute would be the majority function: this function returns 1 if the majority of its inputs are 1, 0 otherwise. The function we've been with working so far is the 3-input majority function.

Parity function

The parity function is useful for a variety of reasons. It is true if and only if the number of inputs that are true is odd. How would you implement this function? Well, it turns out that XOR is the 2-input parity function. What about the 3-input parity function? Here's the truth table: 
a b c  parity(a,b,c)
0 0 0     0
0 0 1     1
0 1 0     1
0 1 1     0
1 0 0     1
1 0 1     0
1 1 0     0
1 1 1     1
One way to implement 3-input parity is using 2 XORs, like this:

Full adder

This brings us to a more useful computation: a full adder circuit. This circuit takes as input three 1-bit numbers and produces the 2-bit sum, which can be 0, 1, 2, or 3. It turns out that the least significant Let's look at a truth table for this function: 
a b c   c0 s0
0 0 0   0  0
0 0 1   0  1
0 1 0   0  1
0 1 1   1  0
1 0 0   0  1
1 0 1   1  0
1 1 0   1  0
1 1 1   1  1
c0 is the most significant bit of the sum, and s0 is the least significant bit. Notice anything? The most significant bit is the majority of the input bits, and the least significant bit is the parity. So we can build a full adder from parity and majority circuits.

Adding two n-bit numbers with combination circuits

We can connect full adders to each other to form a ripple-carry adder. Let's let a full adder be represented by this symbol:

The 3 wires on the left lead from the inputs, and the wires on the right lead from the outputs with the top output being the parity (or "sum") and the bottom output being the majority (or "carry"). Then a 4-bit ripple carry adder looks like this: