Arithmetic logic unit

Various types of UAL

The UAL can be specialized or not. The elementary UAL calculate on integers, and can carry out the common operations, which one can separate in four groups:

arithmetic operations: addition, subtraction, change of sign etc,
logical operations: ones complements, with two, AND, OR, Exclusive-OR, NOT, NON-ET etc,
comparisons: test of equality, superior, inferior, and their equivalents “or equal”,
possibly of the shifts and rotations (but sometimes these operations are externalisées).

Certaines UAL is specialized in the handling of the numbers with Floating decimal point, in simple or double precision double precision (one speaks about Unité of floating-point calculation or floating not links (FPU)) or in vectorial calculations . Typically, these units can achieve the following operations:

additions, subtractions, change of sign,
multiplications, divisions,
comparisons,
modulos

Certain UAL, generally of the class of FPU S, in particular those of the supercomputers, are likely to offer advanced functions:

opposite (1/x),
square root,
logarithms,
functions transcendantales (sin X, cos X, etc),
vectorial operation (produced scalar, vectorial, etc),
etc

A processor calls upon several UAL, at least two: one located in the Way of control to increment the register of program (of 1,2,4 or 8 typically), and another in the Way of data to process the data. To increase their performances, they are generally pipelinées.

Notation

A UAL is schematized as on the figure opposite. This one has two entries has and B on which one will present the data to be treated. The entry F will indicate the operation to be carried out. Lastly, this one has two exits, R which will be the result of the operation, and D the flags which will state either that there was error (divide check, capacity overshooting etc), or of the codes conditions (superior, inferior, equal to zero etc). In the middle of this diagram, the operator is marked (for example the operator +, X,…).

Examples

The UAL 4 bits 74181 (TTL)

Here, for example, a UAL which was very much used at the time when only integrated circuits TTL was available: 74181, a UAL of 4 bits. This circuit is intended to be connected in cascade with others to treat entireties of size higher than 4 bits. It is at the base of the UAL of the Alto and in many models of PDP-11 (16 bits).

Its truth table describing which operation is accomplished according to the values of the entries M and S.

Caution: this table can contain errors.

The diagram of this circuit is given Ci below.

Use of multiplexers 4 towards 1 like UAL

This technique is used in the UAL of IBM Strech alias IBM 7030 (1958). The concept rests on the wise choice of the code function F: it also corresponds to the result of the desired operation. For example, to do one AND, one will choose the code F = 0001, corresponding to the last column of the truth table of AND logical: manufactured by the professor Takesrit Mr.

According to the value presented on the entry F, one generates the desired function: F = 0001 is one AND, F = 0111 one OR, F = 0110 one OR exclusive etc

Thus, the entries (2 bits each one) has and B are used to select the good bit of the entry F. the first multiplexer takes, like entry, the two bits of weak weight of has and B. They are used to select one of the lines of F: 00 will select the first line, 01 the second and so on. The second multiplexer acts in the same way with the two bits of strong weight of has and b:

The diagram of multiplexers 4 towards 1 above is built containing integrated circuits TTL completely standard, whose references are posted. It is about doors NOT, And NON-OU.

See too

Electronic Boolean algebra
Data-processing
List of microprocessors
Microprocessor

External bonds

Site “How it goes”
a simulator of processor MIPS simplified

Random links: