Register Transfer Language in Computer Architecture

Register Transfer Language – A micro operation specifies an operation whose result is stored typically in a register. The term ‘Register Transfer’ indicates the availability of a hardware logic circuit that can perform a stated micro operations and transfer the result of the operation to the same or a new register.

In other words, it implies the availability of the digital hardware circuits to transfer the data or information from the outputs of the source register to the inputs of the destination register, and the destination register has a parallel load capability.

Information transfer from one register to another is represented in symbolic form by means of a replacement operator. For example, the following statement shows the transfer of the contents of register R₁ into register R₂.

R2 ← R1

It means the contents of register R₂ are replaced by the contents of register R₁, but the contents of source register R₁ do not change after the transfer.

What is Register Transfer Language?

The symbolic notation used to describe the micro operation transfers among registers is called a Register Transfer Language (RTL). A register transfer language is a system for expressing the micro-operation sequence in symbolic form among the registers.

The basic notations for the register transfer are shown in the table below. Registers are denoted by capital letters, and numerals follow the letters. Parentheses are used to denote a part of a register by specifying the range of bits.

The comma is used to separate two or more micro operations that are executed at the same time. The arrow represents a transfer of information and the direction of transfer.

Symbol	Description	Examples
Letters and numerals	Register	R1, R2, MAR
Parentheses ( )	Part of register	R1 (0-7), R2(H)
Arrow ←	Transfer of information	R2 ← R1
Comma ,	Two parallel micro operations	R2 ← R1, R1 ← R0

Data Movement Around Register

To understand the data movement around registers, consider an example of a digital system with two 1-bit registers, R₁ and R₂. The micro operation that transfers (copies) the contents of register R₁ to register R₂ can be represented as R₂ + R₁.

The expression does not show how data is copied from register R₁ to register R₂. It simply shows the transfer of contents to be made. This micro operation can be implemented using two methods:

1. Via a direct connection between registers.
2. Via a bus connection between registers.

The above mentioned micro operation (R₂-R₁) using both methods is shown in the figure below.

The set of micro operations for a system is mostly sufficient for designing its data paths. Data paths are defined as the connection between components used for data transfers. The basic limitation of such a system is that it does not show the conditions under which the transfers occur.

The above figure shows that the data may flow from register R₁ to register R₂, but it does not specify when R2 should load this data. To avoid such a limitation, normally, the transfer occurs only under a predetermined control condition.

For example, α is the control input, and data transfer occurs when it is high. This can be shown by means of an if-then statement.

IF (α = 1) THEN (R2 ← R1)

Sometimes, the control function is used to specify the register transfer by separating the control variables from the register transfer micro operation. Hardware Description Languages (HDLs) often use this compact notation. The syntax for this is given as follows:

Conditions: Micro-Operations

When all conditions to the left of the colon are satisfied, the data transfers specified by the micro operations are performed. The above transfer could be written as

α : (R2 ← R1)

Now the data transfer shown in the above figure can be modified with the control input (α) as shown in the below figure.

One way to improve system performance is to perform two or more micro operations simultaneously. In this notation, the micro operations are separated by commas; the order in which they are written is unimportant.

All micro-operations are performed concurrently. For example, if a system is to perform the transfers (R₂ ← R₁) and (R₃ ← R₄) whenever control input α is high (α = 1), this can be written as either.

α : R2 ← R1, R3 ← R4 or α : R3 ← R4, R2 ← R1

Sometimes, it is required to move a constant value into a register, rather than data from another register. For example, register R₁ is required to be loaded with 0 if control input α is high (α = 1) and to be loaded with 1 if control input ẞ is high (ẞ = 1). This can be represented as follows:

α : R1 ← 0
ẞ : R1 ← 1

The above expression shows that the register R₁ will be loaded with 0 for α = 1, irrespective of the control input ẞ, and in the same way, it will be loaded with 1 for ẞ = 1, irrespective of the control input α. If the case α=1 and ẞ=1 occurs, the register will try to clean its value and load the value 1 simultaneously. This is not an acceptable situation. There are two methods to avoid such a condition:

1. If the hardware that generates control inputs α and ẞ can ensure that they are never both set to 1 simultaneously.
2. The conditions can be modified so that they are mutually exclusive.

Therefore, in the previous example, conditions can be modified in three ways:

WAY – FIRST

αẞ1 : R1 ← 0
ẞ : R1 ← 1

WAY – SECOND

α : R1 ← 0
αẞ : R1 ← 1

WAY – THIRD

αẞ : R1 ← 0
αẞ : R1 ← 1

In all three ways, there will be no case when both control inputs α and ẞ are high, and micro-operation is required to perform. When α=ẞ=1, then there will be no change in register R₁.

Related questions and answers