Computer Memory Characteristics | Computer Architecture

Computer Memory Characteristics – Ideally, a processor should have rapid and uninterrupted access to the memory unit available in the system. Memories that work at processor speed are expensive and generally not employed in small digital systems. There are several memory characteristics that determine the type of memory to be used in a system. The primary goal of every memory system is to provide sufficient storage capacity with an acceptable level of performance and cost. The most important memory characteristics are discussed as follows:

Memory Access Time (Latency)

In order to read or write data in a particular place on the disk, the read/write head of the disk needs to be physically moved to the correct place. This process is known as accessing the memory disk. The time taken by the head to move to the right place is access time. Access time for a given disk varies depending on how far the head’s destination is from its origin at the time of each read or write instruction. Therefore, the generally average access time of a disk is calculated 

In a random-access memory, each storage location can be accessed independently of the other locations. There is a separate access mechanism, or read/write head, for every location, as shown in the figure below.

In serial memories, the access mechanism is shared by storage locations and must be assigned to different locations at different times by moving the stored information, the read/write head, or both. In a serial access mechanism, the time required to access memory is more than the time required in a random access mechanism. Therefore, serial access tends to be slower than random access. The serial access mechanism is shown below.

Performance and Cost

The performance of an individual memory device is primarily determined by the rate at which data can be read from and written into memory. A basic measure of performance is the average time to read a fixed amount of information. This has already been defined as access time.

The cost of a memory device is another characteristic of a memory unit. The measure of the cost of a memory device is the purchase price of a complete unit. The price includes not only the cost of the information storage medium but also the cost of the peripheral equipment needed to operate the memory. The cost of two memory devices can be compared on the basis of the cost of the memory devices per unit storage capacity. If the price of a memory device (storage capacity S bits) is CRs, then the cost per bit of the memory is defined as (C/S) Rs per bit 

The desirable memory characteristics are short access time and low cost. But these two characteristics are incompatible. Memory units with high performance (less access time) are expensive, while low-cost memory units have poor performance (slower). The figure below shows the relationship between cost and performance.

The above figure shows that the cost of magnetic tape memories is low, but the performance is also poor (high access time), whereas SRAM has a high cost and good performance (less access time).

Memory Bandwidth

Memory bandwidth is the rate at which data can be read from or stored into a semiconductor memory by a processor. Memory bandwidth is usually expressed in units of bits/bytes per seconds. This is also known as the data transfer rate of the memory unit.

Theoretical maximum memory bandwidth can be computed by multiplying the width of the interface by the frequency at which it transfers data. This is also referred to as the burst rate of the interface. 

Memory Cycle Time

This concept is basically applied to random-access memories. The memory cycle time consists of the access time plus any additional time required before a second access can commence. The additional time may be required to regenerate data if it is destroyed during the reading operation or if signals are not received properly. This characteristic of memory is concerned with the system bus, not the processor.

Reliability

This characteristic of memory is a measure of the mean time before failure. In general, memories with no moving parts have much higher reliability than memories such as magnetic disks, which involve mechanical motion. When a high data transfer rate for a memory is chosen, its reliability reduces. To increase the reliability of any memory, error-detecting and correcting codes are used.