A sort of computer memory in which things may be accessed by matching some portion of their content rather than their address (thus associative memory or Content-addressable memory (CAM)) is called associative memory. Associative memory is significantly slower than RAM and is rarely used in popular computer architectures. For example, functions as an identification tag, associative memory is utilized in multilayer memory systems, where a small fast memory, such as a cache, can store copies of specific blocks of larger memory for quick access.

To recover a word in associative memory, a query key (or descriptor) representing specific values of all or portion of the word's bits must be given. This key is contrasted simultaneously with the appropriate lock and tag bit of all recorded words, and any words that match this key are marked as accessible.

The cost of implementing associative memory as integrated circuits is high. Associative memory can be used in certain extremely fast searching applications. Associative memory may search for data (tag) by content rather than address.

Whenever the associative memory is confronted with an important pattern, it responds by creating one of the previously stored patterns that closely resembles or corresponds to the key pattern.

It may be thought of as data correlation in this case. The information that is input is correlated with the CAM's stored data.

It comes in two varieties:

Autonomous associative memory network (AAMN): A recurring neural network, also known as an auto-associative memory system, is a kind of associative memory which is used for recalling a pattern from incomplete or degraded inputs. The output of an auto-associative network is sent back to its input, enabling the network to acquire and recall the associations it has been trained on. This form of memory network is often employed in applications like as speech recognition and picture recognition, where the information being provided may be missing or noisy.
Associative memory network with heterogeneous associations: A hetero-associative memory system is a sort of associative memory in which one set of patterns is associated with another. The input pattern is connected with a separate output pattern in a hetero-associative system, enabling the network to learn and retain the relationships between both sets of patterns. This form of storage network is often utilized in data compression and retrieval applications.

Associative memory is a type of standard semiconductor memory (often RAM) with additional comparison circuitry that allows the search process to be completed in just one clock cycle. It is a silicon search engine, which is a sort of computer memory that is utilized in particular high-speed searching applications.

Block Diagram of Associative memory

The picture shows an associative memory block diagram. It has a storage array with logic for m bits of n bits each. Each argument registration A and key registration K contain n bits, one per every word bit.
Each memory word is represented by one bit in the match register M. Each word in ram is linked to the contents of an argument register in parallel.
Words that link the argument register bits set a corresponding bit in the corresponding match registers. The bits in the corresponding word registration that is set after the process of matching indicate that the corresponding words have been linked.
Reading is accomplished by sequential memory access for words with corresponding bits set in the match register.
A mask can be used in the key register to choose a certain field or key in an argument word. If the key register contains all 1's, every argument is differentiated with each memory word.
As a result, only bits in the argument with 1s in their equivalent location of the key registers are compared. As a result, the key provides a mask or recognizes an element of data that specifies how the memory reference is formed.
The diagram below depicts the relationship between the array of memory and the external register in memory that is associative.
The array's cells are represented by the initial C and have two subscripts. The first subscript specifies the word number, while the second specifies the bit location inside the word. Cell Cij is therefore a cell for word j in word i.
Each bit in the argument's register is compared to all the bits in the array's column j that support Kj = 1. This is done for all rows j = 1, 2,..., n.
If there is a match between the entire exposed bit of the argument's argument register and each of the bits in phrase i, the match register's corresponding bit Mi is set to 1. Mi is free to proceed if one or more unmasked parts of the argument or the phrase do not match.

How Does Associative Memory Function?

Data in conventional memory is stored in specified places, known as addresses, and is accessed by referring to those addresses. Data is stored in associative memory together with supplementary tags and metadata that define its content. When you run a search, the associative memory matches the search query to the labels of all stored data and gets the material that matches the query.

Associative memory is intended to identify matching facts rapidly, even if the search query is partial or imprecise. This is accomplished by employing parallel processing techniques, which allow several search queries to be executed concurrently. The lookup is also completed in just one phase, as opposed to traditional memory, which requires numerous processes to seek the data.

Applications of Associative Memory

It is only applicable in memory allocation form.
It is commonly used in database management systems and other applications.
In network routing tables, associative memory is employed to quickly discover the path to a target network depending on its address.
Image processing: In image processing applications, associative memory is utilized to look for certain features or patterns inside an image.
Associative memory is employed in applications of AI like expert systems or pattern recognition.
Database administration: In database management systems, associative memory may be utilized to swiftly retrieve data depending on its content.

Advantages of Associative Memory

It is utilized when the search time must be kept to a minimum.
It is appropriate for simultaneous searches.
It is frequently used to speed up databases.
It is utilized in virtual memory page tables as well as neural networks.

Drawbacks of Associative Memory

It is more costly than RAM.
Each cell must include storage and logical circuitry for correlating its content to external arguments.

← Prev Next →

COA Tutorial