Computer Memory


The computing power of the modern information systems has reached an impressive rate. The global society evolves rapidly, and so do the computer systems. However, to maintain the significant progress of the past years it is crucial to develop and pioneer new services and technologies.

.

The purpose of this paper is to outline the main aspects regarding the computer memory. The author defines the notion of the main memory, covers the characteristics of the ideal memory and explores the concepts and ideas that may significantly improve the current memory generation.

The Major Trends Affecting Main Memory Requirements

To outline the major trends defining the memory requirements, it is crucial to understand the essence of the main memory concept. The main memory (sometimes called Random Access Memory or RAM) is the memory where Central Processing Unit temporarily stores all data needed. The CPU copies all processes into the main memory storage. In other words, the main memory serves as a bridge between the CPU and the hard drive. Today we can see two types of RAM presented on the market – Dynamic Random Access Memory (DRAM) and Static Random Access Memory (SRAM).

There are major factors defining the memory requirements. First and foremost it is important to outline that the present-day memory technology scaling cannot be improved significantly. The technical process has almost reached its limits.

The second crucial point is that the main memory power is the greatest of concerns today. There is a strong demand for the increase in the main memory capacity and bandwidth.

The next important issue is the fact that every computer system should be balanced. The point is that the computer will not take advantage of advanced CPU having a poor memory or Graphical Processing Unit. Therefore, the manufacturers develop their products in a synergistic way without direct intention to act that way. Finally, the next important trend affecting the requirements is the competition. It has always been an integral part of any product or service, and the main memory is not the exception. The technological process improvement along with cost cutting reveals great opportunities for both end users and manufacturers.

Ideal Memory Characteristics

The ideal memory concept cannot be interpreted as a constant issue. The ideal memory should have a high capacity, outstanding performance and low production cost. In the best-case scenario, the memory should hold the highest bandwidth and the lowest latency possible. The crucial point is that the energy consumption should be decreased to a minimum. It is important to mention that energy consumption/productivity rate should have the lowest value possible.

The Current Technology

The main technology in use today is DRAM. This technology means that all data presented is stored as the electric charge on the capacitors. The fact that capacitors can be either discharged or charged, all information is perceived as 0 or 1 standing for its absence or presence. The DRAM technology is the dominant technology of main memory because it is relatively cheap and efficient. However, it does have the drawbacks. The issue is that the capacitors tend to lose the charge, thus the cells should be periodically refreshed. The refreshment rate is usually less than 2 ms, but nevertheless it decreases the efficiency when compared to non-refreshable technologies. The next point is that DRAM is facing a threat of scalability. It means that the cells used in the technology cannot be scaled down beyond 42 nm. Another technology widely used today is SRAM, which means Static Random Access Memory. The SRAM technology does not require periodical charging, thus there is no need for refreshment. When DRAM usually requires one transistor per bit, SRAM needs from 4 to 6 transistors per bit. Moreover, the DRAM costs from 8 to 16 times less than SRAM meaning that the latter is preferred choice in faster memory types (e.g. cache memory). Finally, SRAM has less capacity compared to DRAM.

While DRAM is the closest-to-ideal memory concept, there are techniques to improve the actual technology. The RDRAM technology is the successful modification created by Rambus. The main difference is that DRAM uses row/column address strobe while RDRAM uses a split transaction bus. Another modification is called Direct Rambus DRAM, or simply DRDRAM. It uses the half row buffers instead of a full row in RDRAM. Moreover, the DRDRAM has 3-byte wide channel compared to 1-byte channel in RDRAM.

The Possible Alternatives

The scientists are constantly pioneering new technologies trying to find the alternatives. Despite the fact that DRAM is so far the best solution for the industry, the non-volatile types of memory are perceived as the potential replacement for dynamic memory.

The flash memory is the type of non-volatile memory that can be reprogrammed. It retains the data without a power need and is four times denser that DRAM. However, the limited number of erase/write cycles and memory wearing will definitely hold the flash memory off becoming the worthy replacement to DRAM. Also, writing help you can find at the essay professor

The next type is called Phase Change Memories, or simply PCM. The PCM technology has both advantages and disadvantages compared to DRAM. To begin with, PCM has a higher scalability and storage density. Furthermore, it is immune from cross-coupling effects. It also has zero leakage power and low “burst read” latencies. Despite those advantages, PCM has its limitations. It has limited lifetime just like the flash memory, possess higher access latencies and consumes a substantial amount of energy.

The main memory of present day has both pros and cons. In spite of the fact that numerous attempts have been taken to reveal the revolutionary new type of memory, DRAM still is the best solution on the market. The new technologies such as PCM are successfully pioneered, but the final results are still to come.

5 Views