Everyone knows what an invaluable contribution to the development of computer technology made American scientists . Although research has been conducted at many universities in this country, the palm still belongs to Harvard and Princeton. It was there that variants of the architecture of computers were developed that predetermined the concepts of development of computer technology.
How it all started
In the mid-1930s, the United States Department of War instructed Princeton and Harvard Universities to create an electromechanical computer system designed to support the artillery of the United States Navy.
As a result, 2 concepts appeared. Today they are known as Harvard and Princeton architecture.
Conceptually difference
The main difference between these two concepts is that von Neumann (Princeton) architecture used a single memory, i.e., a common data bus. As for her rival, the author of which was Howard Aiken , for its implementation requires the presence of several tires.
In addition, the Harvard architecture of the computing system differs from the Princeton architecture in that its implementation with sufficient complexity of the circuit is faster. This is due to the fact that in the von Neumann version of the program memory and data cannot be accessed simultaneously.
Harvard architecture vs Princeton
As you know, the main components of computers are ALU and memory. Obviously, the fewer conductors between them, the better. From this point of view, and taking into account the technical capabilities that took place before the end of the 60s of the last century, Neumann architecture was certainly in the lead. It was she who formed the basis for the design of RISC processors.
But scientific and technological progress did not stand still, and in the 70s of the twentieth century , semiconductors appeared. With their help, it was possible to create many microconductors, which eliminated the problem of using a large number of contacts and led to the onset of the era of Harvard architecture.
Further confrontation
The appearance of processors developed on the basis of Harvard architecture was met without much enthusiasm, since at that time there was no software capable of making tangible advantages of their use. In particular, due to the fact that they could not work at high frequencies, they were called processors for the poor.
The situation with the demand for Harvard architecture changed after the appearance of the Apple I PC. It operated on an 8-bit MOS 6502 processor operating on Harvard architecture and with Apple DOS OS.
The simplicity of the operating system was offset by a sophisticated processor design called the CISC. It had a separate 16-bit address bus and allowed arbitrary manipulation of registers. The CISC processor had a performance that was several times higher than all existing ones.
Following this, IBM echoed the idea of Apple, creating an IBM-PC personal computer with an Intel processor, operating according to the concept of Harvard architecture. As the OS, a Microsoft company product, Microsoft DOS, was used. Systems with this composition are called Wintel.
Disadvantages of harvard computer architecture
For the performance of the CISC processor, I had to pay double / triple the number of contacts. This not only caused it to overheat, but also imposed restrictions on its size. On average, for every 20 percent increase in Harvard processor performance, its power consumption increased to 50 percent.
To solve this problem, multi-core processors were invented, in which the frequency of operation of each individual core was reduced, but in terms of total performance they exceeded the performance of even overclocked single-core .
The influence of Harvard and Princeton architectures for building computer systems on the further development of computational mathematics
The transition to multi-core BCs led to the end of the era of classical programming, because multi-threaded computing required a change in classical programming algorithms based on flowcharts. All this led to a gap between the capabilities of computers and the existing capabilities of computational mathematics. The problem was aggravated after the proof of Amdahl’s law , according to which it is impossible to develop a fully parallelized computational algorithm, i.e., one in which there would not be a certain fraction of sequential operations.
Modern PC architecture
Today there are computers with different types and even hybrid architectures. However, the basic principles that define them are:
- Program management . It allows you to automate the process of computing on a computer. According to this principle, the solution of any problem is carried out according to a program that determines the sequence of actions of a PC.
- The principle of a program stored in memory. It contains the requirement to issue commands in the form of numbers, as is the case with data, and process them in the same way as numbers. At the same time, it itself is loaded into RAM before starting, which makes it possible to speed up the execution process.
- The principle of random access to computer random access memory. Elements of programs and data are recorded in any place of the OP . This solution allows you to access a specific memory location without looking at the previous ones .
Now you know how Harvard architecture differs from Princeton architecture and what significance they have for the development of computer technology. Perhaps, over time, new principles of building computer systems will appear that will achieve results that seem fantastic today.