In computer science and engineering, computer architecture is the practical art of selecting and interconnecting hardware components to create computers that meet functional, performance and cost goals and the formal modelling of those systems.
The noun computer architecture or digital computer organization is a blueprint, a description of the requirements and basic design for the various parts of a computer. It is usually most concerned with how the central processing unit (CPU) acts and how it accesses computer memory. Some currently (2011) fashionable computer architectures include cluster computing and Non-Uniform Memory Access.
The art of computer architecture has three main subcategories:[1]
The second step of designing a new architecture is often to design a software simulator, and write representative programs in the ISA, to test and adjust the architectural elements. At this stage, it is now commonplace for compiler designers to collaborate, suggesting improvements in the ISA. Modern simulators normally measure time in clock cycles, and give energy use estimates in watts.
Once the instruction set and microarchitecture are described, a practical machine needs to be designed. This design process is called the implementation. Implementation is usually not considered architectural definition, but rather hardware design engineering.
Implementation can be further broken down into several (not fully distinct) steps:
For CPUs, the entire implementation process is often called CPU design.
Contents |
The term “architecture” in computer literature can be traced to the work of Lyle R. Johnson, Muhammad Usman Khan and Frederick P. Brooks, Jr., members in 1959 of the Machine Organization department in IBM’s main research center.
Johnson had the opportunity to write a proprietary research communication about Stretch, an IBM-developed supercomputer for Los Alamos Scientific Laboratory. In attempting to characterize his chosen level of detail for discussing the luxuriously embellished computer, he noted that his description of formats, instruction types, hardware parameters, and speed enhancements was at the level of “system architecture” – a term that seemed more useful than “machine organization.”
Subsequently, Brooks, one of the Stretch designers, started Chapter 2 of a book (Planning a Computer System: Project Stretch, ed. W. Buchholz, 1962) by writing, “Computer architecture, like other architecture, is the art of determining the needs of the user of a structure and then designing to meet those needs as effectively as possible within economic and technological constraints.”
Brooks went on to play a major role in the development of the IBM System/360 (now called the IBM zSeries) line of computers, where “architecture” gained currency as a noun with the definition as “what the user needs to know”. Later the computer world would employ the term in many less-explicit ways.
There are many types of computer architectures:
The quantum computer architecture holds the most promise to revolutionize computing.[3]
Some practitioners of computer architecture at companies such as Intel and AMD use more fine distinctions:
The coordination of abstract levels of a processor under changing forces, involving design, measurement and evaluation. It also includes the overall fundamental working principle of the internal logical structure of a computer system.
It can also be defined as the design of the task-performing part of computers, i.e. how various gates and transistors are interconnected and are caused to function per the instructions given by an assembly language programmer.
Memory organization defines how instructions interact with the memory.
Computer organization helps optimize performance-based products. For example, software engineers need to know the processing ability of processors. They may need to optimize software in order to gain the most performance at the least expense. This can require quite detailed analysis of the computer organization. For example, in a multimedia decoder, the designers might need to arrange for most data to be processed in the fastest data path and the various components are assumed to be in place and task is to investigate the organisational structure to verify the computer parts operates.
Computer organization also helps plan the selection of a processor for a particular project. Multimedia projects may need very rapid data access, while supervisory software may need fast interrupts.
Sometimes certain tasks need additional components as well. For example, a computer capable of virtualization needs virtual memory hardware so that the memory of different simulated computers can be kept separated.
The computer organization and features also affect the power consumption and the cost of the processor.
The exact form of a computer system depends on the constraints and goals for which it was optimized. Computer architectures usually trade off standards, cost, memory capacity, latency and throughput. Sometimes other considerations, such as features, size, weight, reliability, expandability and power consumption are factors as well.
The most common scheme carefully chooses the bottleneck that most reduces the computer's speed. Ideally, the cost is allocated proportionally to assure that the data rate is nearly the same for all parts of the computer, with the most costly part being the slowest. This is how skillful commercial integrators optimize personal computers.
Modern computer architectural performance is often described as MIPS per MHz (millions of instructions per second per millions of cycles per second of clock speed). This metric explicitly measures the efficiency of the architecture at any clock speed. Since a faster clock can make a faster computer, this is a useful, widely applicable measurement. Historic complex instruction set computers had MIPs/MHz as low as 0.1 (See instructions per second). Simple modern processors easily reach near 1. Superscalar processors may reach three to five by executing several instructions per clock cycle. Multicore and vector processing CPUs can multiply this further by acting on a lot of data per instruction, and have several CPUs executing in parallel. Counting machine language instructions would be misleading because they can do varying amounts of work in different ISAs. The "instruction" in the standard measurements is not a count of the ISA's actual machine language instructions, but a historical unit of measurement, usually based on the speed of the VAX computer architecture. Historically, many people measured the speed by the clock rate (usually in MHz or GHz). This refers to the cycles per second of the main clock of the CPU. However, this metric is somewhat misleading, as a machine with a higher clock rate may not necessarily have higher performance. As a result manufacturers have moved away from clock speed as a measure of performance.
Computer performance can also be measured with the amount of cache a processor has. If the speed, MHz or GHz, were to be a car then the cache is like the gas tank. No matter how fast the car goes, it will still need to get gas. The higher the speed, and the greater the cache, the faster a processor runs.
Other factors influence speed, such as the mix of functional units, bus speeds, available memory, and the type and order of instructions in the programs being run.
In a typical home computer, the simplest, most reliable way to speed performance is usually to add random access memory (RAM). More RAM increases the likelihood that needed data or a program will be in RAM. So, the system is less likely to need to move memory data from the disk. The disk is often ten thousand times slower than RAM because it has mechanical parts that must move to access its data.
There are two main types of speed, latency and throughput. Latency is the time between the start of a process and its completion. Throughput is the amount of work done per unit time. Interrupt latency is the guaranteed maximum response time of the system to an electronic event (e.g. when the disk drive finishes moving some data).
Performance is affected by a very wide range of design choices — for example, pipelining a processor usually makes latency worse (slower) but makes throughput better. Computers that control machinery usually need low interrupt latencies. These computers operate in a real-time environment and fail if an operation is not completed in a specified amount of time. For example, computer-controlled anti-lock brakes must begin braking within a predictable, short time after the brake pedal is sensed.
The performance of a computer can be measured using other metrics, depending upon its application domain. A system may be CPU bound (as in numerical calculation), I/O bound (as in a webserving application) or memory bound (as in video editing). Power consumption has become important in servers and portable devices like laptops.
Benchmarking tries to take all these factors into account by measuring the time a computer takes to run through a series of test programs. Although benchmarking shows strengths, it may not help one to choose a computer. Often the measured machines split on different measures. For example, one system might handle scientific applications quickly, while another might play popular video games more smoothly. Furthermore, designers have been known to add special features to their products, whether in hardware or software, which permit a specific benchmark to execute quickly but which do not offer similar advantages to other, more general tasks.
Power consumption is another design criterion that factors in the design of modern computers. Power efficiency can often be traded for performance or cost benefits. the typical measurement in this case is MIPS/W (millions of instructions per watt).
With the increasing power density of modern circuits as the number of transistors per chip scales (Moore's law), power efficiency has increased in importance. Recent processor designs such as the Intel Core 2 put more emphasis on increasing power efficiency. Also, in the world of embedded computing, power efficiency has long been and remains an important goal next to throughput and latency.
|
|||||||||||