A message on the topic of computer processor. The main parameters of processors are: There are three types of cache memory

Abstract on computer science

"CPU"

I've done the work

Gulakov Philip

I checked the work

Kuyantseva L.M.

n. Friendship 2007

Contents___________ 1
Introduction ____________2
Processor __________3
Clock frequency, System bus, Multiplying factor _________4 - 5
Core type and production technology_________6
Differences Pentium processors and Celeron, Athlon and Duron __________ 7
AMD processors and their disadvantages____________8

Introduction

In this essay I will talk about what a microprocessor is, the history of the creation of a microprocessor, why it is needed, and how the processor of one company differs from another.

CPU

Microprocessor is the central device (or complex of devices) of a computer (or computing system), which performs arithmetic and logical operations specified by the information conversion program, controls the computing process and coordinates the operation of system devices (storage, sorting, input-output, data preparation, etc.). First micro Intel processor 4004 was created in 1971 by a team led by the talented inventor, Dr. Ted Hopf. Today his name stands alongside the names of the greatest inventors of all time...Initially, the 4004 processor was intended for... microcalculators and was manufactured to order from a Japanese company. Fortunately, this company went bankrupt, and as a result, the development became the property of Intel. From this moment on, the era of personal computers began. Today's Intel processors are more than ten thousand times faster than their progenitor! And any home computer has power and intelligence many times greater than the computer that controlled the flight of the Apollo spacecraft to the Moon. At first glance, the processor is simply grown by special technology silicon crystal (it’s not for nothing that it’s also called “stone”). However, this pebble contains many individual elements - transistors, which together give the computer the ability to “think”. More precisely, to calculate, performing certain mathematical operations with numbers into which any information entering the computer is transformed. There are many millions of such transistors in any microprocessor. Today's processor is not just a collection of transistors, but the whole system sets important devices. On any processor chip there are:

Processor functions:

Data processing by given program by performing arithmetic and logical operations;

Software control operation of computer devices

Control device (CU). Coordinates the operation of all other devices, performs device management functions, and manages computer calculations.

Arithmetic logic unit (ALU). This is the name of the device for integer operations. Arithmetic operations such as addition, multiplication and division, as well as logical operations (OR, AND, ASL, ROL, etc.) are processed using ALU. These operations make up the vast majority of code in most programs. All operations in the ALU are performed in registers - specially designated cells of the ALU. A processor can have multiple ALUs. Each is capable of performing arithmetic or logical operations independently of the others, allowing multiple operations to be performed simultaneously. An arithmetic logic unit performs arithmetic and logical operations. Logical operations are divided into two simple operations: "Yes" and "No" ("1" and "0"). Usually these two devices are distinguished purely conditionally; they are not structurally separated.

AGU (Address Generation Unit) - address generation device. This device is no less important than the ALU, because it is responsible for correct addressing when loading or saving data. Absolute addressing in programs is used only in rare exceptions. As soon as data sets are taken, program code Indirect addressing is used to make the AGU work.

Mathematical coprocessor (FPU). The processor may contain several mathematical coprocessors. Each of them is capable of performing at least one floating point operation regardless of what the other ALUs are doing. The pipelining method allows one math coprocessor to perform multiple operations simultaneously. The coprocessor supports high-precision calculations, both integer and floating point, and also contains a set of useful constants that speed up calculations. The coprocessor works in parallel with central processor, thus ensuring high performance. The system executes coprocessor instructions in the order in which they appear in the thread. Math coprocessor personal computer The IBM PC allows him to perform high-speed arithmetic and logarithmic operations, as well as trigonometric functions with high accuracy.

Instruction (command) decoder. Parses instructions to extract operands and addresses where results are located. This is followed by a message to another independent device about what needs to be done to carry out the instruction. The decoder allows multiple instructions to be executed simultaneously to load all execution devices.

Cache memory. Special high-speed processor memory. The cache is used as a buffer to speed up data exchange between the processor and RAM, and also for storing copies of instructions and data that have recently been used by the processor. Values from the cache are retrieved directly, without accessing main memory. When studying the features of the programs, it was discovered that they access certain areas of memory with different frequencies, namely: memory cells that the program accessed recently are most likely to be used again. Let's assume that the microprocessor is capable of storing copies of these instructions in its local memory. In this case, the processor will be able to use a copy of these instructions each time throughout the cycle. You will need access to memory at the very beginning. To store these instructions you need absolutely small volume memory. If instructions arrive to the processor quickly enough, then the microprocessor will not waste time waiting. This saves time on following instructions. But for the fastest microprocessors this is not enough. The solution to this problem is to improve memory organization. The memory inside the microprocessor can operate at the speed of the processor itself

First level cache (L1 cache). Cache memory located inside the processor. It is faster than all other types of memory, but smaller in size. Stores most recently used information that can be used when executing short program cycles.

Second level cache (L2 cache). Also located inside the processor. The information stored in it is used less frequently than the information stored in the first level cache, but it has more memory capacity. Also, processors currently use a third-level cache.

Main memory. Much larger in size than cache memory, and much slower. Multi-level cache memory allows you to reduce the performance requirements of the most powerful microprocessors dynamic memory. So, if you reduce main memory access time by 30%, then the performance of a well-designed cache memory will increase by only 10-15%. Cache memory, as is known, can have a significant impact on processor performance depending on the type of operations being performed, but increasing it does not necessarily bring an increase overall performance processor operation. It all depends on how optimized the application is for this structure and uses the cache, as well as whether various program segments fit into the cache entirely or in chunks.

Cache memory not only improves the performance of the microprocessor during memory read operations, but it can also store values written by the processor to main memory; These values can be written later, when the main memory is not occupied. This cache memory is called a cache with write back(write back cache). Its capabilities and principles of operation differ markedly from the characteristics of a write-through cache, which is involved only in read operations from memory.

A bus is a data transfer channel shared by different units of the system. The bus can be a set of conductive lines in a printed circuit board, wires soldered to the terminals of the connectors into which they are inserted printed circuit boards, or flat cable. Information is transmitted on the bus in the form of groups of bits. The bus may have a separate line for each bit of a word (parallel bus), or all bits of a word may use one line sequentially in time (serial bus). Many receiving devices - receivers - can be connected to the bus. Typically the data on the bus is destined for only one of them. The combination of control and address signals determines for whom exactly. The control logic drives special strobe signals to indicate to the receiver when it should receive data. Receivers and senders can be unidirectional (that is, they can only transmit or receive) or bidirectional (that is, they can do both). However, the fastest processor bus won't help much if the memory can't deliver data at the appropriate speed.

Tire types:

Data bus. Serves to transfer data between the processor and memory or the processor and I/O devices. This data can be both commands from the microprocessor and information that it sends to or receives from the I/O ports.
Address bus. Used by the CPU to select the desired memory location or I/O device by setting a specific address on the bus corresponding to one of the memory locations or one of the I/O elements included in the system.
Control bus. It transmits control signals intended for memory and input/output devices. These signals indicate the direction of data transfer (to or from the processor).

BTB (Branch Target Buffer) - branch target buffer. This table contains all the addresses to which a transition will or can be made. Athlon processors also use a branch history table (BHT - Branch History Table), which contains addresses at which branches have already been made.

State Budgetary Educational Institution of Higher Professional Education KemSMA of the Ministry of Health of Russia

Abstract on computer science on the topic:

"CPU"

I've done the work:

Budnikov Vladimir

Student of group 211

Kemerovo 2013

Processor………………………………………………………………………………………...2

Clock frequency, System bus, Multiplying factor …………………4-5

Core type and production technology……………………………………………………………..6-7

Differences between Pentium and Celeron, Athlon and Duron processors ………………………… 7

CPU

Microprocessor is a central device (or set of devices) of a computer (or computing system), which performs arithmetic and logical operations specified by an information conversion program, controls the computing process and coordinates the operation of system devices (storage, sorting, input-output, data preparation, etc. ). The first Intel4004 microprocessor was created in 1971 by a team led by the talented inventor, Dr. Ted Hopf. Today his name stands alongside the names of the greatest inventors of all time...Initially, the 4004 processor was intended for... microcalculators and was manufactured to order from a Japanese company. Fortunately, this company went bankrupt, and as a result, the development became the property of Intel. At this point, the era of personal computers began. Today's Intel processors are more than ten thousand times faster than their progenitor! And any home computer has the power and “smartness” many times greater than the computer that controlled the flight of the Apollo spacecraft to the Moon. At first glance, the processor is simply a silicon crystal grown using a special technology (it’s not for nothing that it is also called “stone”). However, this pebble contains many individual elements - transistors, which together give the computer the ability to “think”. More precisely, to calculate, performing certain mathematical operations with numbers into which any information entering the computer is transformed. There are many millions of such transistors in any microprocessor. Today's processor is not just a collection of transistors, but a whole system of many important devices.

Processor functions:

Processing data according to a given program by performing arithmetic and logical operations;

Software control of computer devices

Control device (CU). Coordinates the operation of all other devices, performs device management functions, and manages computer calculations.

AGU (Address Generation Unit) - address generation device. This device is no less important than the ALU, because it is responsible for correct addressing when loading or saving data. Absolute addressing in programs is used only in rare exceptions. As soon as data arrays are taken, indirect addressing is used in the program code, causing the AGU to work.

Mathematical coprocessor (FPU). The processor may contain several mathematical coprocessors. Each of them is capable of performing at least one floating point operation regardless of what the other ALUs are doing. The pipelining method allows one math coprocessor to perform multiple operations simultaneously. The coprocessor supports high-precision calculations, both integer and floating point, and also contains a set of useful constants that speed up calculations. The coprocessor works in parallel with the central processor, thus providing high performance. The system executes coprocessor instructions in the order in which they appear in the thread. Mathematical coprocessor of personal IBM computer The PC allows him to perform high-speed arithmetic and logarithmic operations, as well as trigonometric functions with high accuracy.

Cache memory. Special high-speed processor memory. The cache is used as a buffer to speed up communication between the processor and RAM, and to store copies of instructions and data that have recently been used by the processor. Values from the cache are retrieved directly, without accessing main memory. When studying the features of the programs, it was discovered that they access certain areas of memory with different frequencies, namely: memory cells that the program accessed recently are most likely to be used again. Let's assume that the microprocessor is capable of storing copies of these instructions in its local memory. In this case, the processor will be able to use a copy of these instructions each time throughout the cycle. You will need access to memory at the very beginning. A very small amount of memory is required to store these instructions. If instructions arrive to the processor quickly enough, then the microprocessor will not waste time waiting. This saves time on following instructions. But for the fastest microprocessors this is not enough. The solution to this problem is to improve memory organization. The memory inside the microprocessor can operate at the speed of the processor itself. First level cache (L1 cache). Cache memory located inside the processor. It is faster than all other types of memory, but smaller in size. Stores most recently used information that can be used when executing short program cycles.

Second level cache (L2cache). Also located inside the processor. The information stored in it is used less frequently than the information stored in the first level cache, but it has more memory capacity. Also, processors currently use a third-level cache.

Main memory. Much larger in size than cache memory, and much slower. Multi-level cache memory allows you to reduce the performance requirements of the most powerful microprocessors for the main dynamic memory. So, if you reduce main memory access time by 30%, then the performance of a well-designed cache memory will increase by only 10-15%. Cache memory, as is known, can have a significant impact on processor performance depending on the type of operations being performed, but increasing it will not necessarily increase the overall processor performance. It all depends on how optimized the application is for a given structure and uses the cache, and also on whether various program segments are cached entirely or in chunks.

Cache memory not only improves the performance of the microprocessor during memory read operations, but it can also store values written by the processor to main memory; These values can be written later, when the main memory is not occupied. This cache is called a write back cache. Its capabilities and principles of operation differ markedly from the characteristics of a write-through cache, which is involved only in read operations from memory.

A bus is a data transfer channel shared by different units of the system. The bus can be a set of conductive lines in a printed circuit board, wires soldered to the terminals of the connectors into which the printed circuit boards are inserted, or a flat cable. Information is transmitted on the bus in the form of groups of bits. The bus may have a separate line for each bit of a word (parallel bus), or all bits of a word may use one line sequentially in time (serial bus). Many receiving devices - receivers - can be connected to the bus. Typically the data on the bus is destined for only one of them. The combination of control and address signals determines for whom exactly. The control logic drives special strobe signals to indicate to the receiver when it should receive data. Receivers and senders can be unidirectional (that is, they can only transmit or receive) or bidirectional (that is, they can do both). However, the fastest processor bus won't help much if the memory can't deliver data at the appropriate speed.

Tire types:

Data bus. Serves to transfer data between the processor and memory or the processor and I/O devices. This data can be both commands from the microprocessor and information that it sends to or receives from the I/O ports.

Address bus. Used by the CPU to select the desired memory location or I/O device by setting a specific address on the bus corresponding to one of the memory locations or one of the I/O elements included in the system.

Control bus. It transmits control signals intended for memory and input/output devices. These signals indicate the direction of data transfer (to or from the processor).

Registers are inner memory processor. They represent a number of specialized additional memory cells, as well as internal storage media of the microprocessor. A register is a temporary storage device for data, numbers or instructions and is used to facilitate arithmetic, logical and transfer operations. Special electronic circuits can perform some manipulations on the contents of some registers. For example, "cut" certain parts of a command for later use or perform certain arithmetic operations over the numbers. The main element of the register is electronic circuit, called a flip-flop, which is capable of storing one binary digit (bit). A register is a collection of triggers connected to each other in a certain way common system management. There are several types of registers, differing in the type of operations performed.

Some important registers have their own names, for example:

adder - an ALU register involved in the execution of each operation.

command counter - register CU, the contents of which correspond to the address of the next executed command; serves for automatic selection of a program from successive memory cells.

command register - a control register for storing the command code for the period of time necessary for its execution. Some of its bits are used to store the operation code, the rest are used to store operand address codes.

Clock frequency.

Speed of work – of course, this is the indicator we pay attention to first! When we talk about processor speed, we mean its clock speed. This value, measured in megahertz (MHz), shows how many instructions the processor can execute within a second. The clock frequency is indicated by a number in the name of the processor (for example, Pentium4-2400, that is, a Pentium4 generation processor with a clock frequency of 2400 MHz or 2.4 GHz).

Clock frequency is undoubtedly the most important indicator processor speed. But far from the only one. How else can we explain the strange fact that Celeron, Athlon and Pentium4 processors operate at the same frequency... with at different speeds?

This is where new factors come into play.

Processor size

Bit capacity – maximum amount bits of information that can be processed and transmitted by the processor simultaneously.

Until recently, all processors were 32-bit (32-bit); this bit depth was reached 10 years ago. For a long time They couldn’t increase the bit depth because the programs were adapted for the old 32-bit platform. And since the buyer looks primarily at clock purity, manufacturers simply did not see the need for such a transition. AMD released the first 64-bit processor, Athlon64, in 2003.

Intel held out until the last until 2005. All Pentium4 processors were still 32-bit. Only in the middle of the year, when new models of the Pentium4 6xx series processor appeared on the market, the first ones had built-in support for 64-bit instructions.

Core type and production technology

The core is the processor chip itself, the part that is directly the “processor”. The crystal itself modern models It is small in size, and the dimensions of the finished processor increase greatly due to its packaging and wiring. The processor crystal can be seen, for example, in Athlon processors; in them it is not closed. P4 has everything top part hidden under a heat dissipator (which also performs a protective function; the crystal itself is not that strong). Processors based on different cores, this can be said different processors, they may differ in cache memory size, bus frequency, manufacturing technology, etc. In most cases, than newer kernel, those better processor. An example is P4, there are two cores - Willamette and Northwood. The first core was produced using 0.18µm technology and operated exclusively on a 400Mhz bus. The lowest models had a frequency of 1.3Ghz, the maximum frequencies for the core were slightly higher than 2.2Ghz. Northwood was later released. It was already made using 0.13 micron technology and supported a bus of 400 and 533 Mhz, and also had an increased cache memory capacity. The transition to a new kernel has significantly increased performance and maximum frequency work. Junior Northwood processors are overclocked well, but in fact the overclocking potential of these processors is based on a more “fine” technical process.

Differences between Pentium and Celeron, Athlon and Duron processors

The Celeron processor is a budget (stripped down) version of the corresponding (more productive, but also much more expensive) main-stream processor, based on the core of which it was created. U Celeron processors two or four times less L2 cache. They also have a lower system bus frequency compared to their corresponding “parents”. Compared to Athlon, Duron processors have 4 times less cache memory and lower system bus 200MHz (266MHz for Applebred), although there are also “full-fledged” Athlons with FSB 200MHz. In the near future, Durons on the Morgan core will completely disappear from sale - their production has already been curtailed quite a long time ago. They should be replaced by Durons on the Applebred core, which are nothing more than AthlonXP Thoroughbreds cut down in cache. Bartons cut down in cache have also already appeared 's, the core of which is called Thorton. The main characteristics of the processors can be seen in the table at the end of the abstract. There are tasks in which there is almost no difference between regular and cut-down processors, and in some cases the lag is quite serious. On average, when compared with uncut-down ones. processor of the same frequency, the lag is 10-30%. But cut-down processors tend to overclock better due to the smaller amount of cache memory and are cheaper, in short, if the price difference between a normal and a cut-down processor is significant, then it’s worth taking. stripped down Although it should be noted here that Celeron processors perform very poorly compared to full-fledged P4s - the lag in some situations reaches 50%. This does not apply to CeleronD processors, in which the second level cache is 256 KB (128 KB in regular Celerons) and the lag is no longer so terrible.

What is a processor?

CPU(from - process) - a device or program whose purpose is to process (process) something (object, process).

The processor is the main chip of the computer, its “brain”. It allows the execution of program code located in memory and controls the operation of all computer devices. The faster the processor speed, the faster the computer will perform. The processor has special cells called registers. It is in the registers that the commands that are executed by the processor are placed, as well as the data that the commands operate on. The processor's job is to select from memory to a certain sequence commands and data and their execution. This is what program execution is based on.

What parameters distinguish one processor from another. This is primarily the clock frequency, bit depth, operating voltage, internal clock frequency multiplication factor and Download and read What is a processor?

Biology Presentation - Coelenterates

In the presentation:
1.Where do coelenterates live?
2. How are coelenterates similar to sponges?
3.Why are they called that?
Download and read Presentation on Biology - Coelenterates

Presentation on biology, conditions for seed germination and the importance of the seed embryo

Biology Presentation - Conditions for seed germination

Abstract on computer science

"CPU"

I've done the work

Gulakov Philip

I checked the work

Kuyantseva L.M.

n. Friendship 2007

Contents___________ 1
Introduction ____________2
Processor __________3
Clock frequency, System bus, Multiplying factor _________4 - 5
Core type and production technology_________6
Differences between Pentium and Celeron, Athlon and Duron processors __________ 7
AMD processors and their disadvantages____________8

Introduction

In this essay I will talk about what a microprocessor is, the history of the creation of a microprocessor, why it is needed, and how the processor of one company differs from another.

CPU

Microprocessor is a central device (or set of devices) of a computer (or computing system), which performs arithmetic and logical operations specified by an information conversion program, controls the computing process and coordinates the operation of system devices (storage, sorting, input-output, data preparation, etc. ). The first Intel 4004 microprocessor was created in 1971 by a team led by the talented inventor, Dr. Ted Hopf. Today his name stands alongside the names of the greatest inventors of all time...Initially, the 4004 processor was intended for... microcalculators and was manufactured to order from a Japanese company. Fortunately, this company went bankrupt, and as a result, the development became the property of Intel. From this moment on, the era of personal computers began. Today's Intel processors are more than ten thousand times faster than their progenitor! And any home computer has the power and “smartness” many times greater than the computer that controlled the flight of the Apollo spacecraft to the Moon. At first glance, the processor is simply a silicon crystal grown using a special technology (it’s not for nothing that it is also called “stone”). However, this pebble contains many individual elements - transistors, which together give the computer the ability to “think”. More precisely, to calculate, performing certain mathematical operations with numbers into which any information entering the computer is transformed. There are many millions of such transistors in any microprocessor. Today's processor is not just a collection of transistors, but a whole system of many important devices. On any processor chip there are:

Processor functions:

Processing data according to a given program by performing arithmetic and logical operations;

Software control of computer devices

Control device (CU). Coordinates the operation of all other devices, performs device management functions, and manages computer calculations.

AGU (Address Generation Unit) - address generation device. This device is no less important than the ALU, because it is responsible for correct addressing when loading or saving data. Absolute addressing in programs is used only in rare exceptions. As soon as data arrays are taken, indirect addressing is used in the program code, causing the AGU to work.

Mathematical coprocessor (FPU). The processor may contain several mathematical coprocessors. Each of them is capable of performing at least one floating point operation regardless of what the other ALUs are doing. The pipelining method allows one math coprocessor to perform multiple operations simultaneously. The coprocessor supports high-precision calculations, both integer and floating point, and also contains a set of useful constants that speed up calculations. The coprocessor works in parallel with the central processor, thus providing high performance. The system executes coprocessor instructions in the order in which they appear in the thread. The mathematical coprocessor of the IBM PC personal computer allows it to perform high-speed arithmetic and logarithmic operations, as well as trigonometric functions with high accuracy.

Main memory. Much larger in size than cache memory, and much slower. Multi-level cache memory allows you to reduce the performance requirements of the most powerful microprocessors for the main dynamic memory. So, if you reduce main memory access time by 30%, then the performance of a well-designed cache memory will increase by only 10-15%. Cache memory, as is known, can have a significant impact on processor performance depending on the type of operations being performed, but increasing it will not necessarily increase the overall processor performance. It all depends on how optimized the application is for a given structure and uses the cache, and also on whether various program segments are cached entirely or in chunks.

Cache memory not only improves the performance of the microprocessor during memory read operations, but it can also store values written by the processor to main memory; These values can be written later, when the main memory is not occupied. This cache is called a write back cache. Its capabilities and principles of operation differ markedly from the characteristics of a write-through cache, which is involved only in read operations from memory.

A bus is a data transfer channel shared by different units of the system. The bus can be a set of conductive lines in a printed circuit board, wires soldered to the terminals of the connectors into which the printed circuit boards are inserted, or a flat cable. Information is transmitted on the bus in the form of groups of bits. The bus may have a separate line for each bit of a word (parallel bus), or all bits of a word may use one line sequentially in time (serial bus). Many receiving devices - receivers - can be connected to the bus. Typically the data on the bus is destined for only one of them. The combination of control and address signals determines for whom exactly. The control logic drives special strobe signals to indicate to the receiver when it should receive data. Receivers and senders can be unidirectional (that is, they can only transmit or receive) or bidirectional (that is, they can do both). However, the fastest processor bus won't help much if the memory can't deliver data at the appropriate speed.

Tire types:

Data bus. Serves to transfer data between the processor and memory or the processor and I/O devices. This data can be both commands from the microprocessor and information that it sends to or receives from the I/O ports.
Address bus. Used by the CPU to select the desired memory location or I/O device by setting a specific address on the bus corresponding to one of the memory locations or one of the I/O elements included in the system.
Control bus. It transmits control signals intended for memory and input/output devices. These signals indicate the direction of data transfer (to or from the processor).

Registers are the internal memory of the processor. They represent a number of specialized additional memory cells, as well as internal storage media of the microprocessor. A register is a temporary storage device for data, numbers or instructions and is used to facilitate arithmetic, logical and transfer operations. Special electronic circuits can perform some manipulations on the contents of some registers. For example, “cut” individual parts of a command for later use or perform certain arithmetic operations on numbers. The main element of a register is an electronic circuit called a flip-flop, which is capable of storing one binary digit(discharge). A register is a collection of triggers connected to each other in a certain way by a common control system. There are several types of registers, differing in the type of operations performed.

Some important registers have their own names, for example:

1. adder - an ALU register involved in the execution of each operation.

2. program counter - register CU, the contents of which correspond to the address of the next executed command; serves for automatic selection of a program from successive memory cells.

3. command register - a control register for storing the command code for the period of time necessary for its execution. Some of its bits are used to store the operation code, the rest are used to store operand address codes.

Clock frequency.

Speed of work – of course, this is the indicator we pay attention to first! When we talk about processor speed, we mean its clock speed. This value, measured in megahertz (MHz), shows how many instructions the processor can execute within a second. The clock frequency is indicated by a number in the name of the processor (for example, Pentium 4-2400, that is, a Pentium 4 generation processor with a clock frequency of 2400 MHz or 2.4 GHz).

Clock speed is undoubtedly the most important indicator of processor speed. But far from the only one. How else can we explain the strange fact that Celeron, Athlon and Pentium 4 processors operate at the same frequency... at different speeds?

This is where new factors come into play.

Processor size

Bit capacity is the maximum number of bits of information that can be processed and transmitted by the processor simultaneously.

Until recently, all processors were 32-bit (32-bit); this bit depth was reached 10 years ago. For a long time they could not increase the bit depth due to the fact that the programs were adapted for the old 32-bit platform. And since the buyer looks primarily at clock purity, manufacturers simply did not see the need for such a transition. AMD released the first 64-bit Athlon 64 processor in 2003.

Intel held out until the last until 2005. All Pentium 4 processors were still 32-bit. Only in the middle of the year, when new models of the Pentium 4 6xx series processor appeared on the market, the first ones had built-in support for 64-bit instructions.

Core type and production technology

The core is the processor chip itself, the part that is directly the “processor”. The crystal itself in modern models is small in size, and the dimensions of the finished processor increase very much due to its packaging and wiring. The processor crystal can be seen, for example, in Athlon processors; in them it is not closed. In P4, the entire upper part is hidden under a heat dissipator (which also performs a protective function; the crystal itself is not that strong). Processors based on different cores are, one might say, different processors; they may differ in cache memory size, bus frequency, manufacturing technology, etc. In most cases, the newer the core, the better the processor. An example is P4, there are two cores - Willamette and Northwood. The first core was produced using 0.18µm technology and operated exclusively on a 400Mhz bus. The lowest models had a frequency of 1.3Ghz, the maximum frequencies for the core were slightly higher than 2.2Ghz. Northwood was later released. It was already made using 0.13 micron technology and supported a bus of 400 and 533 Mhz, and also had an increased cache memory capacity. The transition to a new core has significantly increased performance and maximum operating frequency. Junior Northwood processors can be overclocked well, but in fact the overclocking potential of these processors is based on a more “fine” technical process.

Differences between Pentium and Celeron, Athlon and Duron processors

The Celeron processor is a budget (stripped down) version of the corresponding (more productive, but also much more expensive) main-stream processor, based on the core of which it was created. Celeron processors have two to four times less L2 cache. They also have a lower system bus frequency compared to their corresponding “parents”. Duron processors, compared to Athlon, have 4 times less cache memory and a lower system bus of 200MHz (266MHz for Applebred), although there are also “full-fledged” Athlons with a 200MHz FSB. In the near future, Durons on the Morgan core will completely disappear from sale - their production has already been curtailed quite a long time ago. They should be replaced by Durons on the Applebred core, which are nothing more than AthlonXP Thoroughbreds cut down in cache. Bartons cut down in cache have also already appeared 's, the core of which is called Thorton. The main characteristics of the processors can be seen in the table at the end of the abstract. There are tasks in which there is almost no difference between regular and cut-down processors, and in some cases the lag is quite serious. On average, when compared with uncut-down ones. processor of the same frequency, the lag is 10-30%. But cut-down processors tend to overclock better due to the smaller amount of cache memory and are cheaper, in short, if the price difference between a normal and a cut-down processor is significant, then it’s worth taking. stripped down Although it should be noted here that Celeron processors perform very poorly compared to full-fledged P4s - the lag in some situations reaches 50%. This does not apply to processors Celeron D, in which the second level cache is 256 KB (128 KB in regular Celerons) and the lag is no longer so terrible.

AMD processors

Firstly, with AXP (and Athlon 64) a rating is written instead of frequency, i.e. for example, a 2000+ processor actually operates at a frequency of 1667Mhz, but in terms of operating efficiency it corresponds to Athlon (Thunderbird) 2000Mhz. Temperature has recently been considered the main drawback. But latest models(on Thoroughbred, Barton cores, etc.) the heat dissipation is comparable to the Pentium 4, but the latest, at the time of writing, models from Intel (P4 Extreme Edition) sometimes get warmer and much more. In terms of reliability, processors are now also not much inferior to P4; although they cannot skip cycles (run “idle”) when overheated, they have acquired a built-in thermal sensor (although it appeared in the Palomino core, very few modern motherboards can take readings from this temperature sensor). It should be noted here that Athlon XP on the Barton core acquired similar function BusDisconnect - it “disconnects” the processor from the bus during idle cycles (idle), but it is virtually powerless when overheated due to increased load - here all “responsibility” is shifted to thermal control motherboard. Although the “strength” of the crystal (maximum permissible pressure limits) has increased, due to the reduced core area it has actually remained the same. Therefore, the probability of burning/damaging the crystal, although it has become less, still exists. But the Athlon 64 finally had the processor chip hidden under a heat spreader, so it would be extremely difficult to damage it. All "glitches" attributed to AMD are often the result of uninstalled or incorrectly installed universal drivers for VIA chipsets (VIA 4 in 1 Service Pack) or chipset drivers from other manufacturers (AMD, SIS, ALi). Atholn XP and Pentium 4 processors work in different applications very differently. For example, in complex mathematical calculations(3D modeling, specialized math packages), archiving, encoding in MPEG4, P4 often “beats” AXP. But there are a number of programs that work better with AXP. Basically these are games. For the average user (playing games), it is worth focusing on them, since recoding in any case takes a lot of time, and games, on the contrary, need to carry out all the calculations as quickly as possible. AXP Barton processors with a 400Mhz bus and fundamentally new hybrid processors (32 and 64) have already been released. bit processor"in one bottle") K8.

How good are VIA C3 processors?

Their only advantage is low heat generation. Their power dissipation is 5-20 Watts versus 40-60 (on average) for AXP and P4. C3 are compatible with the outdated (according to Intel) Socket 370, although not with all boards; for example, the new Nehemiah core requires support for Tualatin on the board. In terms of speed, they are very much inferior (up to 50%, sometimes even more) to those with similar frequencies Intel and AMD processors. Even some improvements like SSE support They weren't given anything special. There are almost no such processors on sale and I don’t regret it at all :). If you need a quiet machine (such a processor often only needs a heatsink), but speed is not important, then you can take it. Theoretically, they should overclock quite well (the manufacturing technology is quite advanced), but in practice this is not observed - this is due to the small “safety margin” and ineffective core design.

Hyper Threading.

This technology designed to increase processor efficiency. Intel estimates that most of the time, only 30% of all execution units in the processor are running. Therefore, the idea arose to somehow use the remaining 70% (as you already know, the Pentium 4, which uses this technology, does not at all suffer from excess performance per megahertz). The essence of Hyper Threading is that while one thread is running, programs that are idle actuators can move on to the execution of another “thread” of the program. That is, it turns out something like dividing one physical processor into two virtual ones. Situations are also possible when attempts to simultaneously execute several “threads” will lead to a noticeable drop in performance. For example, because the L2 cache size is quite small, active threads will try to load the cache. It is possible that the fight for the cache will lead to constant clearing and reloading of data in it (therefore, the speed will drop). To use this technology, just one processor with Hyper support Threading is not enough, you need support from the motherboard (chipset). It is very important to remember that there is currently a lack of normal support for this technology from operating systems and, most importantly, the need to recompile, and in some cases change the algorithm, applications so that they can fully take advantage of Hyper Threading. Tests prove this, often there is no increase in speed, sometimes there is even a slight drop in performance. Although there are already a number of applications in which, thanks to optimizations for HT, there is a strong increase in speed. Let's see what will happen next.

Recently new processors of the K8 family appeared and P4 came out “in response”

Extreme Edition (EE), what can we say about them?

P4 EE is essentially a server version of P4 (Xeon on the Gallatin core, “packed” in mPGA478), has all the advantages of regular P4 with 800Mhz FSB, plus 2Mb L3 cache. Athlon 64 supports 32/64-bit computing, has a 1Mb L2 cache, SSE2 support, a built-in controller for initially single-channel, later dual-channel DDR400 and a 200MHz real FSB frequency. Note that the FSB frequency in Athlon 64 systems has a purely formal meaning: in fact, it is simply the frequency of the signal relative to which the operating CPU frequencies and other system components. The Athlon 64 FX is derived from the Operton server processor, and differs from the Athlon 64 in that it is equipped with a dual-channel buffered (registred) DDR400 controller. The general trend is this: the Athlon 64 3200+ loses to the P4 3200Mhz by about 5% in terms of performance on average, although it should be taken into account that the actual processor frequency is about 2Ghz, it turns out that a 2Ghz processor is more than a match for a 3.2Ghz processor! Top on this moment The P4 EE and Athlon 64 FX processors are on par if you average the test results. And if we compare the Athlon 64 3200+ with the regular Athlon 3200+, then the first one is almost always (with the exception of mp3 encoding :)) faster by 10-40%. And now a little about 64-bit. At the moment, the Athlon 64 has practically no use for its support, real applications There are almost none suitable for use by ordinary users. Microsoft is about to release a 64-bit OS suitable for ordinary users. Existing 64-bit Linux in in this case doesn't fit. The most unpleasant thing is that all applications will also require improvements to use all the “power” of the new processors.

In fact, what we call a processor today is correctly called a microprocessor. There is a difference and is determined by the type of device and its historical development.

The first processor (Intel 4004) appeared in 1971 year.

Externally, it is a silicon wafer with millions and billions (today) of transistors and channels for passing signals.

The purpose of the processor is automatic execution programs. In other words, it is the main component of any computer.

Processor device

The key components of the processor are arithmetic logic unit(ALU), registers And control device. ALUs will perform basic mathematical and logical operations. All calculations are made in binary system Reckoning. The control device determines the consistency of the operation of parts of the processor itself and its communication with other (external) devices. The registers temporarily store the current instruction, initial, intermediate and final data (the result of ALU calculations). The size of all registers is the same.

Data and command cache stores frequently used data and commands. Access to the cache is much faster than access to RAM, so the larger it is, the better.

Processor circuit

Processor operation

The processor operates under the control of a program located in RAM.

(The operation of a processor is more complex than is shown in the diagram above. For example, data and commands do not enter the cache immediately from RAM, but through a prefetch unit, which is not shown in the diagram. Also not shown is the decoding unit, which converts data and commands into binary form, only after which the processor can work with them.)

The control unit, among other things, is responsible for calling the next command and determining its type.

The arithmetic-logical unit, having received the data and command, performs the specified operation and writes the result to one of the free registers.

The current team is in a specially designated area command register. In the process of working with the current team, the value of the so-called program counter, which now points to next command(unless, of course, there was a jump or stop command).

Often a command is represented as a structure consisting of a record of the operation (to be performed) and the addresses of the source data cells and the result. At the addresses specified in the command, data is taken and placed in ordinary registers (in the sense not in the command register), the resulting result also first appears in the register, and only then moves to its address specified in the command.

Processor Specifications

Clock frequency The processor today is measured in gigahertz (GHz), previously measured in megahertz (MHz). 1 MHz = 1 million clock cycles per second.

The processor “communicates” with other devices (RAM) using data, address and control buses. The bus width is always a multiple of 8 (it’s clear why if we are dealing with bytes), and is variable during historical development computer equipment and different for different models, and is also not the same for the data bus and the address bus.