Introduction Introduction
The processor (CPU by Central Processing Unit or Central Processing Unit), is to put it in some way, the computer's brain. Allows the processing of numeric data, meaning information entered in binary format and the execution of instructions stored in memory.
The first microprocessor (Intel 4004) was invented in 1971. It was a calculation device 4 bits at a speed of 108 kHz. Since then, the power of microprocessors has increased exponentially. What exactly are these little pieces of silicone that operate a computer?
Operation
The processor (CPU called for Central Processing Unit) is an electronic circuit that operates at the speed of an internal clock thanks to a quartz crystal, subjected to an electric current, sends pulses, called " peaks." The clock speed (also called cycle) corresponds to the number of pulses per second, expressed in Hertz (Hz). Thus, a computer has a 200 MHz clock sends pulses per second 200,000,000. Usually, the clock frequency is a multiple of the system frequency (FSB , Front-Side Bus Bus or front), ie a multiple of the frequency of the motherboard .
With each clock peak, the processor performs an action that in turn corresponds to an instruction or a part of it. The measure CPI (Cycles Per Instruction or Cycles per Instruction) represents the average number of clock cycles required for the microprocessor to execute an instruction. Consequently, the power of the microprocessor may be characterized by the number of instructions per second which is capable of processing. The MIPS (millions of instructions' per second or millions of instructions per second) are the units used, and correspond to the processor's frequency divided by the number of ICC .
Instructions
A statement is an elementary operation that the processor can meet .. The instructions are stored in main memory, waiting be treated by the processor. The instructions have two fields:
- the operation code that represents the action that the processor must run;
- the operating code, which defines the parameters of the action. The code is in turn running the operation. It can be both information and a memory address.
| | Operation Code Field Operation |
The number of bits in an instruction varies according to the type of information (between 1 and 4 bytes of 8 bits).
The instructions can be grouped into different categories. Here are some of the most important
- Access Memory: Memory access or transfer information between registers.
- Arithmetic: operations such as addition, subtraction, division or multiplication.
- Logical Operations: operations such as AND, OR, NOT, NOT EXCLUSIVELY, etc.
- Control: controls sequence, conditional connections, etc.
Records
When the processor executes instructions, the information stored temporarily in small local memory locations 8, 16, 32 or 64 bits, called records. Depending on the type of processor, the total number of records can vary from 10 to several hundred.
most important records are:
- the accumulator (ACC ), which stores the results of arithmetic and logic operations;
- the status register (PSW , Processor Status: Word or Word Processor State), which contains the system status indicators (leading digit, overflow, etc. .)
- the instruction register ( RI), which contains the instruction currently being processed;
- the ordinal counter (OC or by PC Program Counter , Program Counter), which contains the address of the next instruction process;
- the registration buffer that temporarily stores information from memory.
Cache
The cache (also buffer memory) is a fast memory that reduces waiting times of the various information stored in RAM (Random AccessMemory or Random Access Memory). Indeed, the main memory computer is slower than the processor. There are, however, memory types are much faster, but have a higher cost. The solution is therefore to include this type of local memory near the processor and temporarily store the primary data to be processed in it. The latest computer models have many different levels of cache:
- The Level 1 Cache (L1 Cache called for Cache Level 1) is integrated directly into the processor. Is divided into two parts:
- the first part is the instruction cache containing instructions that were decoded RAM during its passage through the pipes.
- second parte es la caché de información, que contiene información de la RAM, así como información utilizada recientemente durante el funcionamiento del procesador.
El tiempo de espera para acceder a las memorias caché nivel 1 es muy breve; es similar al de los registros internos del procesador. - La memoria caché nivel 2 (denominada L2 Cache , por Level 2 Cache ) se encuentra ubicada en la carcasa junto con el procesador (en el chip). La caché nivel 2 es un intermediario entre el procesador con su caché interna y la RAM. Se puede acceder más rápidamente que a la RAM, but not enough to level 1 cache.
- The Level 3 cache (L3 Cache called for Cache Level 3) is located on the motherboard.
All these levels of cache reduces the latency of various types of memory processing or transfer information. While the processor is running, the driver of the level 1 cache controller can interface with the Level 2 cache, in order to transfer information without impeding the operation of the processor. Also, the level 2 cache interfaces with the RAM (cache level 3) to allow transfers without impeding the normal functioning of the processor. Control Signals
The control signals are electronic signals that orchestrate the various processor units involved in the execution of an instruction. These signals are sent using an element called sequencer. For example, the signal Read / Write allows memory learn that the processor wants to read or write.
Functional Units
The processor consists of a group of interrelated units (or units of control). Although microprocessor architecture varies considerably from one design to another, the main elements of the microprocessor are:
- A control unit that links the incoming data, decodes it and send it to the unit of implementation: control unit consists of the following elements:
- sequencer (or monitor and logic unit ) that synchronizes the execution of the instruction clock speed. It also sends control signals
- ordinal counter, which contains the address of the instruction currently being executed;
- instruction register, which contains the following statement.
- A execution unit (or processing unit), which carries out the tasks assigned to the unit of instruction. The execution unit consists of the following elements:
- the arithmetic logic unit (ALU is written) is for the implementation of basic arithmetic and logic functions (AND, OR, XOR, etc.).
- the floating point unit (FPU written) that performs complex calculations partial arithmetic logic unit can not be done
- the status register;
- the accumulator.
- A management unit bus (or drive O) that manages the flow of incoming and outgoing information, and is interconnected with the system RAM ;
The diagram below provides a simplified representation of the elements of the processor (the physical layout of the elements is different from the layout):
Transistor
order to process the information, the microprocessor has a set of instructions, called instruction set ", made possible by electronic circuits. More precisely, the set of instructions is performed with the aid of semiconductors, small "switches de circuito" que utilizan el efecto transistor , descubierto en 1947 por John Barden , Walter H. Brattain y William Shockley , quienes recibieron por ello el premio Nobel en 1956.
Un transistor (contracción de los términos transferencia y resistor ) es un componente electrónico semi-conductor que posee tres electrodos capaces de modificar la corriente que pasa a través suyo, utilizando uno de estos electrodos (denominado electrodo de control). Éstos reciben el nombre de "componentes activos", en contraste a los "componentes pasivos", tales como la resistance or capacitors, with only two electrodes (which is called "bipolar").
The MOS transistor ( metal, oxide, silicon ) is the most common type of transistor used in integrated circuit design. MOS transistors have two negatively charged areas, respectively called source (with near-zero overhead), and drainage (with a load of 5V), separated by a positively charged region, called substrate . The substrate has an overlay control electrode, called door, allowing the load applied to the substrate.
When a voltage is applied to the control electrode, the positively charged substrate acts as a barrier and prevents the movement of electrons from source to drain. However, when the load is applied to the gate, the positive charges of the substrate are repelled and opening a communication channel with a negative charge between the source and drain.
The transistor then acts as a programmable switch, thanks to the control electrode. When a load is applied to the control electrode, it acts as a switch closed and no load, it acts as a switch open.
Integrated Circuits
Once combined, the transistors can make logic circuits which, when combined, form processors. The first integrated circuit dates back to 1958 and was built by Texas Instruments .
MOS transistors are therefore made of silicon wafers (wafers called ), obtained after multiple processes. These silicon wafers cut into rectangular elements to form a "circuit ." The circuits are then placed in cases with input / output connectors, and the sum of these parts makes a " integrated circuit." The extent of the engraving, expressed in microns (micrometers, microns writes ) defines the number of transistors per unit area. There may be millions of transistors on a single processor.
The Moore's Law, written in 1965 by Gordon E. Moore, cofounder of Intel, predicted that the processor performance (by extension the number of transistors integrated in the silicone) would double every 12 months. This law was revised in 1975, and changed the number of months to 18. Moore's Law is still valid today.
Since the rectangular enclosure containing O pins that look like legs, France uses the term "electronic flea " to refer to integrated circuits.
Families
Each type of processor has its own set of instructions. The processors are grouped into the following families, according to their unique instruction sets:
- 80x86: the "x" represents the family. Referred to 386, 486, 586, 686, etc. ARM
-
- IA-64 MIPS
-
- Motorola 6800 PowerPC
-
- SPARC
- ...
This explains why a program produced for a specific type of processor can only work directly on a system with another type of processor if it performs what is called instruction translation or emulation . The term " emulator is used to refer the program that performs the translation.
Instruction Set
A instruction set is the sum of the basic operations can play a processor. The instruction set of a processor is a factor in the architecture of it, but the same architecture can lead to different implementations by different manufacturers.
The processor runs efficiently thanks to a limited number of instructions, permanently connected to the electronic circuits. Most operations can be performed using functions basic. Some architectures, however, do include advanced signal processing functions.
CISC Architecture Architecture
CISC (Complex Instruction Set Computer , Computer Complex Instruction Set) refers to the permanent connection with the instructions of the processor complex, difficult to create from basic instructions.
CISC architecture is especially popular type 80x86 processors. This type of architecture is costly because of the advanced features printed on the silicone.
The instructions are of varying length, and sometimes require more than one clock cycle. Since the processors based on the CISC architecture can only process one instruction at a time, processing time is a function of the size of the instruction.
Architecture RISC processors technology RISC (Reduced Instruction Set Computer , Instruction Set Computer Reduced) do not have advanced features connected permanently.
is why programs must be translated into simple instructions, which complicates the development or necessary to use a more powerful processor. This type of arquitecturaprocesadores CISC. In addition, instructions for simple nature run in one clock cycle, accelerating the implementation of the program when compared to CISC processors. In conclusion, these processors can handle multiple instructions simultaneously, processing them in parallel. production cost is reduced when compared with Technological Improvements
Over time, manufacturers of microprocessors (called founders ) have developed a number of enhancements that optimize the performance of the processor.
Parallel Processing
The parallel processing is the simultaneous execution of instructions from the same program on different processors. Involves dividing the program into multiple processes handled in parallel to reduce execution time.
However, this type of technology requires synchronization and communication between the various processes, similar to what can happen when you break a task into a Company: work is distributed in smaller batch processes that are managed by different departments. Running a business can be affected greatly if the communication between the various internal services not working properly.
Channeling
is called channeling to technology to improve speed of execution of instructions by placing the various stages in parallel.
To understand the mechanism of channeling, you must first understand the stages of instruction execution. The stages of execution of an instruction for a processor with channeling "classic" 5 steps are:
- RECOVERY: ( retrieves the instruction cache;
- DECODING: decodes the instruction and look operands (register or immediate values);
- PERFORMANCE: executes instruction (for example, in the case of an ADD instruction, performing a sum, if an instruction SUB, performing a subtraction, etc.).
- MEMORY : access memory, and writes or retrieves data from there;
- POST SCRIPT (remove) : records the value in a register.
The instructions are arranged in lines in memory and are loaded one after another.
Thanks to the pipeline, the instruction processing requires only five steps above. Since the order of steps is invariable (RECOVERY, DECODING, PERFORMANCE, MEMORY, POST SCRIPT), you can create dedicated circuits for each of them in the processor.
The aim of the pipeline is executing each step in parallel with the preceding and following steps, which involves reading the instruction (RECOVERY) while reading the previous step (decoding), the time step above is being executed (execution) while the previous step is being written in the memory (MEMORY), and the first step in the series is recorded in a register (POST SCRIPT).
In general, should be planned 1 or 2 clock cycles (more rarely) for each step in pipeline, or a maximum of 10 clock cycles per instruction. For two statements, it takes a maximum of 12 clock cycles (10 +2 = 12 instead of 10 * 2 = 20) since the previous statement was already in the pipeline. Both instructions are processed simultaneously, albeit with a delay of 1 or 2 clock cycles. For 3 instructions will need 14 clock cycles, etc.
The principle of the pipeline can be likened to an automotive assembly line. The car moves from one workstation to the other along the line assembly and when it leaves the factory, it is completely finished. To fully understand the principle must be viewed the assembly line as a whole, and not the vehicle by vehicle. It takes three hours to produce each vehicle, but in reality there is one per minute.
should be noted that there are many different types of pipes, with amounts ranging from 2 to 40 steps, but the principle remains the same.
Superscaling
Superscaling technology is to place multiple parallel processing units to to process multiple instructions per cycle.
HyperThreading technology HyperThreading (HT is written) is to place two logical processors with a physical processor. The system recognizes and two physical processors and behaves as a multitasking system, thus sending two simultaneous threads called SMT (Simultaneous Multi Threading , Simultaneous multiprocessing). This "cheating", as it somehow allows better use of processor resources, ensuring the mass mailing of information it.
Last updated on Friday, April 24, 2009, 12:27:17
