Showing posts with label memory. Show all posts
Showing posts with label memory. Show all posts

Thursday, October 23, 2008

Intel cache Evolutions


Problem

1. External memory slower than the system bus.

  • Solution : Add external cache using faster memory technology.
  • Processor on which feature first appears : 386
2. Increased processor speed results in external bus becoming a bottleneck for cache access.
  • Solution : Move external cache on-chip, operating at the same speed as the processor.
  • Processor on which feature first appears : 486
3. Internal cache is rather small, due to limited space on chip.
  • Solution : Add external L2 cache using faster technology than main memory.
  • Processor on which feature first appears : 486
4. Contention occurs when both the Instruction Prefetcher and the Execution Unit simultaneously require access to the cache. In that case, the Prefetcher is stalled while the Execution Unit’s data access takes place.
  • Solution : Create separate data and instruction caches.
  • Processor on which feature first appears : Pentium
5. Increased processor speed results in external bus becoming a bottleneck for L2 cache access.
  • Solution : - Create separate back-side bus that runs at higher speed than the main (front-side) external bus. The BSB is dedicated to the L2 cache.
  • Processor on which feature first appears : Pentium Pro
-Move L2 cache on to the processor chip.
  • Processor on which feature first appears : Pentium II
6. Some applications deal with massive databases and must have rapid access to large amounts
of data. The on-chip caches are too small.
  • Solution : - Add external L3 cache.
  • Processor on which feature first appears : Pentium III
- Move L3 cache on-chip.
  • Processor on which feature first appears : Pentium 4

Wednesday, October 22, 2008

CPU Functions

CPU functions

  • Running programs - programs that are stored in the main way to take instruction - instruction, testing and instruction exsecution one by one according to the groove.
  • The view most simple program execution process is to take the processing instruction that consists of `two steps, namely:
* Operation reading the instructions (Fetch)
* Operation of instruction (execute)

Instruction cycle
  • Consists of Fetch cycle and the cycle execution

Fetch cycle - execution
  • At each cycle instruction, the CPU will initially read instructions from memory.
  • There are registers in the CPU, which functions control and calculate the next instruction, called the Program Counter (PC)
  • PC will add one calculation each time the CPU to read the instructions.
  • Instructions - read the instructions that will be made in the Instruction Register (IR).
  • Instructions - in the form of this instruction code - the code binner the interpretation can be done by the CPU and the action required.

The CPU Action
  1. CPU - Memory, the movement of data from the CPU to memory and vice versa.
  2. CPU and I / O, the movement of data from the CPU to the module I / O and vice versa.
  3. Data processing, the CPU establish a number of arithmetic and logic operations of the data.
  4. Control, is the instruction to control functions or work. For instance affairs conversion instruction execution.

Details cycle execution
  • Instruction Address Calculation (IAC), namely Calculation or determine the address of the next instruction that will be executed.
  • Involving the number of permanent additions to address previous instructions.
>> For example, if the length of each 16 bit instructions have
a long memory when 8 bits, then add 2 to address previously.

Cycle execution
  • Cycle for an instruction execution can involve more than a reference to the memory.
  • In addition, an instruction can determine an operation I / O.
  • Instruction Fetch (IF), read the instructions from the authorities or the location of memory to the CPU.
  • Instruction Operation Decoding (IOD), which analyzes the instructions to determine the type of operations that will be formed and operand that will be used.
  • Operand Address Calculation (OAC), which determine the address operand, this is done when the reference operand involved in memory.
  • Operand Fetch (OF), is taking operand from memory or from the module I / O.
  • Data Operation (DO), which is formed in which the operating instructions.
  • Operand store (OS), the results of execution to save the memory.

Figure cycle instructions



Function Interruptions
  • Mechanism termination or transfer of the processing instructions in the CPU to the routine interruptions.
  • Almost all modules (memory and I / O) have a mechanism that can work to interrupt CPU.

Goals Interrupts
  • In general, for management execution routine instructions to effective and efficient inter-module and the CPU - the module I / O and memory.
  • Each component of the computer can run tasks simultaneously, but the control is located on the CPU in addition to the speed of execution - each module is different.
  • Can work as the synchronization between the modules

Kind of class signal interruptions
  • Program, the interruptions are raised with several conditions that occur on the results of program execution.
* Example: aromatic overflow, the division of zero, operation illegal.
  • Timer, is the interruption of the timing raised in the processor.
* Signals allows operating systems to run certain functions regularly.

Kind of class signal interruptions
  • I / O, signal interruptions that are raised by the module I / O notification in respect of conditions of error and a resolution of the operation.
  • Hardware failure, is raised by the interruption of power failure or memory parity errors.

The process Interrupt
  • With the mechanism interruptions, the processor can be used for executing other instructions.
  • When a module is finished and ready to run the duty to receive the next assignment, the module will send a request interruptions to the processor.

The process Interrupt
  • Then, the processor will stop the execution of run to manhandle routine interruptions.
  • Once the program is completed, the interruption processor will continue the program execution again.
  • When interruptions received signal processor there are two possible actions, namely interruptions accepted or rejected suspended and interruptions.

Interruptions Suspended
What do processors?
  • Processors suspend the execution of programs that run and save the context. This is the address of store instructions that will be exercised and other relevant data.
  • Processors set the program counter (PC) to address early routine Interrupt handlers.

Cycle execution by the processor with the function of interruptions


Operating system complex
  • Interruption double (multiple Interrupt).
* For example, a computer will receive the request interruptions
during the printing process is completed with
the printer, in addition to the possible channels
of communication will submit a request
each time the interruption of data arrived.
  • Can be taken two approaches to handle this double interruption.

Interrupts dual approach
There are 2 approaches:
  • This approach is called sequential processing interruptions / sequential.
* Reject or other interruptions do not allow the interruption of a processor handled.
* Once completed processor handles the interruption of a new handled other
interruptions.

Interrupts dual approach
  • Processing interruption nest that is defining priorities for the interruption.
* Interrupt Handler: allow interruptions priorities higher handled first.

Multiple Interrupts - Sequential


Multiple Interrupts - Nested


Case
  • A system has three devices I / O: printers, disk, and communication channels, each - their priorities 2, 4 and 5. How will the process interruptions?
  • At the beginning of the printing system with a printer, when there is data on the delivery channel of communication so that communication module to request interruptions.
  • The transfer of the execution of interruption of communication modules, the printer interruptions while suspended.
  • When execution communication module, the interruption disk, but a lower priority because of the interruption suspended disk.
  • After the interruption of communication modules will be completed by interruptions that have higher priority, namely the disk.
  • When the disk is finished interruption, the execution interruptions printer. `Then the execution of the main program.

CPU (Central Processing Unit)

CPU

  • Central Processing Unit.
  • It is the most important components of the computer system.
  • Components of data processing based on the instructions given to him.
  • In creating the functions and tasks, the CPU composed of several components.

Main components of the CPU
  • Arithmetic and Logic Unit (pestle)
  • Control Unit
  • Registers
  • CPU Interconnections

CPU Chart



Arithmetic and Logic Unit (pestle)
  • The task of shaping functions - data processing functions of the computer.
  • Pestle is often called the machine language because the instructions do this - the machine language instructions given to them.
  • Pestle consists of two parts, namely the unit and the unit arithmetic Boolean logic, that each - each have the task specification work.

Control Unit
  • Duty to control the operation of the CPU and the overall control the computer so that the synchronization of work between the components in the run function - a function of operations.
  • Included in the responsibilities of the unit is to take control instructions - instructions from main memory and determine the type of instruction.

Registers
  • Media storage internal CPU that is used when processing data.
  • Memory is temporary, usually in use to store data when the data is processed or for further processing.

CPU Interconnections
  • System connection and a bus connecting the internal components and a bus - the bus external CPU.
  • CPU internal components, namely pestle, the control unit and the register - the register.
  • Component external CPU: other system, such as main memory, a tool inputs / outputs.

CPU
internal components



Detail the structure of the internal CPU

Tuesday, October 21, 2008

Computer Evolution and Performance Part 14


History of the Intel Processor

- 8080 (1972)

  • Intel's first microprocessor.
  • 8 bit machines and data to the memory bus, 8 bits.
  • 66 with the number of instructions ability 16KB.
- 8086 (1974)
  • 16 bit microprocessor technology with the instruction cache.
  • The number of instructions reached 111 and the ability to pengalamatan memory 64KB.
- 80,286 (1982)
  • Development from 8086.
  • Pengalamatan ability to reach 133 with 1MB instructions.
- 80,386 (1985)
  • 32 bit machine.
  • The system supports multitasking.
  • Best at that time.
- 80,486 (1989)
  • Progress in the cache memory technology and instruction pipelining.
  • It is equipped with a math co-processor.
- Pentium (1993)
  • Use technology that allows superscalar execution of instructions in parallel.
- Pentium Pro (1995)
  • Increase superscalar organization parallel to the process,
  • Found the system branch prediction, data flow analysis and cache memory system that increasingly sophisticated.
- Pentium II (1997)
  • MMX technology, capable of handling multimedia needs.
  • Commercial Pentium II has been using RISC technology.
- Pentium III
  • Floating point instructions ability to handle 3D graphics.
- Pentium IV
  • Floating point capabilities and increasingly sophisticated multimedia.
- Itanium
  • 2 units have the ability to floating point, integer 4 units, 3 units furcation, Internet streaming, 128 interger register.

Computer Evolution and Performance Part 13


Organization and Architecture Chip

- Increase the speed processor.
  • In carried out primarily by reducing the size of "Logic gate."
>> "Gate" more, more meetings, the speed increases.
>> Processing time can be shortened.
- Addition of the cache size and speed.
  • It is part of the processor chip.
>> Cache to reduce the access time is significant.
- Changing the organization and processor architecture.
  • Improving the effectiveness and speed of execution.
  • Processing in parallel.

Intel's Microprocessor Performance


The increase in capacity Cache
- Consists of two or three levels of cache between the processor and main memory.
- The increase in chip density.
  • The more cache memory in the chip.
>> Faster access the cache.
- In the Pentium 10% allocated to the cache.
- In the Pentium 4 is allocated 50%.

New Approach - Multiple Cores
- Multiple processors on one chip.
  • Large shared cache
- With one processor, proportionate increase performance straight with the increase of complexity.
- If the software can utilize multiple processor, the processor can mean two performance fold.
- So, take two simple processor in the chip better than one processor that is more complex.
- With two processors, cache it bigger.
  • Consumption of power in the logical memory is smaller than the logical process.
- Example: IBM POWER4
  • Two cores based on the Power PC.

POWER4 Chip Organization

Computer Evolution and Performance Part 12

Solutions
- Increasing the number of bits in one time

  • Create greater DRAM
- Changing the interface DRAM
  • Cache
- Reduce frequency access memory
  • Cache more complex and is located on the chip
- Improve the bus system
  • High speed buses
  • Hierarchy of buses


I / O Devices
- The device needs for operation of I / O
- Need Throughput data that is large enough
- Processor can handle this task
- Problems in the transfer of data
- Solution:
  • Caching
  • Buffering
  • High-speed interconnection buses
  • More elaborate bus structures
  • Multiple-processor configurations

Typical I / O Device Data Rates


The key: balance
- Processor components
- Main memory
- I / O devices
- Interconnection structures

Computer Evolution and Performance Part 11


Intel

  • 1971 - 4004
  • Microprocessor first
  • All components of the CPU is a single chip
  • 4 bits
  • Following the emergence of 8008 since 1972
  • 8 bits
  • Microprocessor with special design application
  • 1974 - 8080
  • Intel is microprocessor with the general usefulness.

Increased speed
  • Pipelining
  • On board cache
  • On board L1 & L2 cache
  • Branch PREDICTION
  • Data-flow analysis
  • Speculative execution

Comparison of performance
  • Increased processor speed.
  • Increased memory capacity.
  • Speed memory behind than processor speed.

Logic and Memory Performance Gap

Computer Evolution and Performance Part 10


DEC PDP-8

  • Introductions 1964.
  • Minicomputer first time (after the miniskirt).
  • Does not require air-conditioned room.
  • The form small, can be put on the table lab.
  • The price is $ 16,000
  • $ 100 + for the IBM 360
  • Already there are applications & OEM.
  • Bus architecture such as PDP-8 is later used by computers - the modern computer.

DEC - PDP-8 Bus Structure


Semiconductor memory
  • Start the 1970.
  • Fairchild.
  • Size of a single core
  • i.e. 1 bit of Magnetic core storage
  • Holds 256 bits
  • Non-destructive read
  • Faster than much core
  • Capacity approximately doubles each year

Computer Evolution and Performance Part 9


IBM 360 Series

  • Produced 1964.
  • Instruction Set Similar or identical, in these various groups of computer models issued using the same set of instructions so that the support system and device compatibility severity.
  • Operating System Similar or identical, this is a feature that consumers demand, so when the computer does not require the replacement of the difficulties in the operating system because the same.

IBM 360 Series
  • The increased speeds, models are offered ranging from low speed to high speed for the use of the consumers themselves can be adjusted.
  • Memory size is greater, the higher models will be increasingly large memory.
  • Price increases, the higher the style then the more expensive price.

IBM 360 images


Saturday, October 18, 2008

Computer Evolution and Performance Part 3


IAS

  • 1946, with the computer stored-program concept publised.
  • Known by the name Computer IAS (Computer of the Institute for Advanced Studies).

IAS structure
Here is the structure of the computer IAS:
  • Main memory, to store data and instructions.
  • Arithmetic Logic Unit (pestle), for binary data.
  • Control Unit, to control the instructions in memory.
  • I / O, to interact with the environment outside.

Structure computer IAS


IAS - details

>> Memory IAS consists of 1,000 storage locations called the word.
  • - Word consists of 40 bits.
  • Data and Instructions in a code in binary.
  • One word consists of 20 bit instructions, each - each 8 bit code operation (op code) and 12-bit address.
>> Set of Registers (storage in the CPU)
  • Memory Buffer Register (MBR), contains one word, stored in memory or used to receive word from memory.
  • Memory Address Register (MAR), to define the word address in memory to be written or read from the MBR by the MBR.
  • Instruction Register (IR) port, 8 instruction beet contains code that the operation will be executed.
  • Instruction Buffer Register (Heb), used to store instructions while the right word in memory.
  • Program Counter (PC), contains the address of the next pair of instructions will be retrieved from memory.
  • Accumulator (AC) and the multiplier quotient (MQ), used for temporary storage operand and the results pestle.
>> IAS operate repeatedly, forming cycle instructions.
  • Computers IAS has 21 instructions, which are grouped as follows:
  • Data transfer, transfer of data between memory with the register - register pestle or between two registers pestle own.
  • Unconditional branch, the command - the command execution without branching certain requirements.
  • Conditional branch, the execution branching certain requirements that need to produce a value of these ramification.
  • Arithmetic, the operation - operations which was formed by the pestle.
  • Modify Address, the instructions - instructions that allows changing the address on the so Computational address that allows high flexibility in programs.

Struktur IAS Detail


IAS Image