keywords:
Bookmark and Share



Front Back
What is the CPU?
  • Central processing unit or processor.
  • Brains of the computer.
  • Chip that processes instructions, manipulates data, and controls the interactions of the other circuits in your computer.
  • CPUs contained on a single chip are called microprocessors.
What are the three cpu components?
  1. Control Unit
  2. One or more execution units.
  3. Registers
What is the purpose of the control unit?
It is responsible for managing the flow of a program. It is the component that retrieves the next instruction to be acted upon or the data to be processed.
What's an execution unit?
Responsible for processing instructions and data. They are built from two units, ALU and FPU.
What does ALU stand for?
What does it do?
  • Arithemetic Logic Unit
  • It calculates and compares numbers. It does most of the work of the processor but its best suited to working with operations that act on whole numbers not fractions.
  • An execution unit.
What does FPU stand for? 
What does it do?
  • Floating-Point Unit
  • Is designed specifically to work with real numbers (fractions, very large, small, or precise numbers). It's faster and more efficient at performing mathematical manipulations than the ALU.
What does core mean?
  • A core is an execution unit.
  • A single core processor has one execution unit.
  • A dual-core has two, a triple-core has 3, quad core 4.
How does a single-core processor work?
  • The processor orders, executes, and then selectively stores strings of instructions in it cache (called registers, one cache has multiple registers).
  • When the processor needs data outside of its cache it must retrieve the data from RAM through the system bus or from a storage device such as the hard disk.
  • Performance is limited to the max speed of the bus, memory, or storage device.
  • When multi-tasking the processor switches back and forth between multiple sets of instructions.
How does a multi-core processor work?
  • Each core handles incoming strings of instruction simultaneously.
  • When one core is executing instructions, other cores can be accessing the system bus or executing their own string of instructions.
What is needed in order to utilize a multi-core processor?
  • The OS must be able to recognize multi-threading and the software must have simultaneous multi-threading technology (SMT) written into its code.
  • SMT enables parallel mutli-threading, meaning that multi-threaded instructions are delivered to the cores in parallel.
  • Without it the software will recognize only one core.
What is the difference between a multi-core processor and a multi-processor system?
  • In a multi-core processor system, system resources are shared and all cores reside on the same chip.
  • In a multi-processor system, there are two separate processors, each with its own system resources.
  • A multi-processor system is faster.
What is a register?
  • very small yet very fast memory locations for holding instructions or units of data.
  • Registers operate at the same speed as the CPU, whereas normal system memory can be many times slower.
How do registers work?
  • CPU's store data and instructions in registers during operations. That information is then transferred back to main system memory. To speed operations, the control unit can  "prefetch" instructions and data from system memory and store it in the CPU's registers.
  • CPUs can have many registers, with groups of registers devoted to a specific purpose. Some modern processors can use registers as needed for the task at hand, rather than being limited by special purpose registers.
What is an instruction?
  • The low-level hardware-specific command to be acted upon by a processor.
  • Something like "SUM 5, 10" is an instruction. Before it can be executed a previous instruction must have fetched the value 5 from a location in memory and stored it in a register, the same for 10. One more instruction is required to write the resulting sum back into the register, from where it is transferred by the control unit to a new memory location.
How do you determine the performance of a processor?
By examining the number of instructions it can perform in a second. It is usually rated in millions of instructions per second (MIPS).

cpu performance

What is addressable RAM?
The total amount of memory that's accessible to the processor.
cpu performance


What is Branch prediction?
A technique by which the processor anticipates the code that will be used next and loads that code to try to "get ahead" of the program. When a processor guesses correctly, the program speeds up. Otherwise performance slows while the correct instruction is retrieved.
cpu performance

How is Bus, address related to performance?
The bus (pathway) that connects the processor to main memory. The wider the address bus, the more memory can be accessed. Data isn't transferred over this bus.
cpu performance

How is Bus, data related to performance?
  • The number of bits of data or instructions that can be transferred in a single operation.
  • The larger the data bus, the more data that can be moved and thus the faster the processor can operate.
cpu performance

How is the Bus, internal related to performance?
  • The bus that determines how many bits of information the processor can work with at once.
  • If the internal bus is smaller than the data bus, data and instructions must be manipulated in parts. For example, a processor with a 32-bit internal bus and a 64-bit data bus must deal with data in two halves.
cpu performance

How does cache relate to performance?
  • High-speed memory contained within or directly coupled to the processor.
  • Accessing data from cache is considerably faster than accessing it from main memory. Processors use levels 1, 2, and 3 caching, where level 1 cache is the fastest and most closely coupled to the processor, level 2 less so, and level 3 even less. All are still faster than normal system memory.
  • Processors save instructions, the data to be processed and the results in the caches.
cpu performance

What is clock speed?
  • The number of cycles per second of the computer's synchronization clock, measured in hertz (Hz), millions of cycles per second (megahertz or MHz), or billions of cycles per second (gigahertz or GHz).
  • A modern processor performs more than one instruction during every clock cycle, older processors performed one or fewer.
  • Normally, a clock speed rating refers to the internal or core speed of the processor rather than to the actual speed of a computer's synchronizing clock chip.

cpu performance
What is a Dual Indpendent Bus (DIB)?
A processor architecture that includes to buses: one to the main system memory, and another to the level 2 cache. The processor can access both buses simultaneously for improved performance.
cpu performance

What is the front-side bus speed?
  • The speed at which the processor interacts with the rest of the system.
  • A processor's internal core speed can be many times higher than its front-side bus speed. If the core speed is too much higher than the front-side bus speed, the processor can sit idle, waiting for data to be moved in or out and made available for processing.
cpu performance

What is hyperthreading?
An Intel technology that enable a single processor to execute two streams of instructions at the same time, as if it were two processors.
cpu performance

What are Multimedia extensions (MMX)?
  • An expanded set of instructions supported by a processor that provides multimedia-specific instructions.
  • Without MMX a programmer might have to implement multiple low-level commands to perform a multimedia operation. With MMX, the same function would involve a single instruction.
cpu performance

What is out-of-order completion?
A technique by which instructions can be executed out of order when order isn't important and the processor determines a more efficient sequence.

cpu performance

What is overclocking?
Running the CPU at a higher speed than it was rated to run at. overclocking increases performance, but also increases the potential for errors. Also more heat is generated by an overclocked cpu.

cpu performance
What is pipelining?
  • The overlapping of the steps involved in processing instructions. Instructions are normally fetched, decoded, and executed, and the results are written out to the memory.
  • Modern processors overlap these steps to speed overall execution. While one instruction is being executed, another is being decoded, and a third is being fetched.
cpu performance

What is register renaming?

  • A technique by which modern processors can rename registers so that instructions can access their own set of registers and not interfere with other instructions.
  • When multiple instructions are running at the same time, there's a chance that two will attempt to read or write to the same register simultaneously. Register renaming prevents this.

cpu performance

What is Single Instruction Multiple Data (SIMD)?
  • A technique by which a single instruction can be applied to more than one piece of data. For example, with SIMD, fiver numbers might be moved into the processor along with the single command to add them up in one operation. The next operation would carry out that instruction.
  • Without SIMD. the numbers would be added together one by one in a longer sequence of instructions.

cpu performance

What is speculative execution?
  • A technique by which a processor executes an instruction in the expectation that the result is needed.
  • This technique can improve performance when the program branches, as in an "if this condition is met, do that operation" situation. The processor can begin performing the operation before the condition is fully evaluated in the expectation that it will be met.

cpu performance

What is superpipelining?
An improvement over pipelining. Superpipelining uses a larger number of shorter stages and support for a higher clock rate to improve performance.
cpu performance

What is superscalar?
A tehcnique that enables a processor to execute more than one instruction in a single clock cycle.
cpu performance

What is throttling?
A technique by which the speed of the processor is scaled back so that it uses less power and creates less heat. Throttling reduces performance. It is most useful with portable computers, for which low power consumption and low heat production are critical design factors.
That operating systems support more than one processor?
  • The operating system and applications must include symmetric multiprocessing code.
  • Windows 2000 Professional, XP Pro, and Vista Business, Ultimate, and Enterprise include symmetric multiprocessing code.
  • Linux is available in multiprocessing editions.
What are the three primary categories of processor specifications?
  1. Primary- Clock speed, front-side bus speed, addressable RAM, and cache sizes.
  2. Bus width
  3. Internal- internal and external voltages used by chips, number of transistors that make up the processor, and whether it includes an integrated FPU.
CPU and Bus width specifications detailed on 4-7 to 4-10
See book II
What are the internal specifications of a processor?
  • The internal specs refer to the way the CPU's circuits are constructed.
  • The core voltage value describes the voltage level required by the core processing components of the CPU.
  • The I/O voltage value sometimes called the external voltage, is the level required by the CPU's input and output circuitry.
  • In modern processors I/O and the core voltage values typically differ.
What do voltage ID (VID) pins do?
  • On modern processors VIDs send a signal containing the exact voltage requirements to the voltage regulator module (VRM) on the motherboard.
  • The RM then supplies that specific power to the CPU. Not all motherboards include a VRM. On a few motherboards, you must set jumpers or switches to match the voltage supplied to the CPU with what it requires. If you supply the wrong voltage you can ruin the CPU.
What is the significance of the number of transistors in a CPU?
The count of the transistors (microscopic electronic switches) provides a rough estimate of the size and complexity of the chip.
Internal specifications for Processors listed on 4-11 to 4-12.
See book II
What is a chipset?
  • One or more chips, packaged into a single unit and sold together, that perform a set of functions in a computer.
  • The chipset doesn't actually include the CPU. However each chipset is designed to support a select few CPUs.
  • Sometimes the "term" chipset is used with video adapter cards. In those cases the video chipset combines what used to be separate video chips into a single chip or unit.
The term chipset is most often used to describe the core features of a computer which often include:
  • memory control
  • system bus functions
  • audio functions
  • video display functions
  • system management functions
What is the Northbridge chip and what does it do?
  • One of the most important components of a PC chipset. works with CPU.
  • The Northbridge chip controls interactions between the CPU, memory, AGP video control circuitry, and the Southbridge.
What is the Southbridge chip and what does it do?
  • One of the most important components of a PC chipset. works with CPU.
  • The Southbridge chip controls interactions between buses and devices not controlled by the Nothrbridge, including the standard PCI expansion bus, floppy drive controller, and serial, parallel, and PS/2 keyboard and mouse ports.
What is a die?
  • Any type of microchip is made up of microscopic wires, transistors, and other components.
  • This plain chip is called a die.
  • Due to the size differences between the wires on teh die and the circuit boards of your computer, the die can't be connected directly tot he circuit board.
What is a package?
  • Houses the die.
  • A case made from plastic, ceramic, glass, metal or other material, plus wires and connectors that bridge the microscopic connections on the die with the external circuitry.
  • A package may also include support function chips, memory, and cooling-related components.
A PDIP (plastic dual inline package). Older package that used connectors that were large compared to the die.
Pin Grid Array Package
x of y cards Next >|