Processor
At the heart of every computer is a special motherboard chip called the processor, which determines, to a
great extent, the power of the computer. The processor is also called the central processing unit (CPU) or
microprocessor. The processor executes instructions, performs calculations, and coordinates input/output
operations. Each motherboard has electronic chips that work with the CPU and are designed to exact
specifications. Whether these other electronic components can keep up with the processor depends on the
individual component’s specifications. The major processor manufacturers today are Intel, Motorola, VIA,
Samsung, NVIDIA, Apple Inc., Qualcomm, and AMD (Advanced Micro Devices, Inc.). Intel and AMD are the
predominant manufacturers for desktop and laptop processors, and the other manufacturers target the
mobile/smartphone markets.
Processor Basics
Processors come in a variety of speeds, measured in gigahertz (GHz). Hertz is a measurement of cycles per
second. One hertz equals one cycle per second. One gigahertz is 1 billion cycles per second, or 1GHz. The
original PC CPU, the 8088 microprocessor, ran at 4.77MHz. Today’s processors run at speeds near 5GHz.
The number of bits processed at one time is the processor’s register size (word size). Intel’s 8086 processor’s
register size is 16 bits, or 2 bytes. Today’s CPUs have register sizes of 64 or 128 bits.
Buses
Processors operate on 1s and 0s. The 1s and 0s must travel from one place to another inside the processor,
as well as outside to other chips. To move the 1s and 0s around, electronic lines called a bus are used. The
electronic lines inside the CPU are known as the internal data bus or system bus. In the 8086, the internal
data bus comprises 16 separate lines, with each line carrying one 1 or one 0. The word size and the number
of lines for the internal data bus are equal. The 8086, for example, has a 16-bit word size, and 16 lines carry
16 bits on the internal data bus. In today’s processors, 64 or 128 internal data bus lines operate concurrently.
For a CPU to communicate with devices in the outside world, such as a printer, the 1s and 0s travel on the
external data bus. The external data bus connects the processor to adapters, the keyboard, the mouse, the
hard drive, and other devices. An external data bus is also known as an external data path. You can see the
external data lines by looking between the expansion slots on the motherboard. Some solder lines between
the expansion slots are used to send data out along the external data bus to the expansion slots. Today’s
processors have 64- and 128-bit external data paths.
ALUs
A processor has a special component called the arithmetic logic unit (ALU), which does all the calculations
and comparison logic that the computer needs. The control unit coordinates activities inside the processor.
The I/O unit manages data entering and leaving the processor. The registers in the CPU make up a high-
speed storage area for 1s and 0s before the bits are processed.
To make sense of all of this, take a look at a letter typed on a computer that starts out DEAR MOM. To the
computer, each letter of the alphabet is a different combination of eight 1s and 0s. For example, the letter
D is 01000100, and the letter E is 01000101.
Pipelines
Processors have multiple pipelines (separate internal buses) that operate simultaneously. To understand
pipelining, take the example of a fast-food restaurant. In the restaurant, assume that there are five steps
(and one employee per step) involved in making a burger and giving it to the customer. First, (1) take the
order and input it into the computer system; (2) brown the buns and cook the burgers; (3) add the
condiments to the buns and burgers; (4) wrap the burgers, add fries, and insert them into the bag; and then
(5) take the customer’s money and give the bag to the customer. Keep in mind that the person taking the
customer’s order and inputting the order can serve another customer once he or she has completed this
task for the first customer. The same is true for each person along the line. To make this burger process go
faster, you could (maybe) do one of the things
To relate this to processors, making the employees work faster is the same as increasing the CPU clock speed.
Breaking the tasks into smaller tasks is the same as changing the structure of the CPU pipeline. Instead of
performing the standard 5 tasks, the CPU might perform 6, 7, 14, 20, or even more steps. This allows each
step to be acted upon more quickly, the task to be smaller, and production to be faster. Having more lines
of people doing the same complete process is like having multiple pipelines.
A 32- or 64-bit CPU can have separate paths, each of which handles 32 or 64 bits. For example, if a processor
has two pipelines, the Dear Mom letter can be in one pipeline, while a photo upload using a different
application can be in the other pipeline.
A processor might have 12 pipelines for integers and 17 pipelines for floating-point numbers. (A floating-
point number is a number that can include a decimal point.) Other processors contain anywhere from 20- to
31-stage pipelines. Debate continues about whether a longer pipeline improves performance.
Speeding Up Processor Operations Overview
You can determine the speed of a processor by looking at the model number on the chip, but processors
frequently have devices attached to them for cooling, which makes it difficult to see the writing on the chip.
A processor commonly does not use its maximum speed all the time in order to save power or stay cool.
Also, a processor is not always functioning at its maximum potential for a lot of reasons including coding
used within an application, the user switching from application to application, inadequate bus width, or the
amount of RAM installed. The processor can also operate beyond its rated specifications. Intel Turbo Boost
allows the processor to operate faster than it is rated in order to handle periods of increased workload.
Cache
An important concept related to processor speed is keeping data flowing into the processor. Registers are a
type of high-speed memory storage inside the processor. They are used to temporarily hold calculations,
data, or instructions. The data or instruction the CPU needs to operate on is usually found in one of three
places: cache memory, the motherboard memory (main memory), or the hard drive.
Cache memory is a very fast type of memory designed to increase the speed of processor operations. CPU
efficiency is increased when data continuously flows into the CPU. Cache provides the fastest access. If the
information is not in cache, the processor looks for the data in motherboard RAM. If the information is not
there, it is retrieved from the hard drive and placed into the motherboard memory or the cache. Hard drive
access is the slowest of the three.
An analogy best explains this. Consider a glass of cold lemonade, a pitcher of lemonade, and a can of frozen
lemonade concentrate. If you were thirsty, you would drink from the glass because it is the fastest and most
easily accessible. If the glass were empty, you would pour lemonade from the pitcher to refill the glass. If the
pitcher were empty, you would go to the freezer to get the frozen concentrate to make more lemonade.
Usually, the more cache memory a system has, the better that system performs, but this is not always true.
System performance also depends on the efficiency of the cache controller (the chip that manages the cache
memory), the system design, the amount of available hard drive space, and the speed of the processor.
When determining memory requirements, you must consider the operating system used, applications used,
and hardware installed. The Windows XP operating system takes a lot less memory than Windows 10. High-
end games and desktop publishing take more RAM than word processing. Free hard drive space and video
memory are often as important as RAM in improving a computer’s performance. Memory is only one piece
of the puzzle. All of the computer’s parts must work together to provide efficient system performance.
Clocking
The motherboard generates a clock signal that is used to control the transfer of 1s and 0s to and from the
processor. A clock signal can be illustrated as a sine wave. One clock cycle is from one point on the sine wave
to the next point that is located on the same point on the sine wave later in time.
In older computers, data was sent to the CPU only once during a clock cycle. Then, newer memory
technologies evolved that allow data to be sent twice during every clock cycle. Today, data is sent four times
during a single clock cycle.
Threading Technology
Several threading techniques are used to speed up processor efficiency: multithreading and HT (Hyper-
Threading Technology). A thread is a small piece of an application process that can be handled by an
operating system. An operating system such as Windows schedules and assigns resources to a thread. Each
thread can share resources (such as the processor or cache memory) with other threads. A thread in the
pipeline might have a delay due to waiting on data to be retrieved or access to a port or another hardware
component. Multithreading keeps the line moving by letting another thread execute some code. This is like
a grocery cashier taking another customer while someone goes for a forgotten loaf of bread.
Intel’s HTT (Hyper-Threading HT or HT Technology) allows a single processor to handle two separate sets of
instructions simultaneously. To the operating system, HT makes the system appear as if it has multiple
processors. Intel claims that the system can have up to a 30 percent increase in performance, but studies
have shown that the increase is application dependent. If the application being used cannot take advantage
of the multithreading, then HT can be disabled in the system BIOS/unified extensible firmware interface
(UEFI).
Connecting to the Processor
We have considered various ways to speed up processor operations, including having more stages in the
processor, increasing the speed of the clock, and sending more data in the same amount of time. Accessing
L2 cache and motherboard components was a bottleneck in older systems because the CPU used the same
bus to communicate with RAM and other motherboard components as it did with L2 and motherboard
cache. The solution is DIB (dual independent bus). With DIB, two buses are used: a back side bus and a front
side bus. The back side bus connects the CPU to the L2 cache. The FSB (front side bus) connects the CPU to
the motherboard components. The FSB is considered the speed of the motherboard.
Many people think that the higher the CPU speed, the faster the computer. This is seldom true. Several
factors contribute to computer speed. One factor is bus speed. Bus speed describes how fast the CPU can
communicate with motherboard components, such as memory, the chipset, or the PCI/PCIe bus. The first
Pentium CPUs ran at the same speed as the bus (60MHz); in time, CPUs got faster and buses stayed the
same. Advances in technology have not reached the rest of the motherboard components (and it would cost
too much to try to have them keep pace).
Intel and AMD have technologies to replace the front side bus in some parts. AMD’s solution is Direct
Connect. Direct Connect allows each of the processor cores to connect directly to memory, to the other
motherboard components such as the expansion slots, and to other processor cores using a high-speed bus
called HyperTransport. Intel has QuickPath Interconnect (QPI), Serial Peripheral Interface (SPI), and Direct
Media Interface (DMI), which are full-duplex (that is, traffic can flow in both directions simultaneously) point-
to-point connections between the processor and one or more motherboard components. This type of
connectivity used with Intel-based processors and chipsets.
Multi-Core Processors
In the past, when two processors were installed, software had to be specifically written to support having
multiple processors. That is no longer true. A dual-core processor combines two CPUs in a single unit. A tri-
core processor has three processors in a single unit. Both Intel and AMD have quad-core CPU technologies,
which is either two dual-core CPUs installed on the same motherboard, two dual-core CPUs installed in a
single socket, or today’s model of all four cores installed in one unit. Now there are also hexa-core (six cores)
and octa-core (eight cores) processors. IT professionals in the field find it easiest to just say multi-core to
describe the multiple cores contained in the same processor housing. Single-core processors and early dual-
core processors accessed memory through a memory controller. Today, the processor cores have their own
memory controller built into the processor. Below Image shows how an AMD quad-core processor has an
integrated controller and interfaces with the rest of the motherboard using a high-speed bus called
HyperTransport. HyperTransport is a feature of AMD’s Direct Connect architecture. With Direct Connect,
there are no front side buses. Instead, the memory controller and input/output functions directly connect
to the CPU.
All applications can take advantage of the multi-core technology and the background processes that are
associated with the operating system and applications. This improves operations when multitasking or when
running powerful applications that require many instructions to be executed, such as drawing applications
and games.
Graphics Processing Unit (GPU)
Another bottleneck for computer performance is video. Computer users who want better video performance
buy a separate video adapter that contains a GPU. Both Intel and AMD have a graphics processing unit (GPU)
within the CPU on some of their processor models. With an integrated GPU (iGPU), sometimes called an
integrated graphics processor (IGP), an external video card with a GPU is not required, and graphical data is
processed quickly, with reduced power consumption. Today’s CPUs contain multiple core processors,
whereas GPUs contain hundreds of smaller core processors. GPUs can also be used for other purposes that
are not directly related to graphics that increase system performance. These GPUs are sometimes referred
to as a general-purpose GPU (GPGPU). A computer system can also have multiple GPUs. AMD provides
information about the number of “compute cores.” For example, an AMD system that has four CPUs and
two GPUs would have six compute cores.
Integrated GPUs can either share part of the motherboard RAM with the rest of the system or have a
separate block of memory dedicated for video. Integrated GPUs can have their own cache memory or share
with the CPU. IGPs can be part of the chipset or be included as part of the CPU housing (on-die). AMD calls
its processors that have a GPU integrated with the CPU an accelerated processing unit (APU). Intel calls its
integrated GPU Intel HD Graphics and Intel Iris Graphics.
Virtualization
One advantage of having multiple processor cores is that home and business computers can take advantage
of virtualization. Virtualization is having one or two virtual machines on the same computer. Virtualization
software, such as VMware Workstation, Oracle VM VirtualBox, or Microsoft Hyper-V, enables one computer
to act as if it were two or more computers. The computer can have two or more operating systems installed
through the use of the virtualization software. Each operating system would have no knowledge of the other
operating system.
Windows 7 has Virtual PC and Windows 8 has Hyper-V, which allow an application to run in a virtual
environment as if an older operating system had been installed. The concept of virtualization is of interest
to businesses so that legacy software can be put on a newer machine but kept separate from the main
operating syshome and business computers can take advantage of virtualization. Virtualization is having one
or two virtual machines on the same computer. Virtualization software, such as VMware Workstation, Oracle
VM VirtualBox, or Microsoft Hyper-V, enables one computer to act as if it were two or more computers. The
computer can have two or more operating systems installed through the use of the virtualization software.
Each operating system would have no knowledge of the other operating system.
Windows 7 has Virtual PC and Windows 8 has Hyper-V, which allow an application to run in a virtual
environment as if an older operating system had been installed. The concept of virtualization is of interest
to businesses so that legacy software can be put on a newer machine but kept separate from the main
operating system or another virtualized machine on the same computer. Reduced costs and physical space
are benefits of virtualization. Home computer users can install multiple operating systems in separate VMs
(virtual machines) within the same physical box, with each VM being seen as a separate computer.
Selecting a motherboard and processor is important when in a virtual environment. Not all processors were
designed for virtualization. Refer to the virtualization software documentation to determine whether the
CPU used is allowed to be used in a virtual environment. Another issue regarding processors and
virtualization is licensing. For virtualization software that must be purchased (that is, is not freeware), the
software manufacturer can charge on a per-processor or per-socket license basis or a per-core basis. If a CPU
has four cores, then pricing might play into what virtualization software is purchased.
Intel Processors
Traditionally, Intel has rated its processors by GHz and people have compared processors based on speed
alone. Now, Intel arranges its products by family numbers. In a family of processors, you can compare
attributes such as speed and the amount of cache memory and other technologies.
CPU Sockets
A processor inserts into a socket or slot, depending on the model. Most processors today insert into a socket.
There are different types of sockets: pin grid array (PGA), which has even rows of holes around a square
socket; staggered pin grid array (SPGA), which has staggered holes so more pins can be inserted; plastic pin
grid array (PPGA); micro pin grid array (µPGA); flip chip ball grid array (FCBGA); and land grid array (LGA) are
all used with either AMD and/or Intel processors.
Processor sockets are also called zero insertion force (ZIF) sockets; they come in different sizes. A processor
socket accepts one or more specific processor models. The socket has a small lever to the side that, when
lifted, brings the processor slightly up and out of the socket holes. When installing a processor, the CPU is
aligned over the holes and the lever is depressed to bring the processor pins into the slot with equal force
on all the pins.
Processor Cooling
Keeping the CPU cool is critical. Both Intel and AMD have technologies that reduce processor energy
consumption (and heat) by turning off unused parts of the processor or slowing down the processor when
it starts to overheat. But these measures alone are not enough.
Installing CPU Thermal Solutions
Some CPUs come with a thermal solution such as a heat sink and/or fan. The thermal solution commonly
comes with a preapplied thermal paste or attached thermal pad. Heat sinks and fans attach to the processor
using different methods. The most common methods are screws, thermal compound, and clips. Clips can
use retaining screws, pressure release (where you press down on them, and they release), or a retaining slot.
Small screwdrivers can be used to release the clips that attach using the retaining slot. Clips for fans or heat
sinks can be difficult to install. The type of heat sink and/or fan installed must fit the processor and case.
Additional hardware may have to be installed on the motherboard to be able to attach the CPU thermal
solution.
CPU fans frequently have a 3- or 4-pin cable that attaches to the motherboard. The motherboard might have
a 3- or 4-pin connector. A 3-pin fan can be attached to a 4-pin motherboard connector, and a 4-pin fan cable
can be connected to a 3-pin motherboard connector. Note that when a 3-pin cable attaches to 4-pin
connector, the fan is always on and cannot be controlled, like a 4-pin cable to 4-pin connector can.
Overclocking Processors
Overclocking is changing the front side bus speed and/or multiplier to boost CPU and system speed.
Overclocking has some issues:
• CPU speed ratings are conservative.
• The processor, motherboard, memory, and other components can be damaged by overclocking.
• Applications may crash, the operating system may not boot, and/or the system may hang (lock up)
when overclocking.
• The warranty may be void on some CPUs if you overclock.
• When you increase the speed of the CPU, the processor’s heat increases. Extra cooling, using fans
and larger heat sinks, is essential.
• Input/output devices may not react well to overclocking.
• The memory chips may need to be upgraded to be able to keep up with the faster processing.
• You need to know how to reset the system BIOS/UEFI in case the computer will not boot properly
after you make changes.
Many motherboard manufacturers do not allow changes to the CPU, multiplier, and clock settings. The
changes to the motherboard are most often made through BIOS/UEFI Setup. However, CPU manufacturers
may provide tuning tools in the form of applications installed on the computer for overclocking
configuration. Keep in mind that overclocking is a trial-and-error situation. There are websites geared toward
documenting specific motherboards and overclocked CPUs.
