Picking Your Processor

Understanding Processor Terminology

In this section, I cover some basic terms that describe characteristics of different processors, past and present. The A+ Exams might not ask for the specific definition of each term, but understanding the terms will help you answer the related questions in this topic area.

Processor speed

Processor speed is how fast a processor executes its instructions or commands. This speed was originally measured in millions of hertz, or megahertz (MHz), per second. A hertz is also known as a clock cycle, and a processor can execute code at every clock cycle. Thus, a processor operating at a measly 1 MHz per second can execute one million tasks every second. Processors today now measure their speed in gigahertz (GHz) per second. A gigahertz is one billion clock cycles per second — so the CPU can execute tasks a billion times per second!

Original CPUs had a speed of 4.77 MHz, and systems at the time of this writing are running over 3.0 GHz. Although processor speed is not the only factor affecting performance, in general, the faster the processor, the faster the system.

Data bus

A city bus is responsible for transferring people from one location to another. In the world of computers, a bus is responsible for delivering data from one location on the PC to another. Data bus is the term used to define the pathway between the processor and memory. Because the processor accesses information from memory so often, an entire bus — the data bus — is dedicated to this action. The larger the data bus, the more data can be carried from the CPU to memory in one clock cycle.

Here’s an illustration. Say 50 people needed to go from one end of the city to the other, but a city bus had only 25 available seats. The solution? The bus would make two trips. Hmm. Wouldn’t getting a larger bus be more efficient? If you upgraded the bus to 50 seats, the bus would have to make only one trip to transfer the 50 people from one end of the city to the other, which increases the efficiency of the public transit system.

The data bus works the same way, only it transfers data in the form of bits. A single bit is either a one or a zero. All data processed by the computer is in the form of bits. The data bus has a full capacity point at which it cannot handle any more bits of data, just like the bus system in the city has a full capacity point (measured in seats).

If a processor has a 32-bit data bus, it can deliver — at most — 16 bits during a single clock cycle. If the same processor needs to deliver 64 bits of information, it has to take two trips: sending 32 bits during the first clock cycle and the remaining 32 bits during the next clock cycle. Taking that same 64 bits of information and processing it on a 64-bit processor means that the information will be delivered in one trip — one clock cycle — as opposed to two, which increases the overall efficiency of the system.

Address bus

Figure 2-1 shows how system memory is organized like a spreadsheet, in rows and columns. These rows and columns make up blocks that can be written to and read from. If you want to store information in one of the blocks, you have to reference the location by address. For example, you may store data in cell B2.

FIGURE 2-1: How system memory is organized.

To store information into system memory, your processor has to give an address that points to a particular storage location, only the address doesn’t look like B2. It looks something like 10, or maybe 11, which are two completely different memory locations. As a result, the data would get stored in two different blocks.

Your processor accesses memory locations through the address bus. If, for example, the address bus is two-bit, the processor has two address lines from the processor to system memory. The address lines carry signals that specify locations in memory, each with an on/off state. A 1 represents an on state, and 0 represents an off state. The combination of the on/off states of both address lines at any given time is how a reference to an area in memory is made. The left side of Figure 2-2 illustrates a processor making a reference — or call — to address 10. The left side of the figure shows a reference to address 11. These two address calls reference completely different locations in memory.

FIGURE 2-2: Accessing two different memory addresses with the address bus.

If you add another address line to the address bus, the processor can access even more possible addresses because the processor has more variations with three bits than with two. A two-bit address bus can make a reference to four possible memory addresses (2 × 2), but a three-bit address bus can make a reference to eight possible memory addresses (2 × 2 × 2).

Therefore, the address bus dictates how much physical memory the processor can access. For example, an old 80286 processor has a 24-bit address bus, which means that it can access 16,777,216 (2²⁴) memory addresses, or 16MB of system memory. A few years back, some processors had a 36-bit address buses, which allows them to access 68,719,476,736 memory addresses, or 64GB of memory. Newer processors are exceeding the 36-bit address bus, and one more example is a 40-bit address bus which allows for 1,099,511,627,776 addresses (or 1 TB of RAM)!

Registers

Registers are storage areas within the processor used to store data temporarily for manipulation later. They are used to store and process data and perhaps write back the result of the processed data. The benefit of storing this information in the registers — instead of in memory — is that the processor contains the information and does not have to retrieve it from memory, which takes time. It is as if information to be processed were in your pocket, rather than across a room, where you would have to walk all the way over and pick it up. Having information in your pocket means it can be accessed much more quickly, saving time and increasing performance. Registers give a processor quicker access to data; the more registers a processor has, the more data it can store.

Registers are measured in bits. A processor with 32-bit registers has 32 containers into which a programmer can choose to store information. Comparatively, a processor with 64-bit registers has twice as many containers that it can use to store information.

Cache memory

The processor accesses information that resides in system memory, which is a slower process than if the information is stored in the processor’s own special high-speed memory, known as cache memory. When the information is sitting in system memory and the processor sends a request for that information, the request goes to the memory controller, which manages data in memory. The memory controller finds the data in memory, retrieves it, and delivers it to the processor. Throughout this entire process, the processor is simply waiting around for the information. Thus, many newer processors include their own special high-speed memory within the processor’s chip.

When the processor retrieves information from slower system memory, it then stores it in the high-speed cache in case the processor wants to access the information a second time. The benefit is that the second time the data is needed, it is sitting in the high-speed memory located on the processor chip. The processor does not need to sit around and wait for the data to come from system memory — again, increasing overall performance.

Cache memory is integrated right into the processor’s chip and is made up of static RAM (SRAM). For more information on SRAM check out Book 2, Chapter 3. Cache memory is very expensive because it is much quicker than regular system memory. As a result of this extra memory being integrated into the processor chip, the processor becomes more expensive than a processor that has less or no cache memory.

The three types of cache memory are Level 1 (L1) cache, Level 2 (L2) cache, and Level 3 (L3) cache. L1 cache is built into the processor, whereas L2 cache resides outside the processor. In the past, L2 cache resided on the motherboard, but newer processors have a bit of L1 and L2 cache for each core in the processor and then a shared block of L3 cache for all cores in the processor.

Math co-processor

The math co-processor, also known as the arithmetic logic unit (ALU), is the processor’s sidekick. Systems with math co-processors can well outperform systems that do not because the math co-processor takes some of the workload off the CPU. For example, it performs many of the large calculations that applications may require, such as floating point arithmetic. Overall system performance increases because the CPU can focus on logic functions while the math co-processor executes complicated mathematical functions.

If you have large spreadsheets or use large graphics applications, you might find that applications run very poorly or not at all on systems without a math co-processor. If you are running a system that does not have a math co-processor integrated into the CPU, you can add one to the motherboard, or perhaps upgrade the main processor.

In earlier computers, the processor was one chip, and the math co-processor was a separate chip on the motherboard. Today’s processors all have the math co-processor integrated into the CPU, so you will not purchase or add an NPU to the system — or even hear much about a math co-processor.

MMX

After the Pentium was developed, Intel introduced a feature called MultiMedia eXtensions, or MMX. MMX added 57 new instructions that were built into the processor and told the system how to work with audio, video, and graphics. If these instructions were not built into the processor, the processor would have to retrieve them from somewhere else.

When MMX was developed, both home and business users seemed to be heading toward the world of multimedia, and it made sense to enhance the processor and make it “multimedia aware.” Running any kind of multimedia application on a processor that supports MMX gives you a major performance increase over a processor that doesn’t support MMX technology.

MMX is an old instruction set, but my point with putting this in here is for you to see that processor manufacturers will install instructions into the processor for whatever is a hot technology at the time in order to increase performance when using that technology.

Hyperthreading

Hyperthreading is a feature designed by Intel that was placed in the Pentium processors. Hyperthreading technology, or HTT, allows a processor to logically act as two different processors by being able to execute simultaneous threads. A thread is a part of an application that executes at any given time. For example, when running Microsoft Word, one thread accepts keystrokes, and another thread runs the spell checker while you type. Thus, two parts of the application run at the same time.

For a system to truly be able to take advantage of multithreaded applications, you normally need a system that has multiple processors — one processor to run one thread at a time. With hyperthreading, one processor can run more than one thread at a time, increasing performance by 15 to 30 percent.

Multicore

A multicore processor combines a number of independent processors and the L1 cache from those processors onto a single processor chip. The benefit of a multicore processor is that it can execute multiple threads at the same time without hyperthreading because you essentially have multiple processors in one chip package.

A multicore processor has the benefit of having multiple processors’ core features — such as superscalar execution, pipelining, and threading — all packaged into one physical processor. The core features also include each core having its own L1 cache memory. Multicore processors also have a block of shared L2 cache between the two processors in the multicore chip.

A huge benefit of being only one chip on the motherboard is that the one multicore chip draws less power than two separate processors would. Figure 2-3 shows the logical view of a dual-core processor.

FIGURE 2-3: Looking at the logical structure of a dual-core processor.

A number of different flavors of multicore processors are available today, such as dual-core, triple-core, and quad-core processors. Here are the differences between the three:

Dual core: Has two cores in one chip package, with each core typically having 128K of L1 cache and 512K of shared L2 cache.
Triple core: Has three cores in one chip package with each core typically having 128K of L1 cache. Most triple-core processors also have 512K of L2 cache per core and share a block of cache memory, known as L3 cache.
Quad core: Has four cores in one chip package with each core typically having 128K of L1 cache. Most quad-core processors also have 512K of L2 cache per core and share a block of L3 cache (2MB–6MB).

Virtualization support

One of the newest features to come from processors in recent years is virtualization support. Virtualization allows you to run multiple computers, known as virtual machines (VMs), on one physical computer. Each virtual machine runs a separate operating system with virtual resources assigned to it such as CPUs, virtual hard drives, and memory.

In order to use virtualization software with your system, your processor must support it and virtualization must be enabled in the BIOS. I don’t know how many times I have installation virtualization software such as Hyper-V and then go to start a VM and I receive an error. After this happens, I check the BIOS settings and find out that virtualization support is disabled. A quick change of that setting to “enable” and we are able to run virtual machines!

Depending on the manufacturer of the processor you will find that the virtualization extensions are called something different. If you have an Intel processor the virtualization feature is called Intel VT (for Intel Virtualization Technology), but if your processor is an AMD processor then the virtualization feature you must enable is called AMD-V (for AMD Virtualization).

To run virtualization software such as Microsoft’s Hyper-V, you must have a processor that supports it and it must be enabled in the system BIOS.

Architecture (32-bit or 64-bit)

When purchasing a system, you should always investigate whether the system has a 32-bit processor or a 64-bit processor. The 64-bit processors are needed to run today’s 64-bit operating systems and software, and well outperform the 32-bit systems.

Virtualization software — such as Microsoft’s Hyper-V — is only available with 64-bit editions of Windows, such as Windows 8.1 or Windows Server. If you do not have a 64-bit computer and a 64-bit edition of Windows 8.1, you are unable to run Hyper-V. My point being if that was your goal and you did not investigate the architecture of the processor before purchasing the second-hand computer, you may not be able to run the software you want.

Integrated GPU

New processors today may have an integrated graphics processing unit (GPU), which would take care of the video calculations and processing for the system. If the processor does not have an integrated GPU and you need to run heavy graphics applications, you will most likely need to purchase a video card with a dedicated GPU in it.

One of the downfalls of using an integrated GPU is that it will use system RAM as its memory, while a dedicated GPU is an expansion card that has the GPU and its own memory for video processing. This is important because if you have 16GB of RAM in the system and an integrated GPU, the GPU is using some of that 16GB of RAM. A dedicated GPU would not, as it has its own memory on the graphics card.

The benefit of the integrated GPU is that it is less expensive and also uses less power than a dedicated GPU. The integrated GPU would also produce less heat, meaning you wouldn’t need to worry about cooling as much.

Remember that the CPU performs logic operations, the NPU performs mathematical calculations, and the GPU takes care of video operations.

Execute Disable Bit

Execute Disable Bit (EDB) is a security feature that allows the processor to flag different areas of memory that are, and are not, allowed to run code. This is a great security feature as it is a method to help prevent buffer overflow attacks. Buffer overflow attacks are when the hacker submits code to areas of memory in order to run that code. Now with EDB, the code submitted into memory by the hacker would not have permission to execute unless it is running from an area flagged for code execution.

To take advantage of EDB you must have a processor that supports it, and you must enable it in the BIOS/UEFI.

Throttling

Throttling (a feature built into a lot of newer processors today) involves the CPU sensing when it is going to overheat and then reducing its speed to lower the heat to an acceptable range.

Processors that support throttling have a built-in thermal sensor (a high-tech thermometer) that monitors the temperature of the processor. When the processor detects that it is going to overheat — maybe, because of a fan failure — the processor drops its speed so that the temperature drops to an acceptable range.

Overclocking

Overclocking, a big feature for PC enthusiasts, involves running a piece of hardware faster than the speed at which it is rated. A number of devices can be overclocked, such as video adapters and (of course) processors.

Although you might be able to overclock the processor, it is not recommended because overclocking can result in an unstable system or even hardware failure.

VRM

The voltage regulator module (VRM) is responsible for regulating the voltage that is delivered to the processor. The VRM is located on the motherboard (or appears as its own device in the system) and provides the correct running voltage to the processor.

Some VRMs use a jumper on the motherboard to determine how much voltage is supplied to the processor, and other VRMs sense what the processor needs on startup. Typically, VRMs on the motherboard sense what voltage the processor needs and then supply that voltage.

Chip packaging

Chip packaging refers to how the chip is constructed and delivered to the consumer. The chip package defines the appearance or form factor of the chip. Many chip packages have been used over the years.

The chip packages you should be familiar with for the A+ Exams are as follows:

Dual inline package (DIP) chip: A rectangular chip with two rows of 20 pins. Pin 1 is located at the end of the chip that has a square notch carved into it. It is important to identify pin 1 because when you add a DIP chip to the motherboard, you have to match pin 1 on the chip with pin 1 in the chip socket.

Older processors (such as the 8088) and many math co-processor chips use the DIP chip style. Although they are no longer used for CPUs, DIP chips are still used for cache memory and BIOS chips on motherboards. They are also found on memory modules. (See Book 2, Chapter 3 for a discussion of memory modules.)
Pin grid array (PGA) chip: One of the most popular processor chip packages in use today, the PGA chip is a square chip with an array of pins filling up the shape of the chip. In general, the PGA chip uses hundreds of pins. You can locate pin 1 on the PGA by identifying the corner of the PGA chip that has the corner cut off — that corner is where pin 1 is located. Figure 2-4 compares a DIP (right side) with a PGA (left side) chip type.

Today’s implementations of processors typically fit into a zero insertion force (ZIF) socket. The ZIF socket is ideal for upgrading processors (especially compared with the days before ZIF sockets were used) because the ZIF socket has a lever (on the side of the socket) that you lift to raise the chip from the socket. Because the chip is automatically raised out of the socket, you can simply remove the chip out of the socket with little effort!

Before ZIF sockets were used, you had to pry the chip out of the socket while trying to ensure that you did not damage the chip or the pins. With the ZIF socket, after the processor is raised, you can replace the old chip with a new one. In the past, not all boards used ZIF sockets, so you had to get some special extractors to pull the chip out (carefully!). A ZIF socket is shown in Figure 2-5.
Land grid array (LGA) chip: A chip packaging style similar to PGA is the LGA, which has a rectangular packaging style and contacts instead of pins on the bottom side of the chip. These contacts make a connection to the contacts within the LGA socket on the motherboard when the processor is placed in the LGA socket. An LGA chip and an LGA socket are shown in Figure 2-6.

The LGA socket is similar to a ZIF socket in the sense that it is designed to make it easy for someone to remove the processor from the system. The LGA socket uses a surface plate to hold the chip in place.
Single Edge Contact (SEC) chip: A chip package type that was popular with the Pentium II processors, the SEC chip is a huge cartridge surrounded by a plastic casing. The newer version, SEC2, is implemented as a card that is inserted into a slot on the motherboard and doesn’t have the big plastic casing around it. It is important to stress that the SEC and SEC2 are inserted into a slot and not a socket. For more information on slots and sockets, read the next section. Figure 2-7 shows an SEC chip package along with some PGA chip packages.

FIGURE 2-4: Comparing a DIP chip package (right) to a PGA package (left).

FIGURE 2-5: A ZIF socket on the motherboard holds a processor.

FIGURE 2-6: LGA processor and LGA socket.

FIGURE 2-7: An SEC chip package along with some PGA chip packages.

Be sure to remember the different chip package types for the A+ Exams. The older Pentium II processor used the SEC, while newer processors, such as the Pentium 4 and Intel Core i7 series, are using the PGA and LGA.

Identifying Socket Types

Intel decided to develop a new standard for upgrading a processor on motherboards, beginning with the 80486 chips and continuing with the Pentium-class processors. This standard — a processor socket designed to hold a specific processor chip with the appropriate number of pins — enabled Intel to develop new chips with compatibility of a particular socket in mind. For example, if a socket is developed with 321 pins, Intel could develop a new processor that has 321 pins and know that the processor will work with any motherboard that has the right socket. This allows the consumer to upgrade a processor much easier than in the past. Intel could design a new chip for an old socket so that customers could update their computers by dropping the new processor in the compatible socket.

Original Pentium processors supported mainly Socket 5 with 320 pins or Socket 7 with 321 pins. Thus, to add a Pentium processor to a motherboard, you determine what socket exists on that board and then purchase a CPU to fit in that socket. You also have to remember to match the voltage of the board to the voltage required by the CPU. Figure 2-8 helps you identify a CPU socket in your system.

FIGURE 2-8: Identifying a processor socket.

Sockets are normally labeled by type along the side of the socket. For example, in Figure 2-8, the socket is labeled as PGA 370, meaning that it is Socket 370 and will hold any processor designed for Socket 370. (Socket 370 is a socket that holds a processor containing 370 pins.)

Table 2-1 lists the different types of sockets and the processors that are placed in the sockets. For more information about the processors, read the sections “Looking at Intel Processors” and “Don’t Forget Non-Intel Chips,” later in this chapter. Table 2-1 also shows the number of pins associated with the different types of sockets and slots.

TABLE 2-1 Processor Socket Types and Slots

Socket	Processor	Number of Pins
Socket A	Later Athlon, Duron, and Athlon XP	462
Socket 4	Pentium 60/66	273
Socket 5	Pentium 75-133	320
Socket 7	Pentium 75-200	321
Socket 8	Pentium Pro	387
Socket 370	Celeron and Pentium III	370
Socket 418	Itanium	418
Socket 423	Pentium 4	423
Socket 478	Later Celerons and Pentium 4	478
Socket 603	Xeon (Pentium 4 version)	603
Socket 611	Itanium	611
Socket 940	Opteron	940
Slot A	Athlon	242
Slot 1	Pentium II and Pentium III	242
Slot 2	Xeon	330
Socket 754	Turion 64, Athlon 64, Sempron	754
Socket 775 or LGA775 (also called Socket T)	Pentium 4, Pentium D, Celeron D, Core 2 Duo	775
LGA1366 (aka Socket B)	Intel Core i7 processor	1366
Socket 1156 (LGA)	Intel Core i3, Core i5, and Core i7 – 1st Generation (Lynnfield/Clarkedale)	1156
Socket 1155 (LGA)	Intel Core i3, Core i5, Core i7 – 2nd and 3rd Generation (SandyBridge and Ivy Bridge)	1155
Socket 1150 (LGA)	Intel Core i3, Core i5, Core i7 – 4th Generation (Haswell-DT)	1150
LGA 2011 (aka Socket R)	Intel Core i7 processor (high-end workstations and servers)	2011
AM2	Opteron, Sempron, Athlon 64, Athlon 64 X2	940
AM2+	Athlon 64, Athlon 64 X2, Phenom, Phenom II	940
AM3	Phenom II	941
AM3+	AMD FX	942
FM1	AMD Llano	905
FM2	Athlon X2 and Athlon X4	904
FM2+	AMD Kaveri APU	906
F	AMD Athlon 64 FX, Opteron	1207

Know the socket types used to hold the Pentium 4, Celeron, Intel Core i7, Athlon XP, Athlon 64, Phenom, and Turion processors. You will not be expected to memorize the entire Table 2-1, but you should be familiar with the sockets used by today’s popular processors.

Originally, the sockets were simply called Socket 1, Socket 2, and so on up to Socket 8. To make it easier to understand what processors went into which sockets, Intel started naming the sockets after the number of pins that existed on the processor that the socket would support. For example, Socket 370 holds a processor with 370 pins, and Socket 478 holds a processor with 478 pins. It is much easier now to identify what processors go into which sockets!

Now that you understand some of the characteristics of processors and you understand what a socket is, take a look at some of the popular Intel and AMD chips you are expected to know for the A+ Exams.

Looking at Intel Processors

In this section, I provide an overview of the Pentium-class processors and their characteristics, including data bus, address bus, registers, and the amount of cache memory supported on these processors. You also are introduced to any new or unique processor features that each processor offers.

Pentium

The original Pentium processor, released in 1993, was developed at speeds of 60 MHz and 66 MHz. The Pentium processor was a PGA chip that was placed in Socket 5 or Socket 7. Soon after its release, Intel marketed Pentium processors in 75, 90, 100, 120, 133, 150, 166, and 200 MHz flavors, which were really just clock multipliers of the original 60 MHz or 66 MHz systems.

Clock multiplying is the concept that the processor will run faster than the motherboard that the processor sits in. For example, the original Pentium processor ran on 60 or 66 MHz motherboards. Say that the computer is marketed as a Pentium 90. Because you know that the motherboard runs at 60 or 66 MHz, you can determine that the 90 comes from 60 × 1.5 — meaning that the processor runs 1.5 times the speed of the motherboard. This is important because as a consumer, when you purchase a computer, you also want to make sure you know what the motherboard speed is — not just the advertised speed of the processor.

From a consumer’s point of view, clock multipliers become important when you take a look at computers such as the Pentium 133 and the Pentium 150. Which is faster? The obvious answer is the Pentium 150, the system with the higher megahertz speed. But is it really? The Pentium 133 is a clock double of the 66 MHz board, and the Pentium 150 is a clock double and a half of the 60 MHz board. My point being that the overall performance of the system is controlled by more than just the speed of the processor — you need to consider other components, such as the speed of the motherboard.

By looking at the motherboard speeds of the Pentium 133 and the Pentium 150, you could assume that the computer running the Pentium 133 might be able to keep up with (if not outperform) the one running the Pentium 150. Table 2-2 compares the speed of the motherboard and processor for the different Pentium systems. You do not need to memorize this table for the A+ Exams; it is here to help you visualize clock multipliers as that is still being used in today’s systems.

TABLE 2-2 Pentium Clock Multipliers

Processor	Motherboard Speed (MHz)	Multiplier	Processor Speed (MHz)
Pentium 90	60	1.5	90
Pentium 100	66	1.5	99
Pentium 120	60	2	120
Pentium 133	66	2	132
Pentium 150	60	2.5	150
Pentium 180	60	3	180
Pentium 200	66	3	198
Pentium II	100	4.5	450

The Pentium processor has a 32-bit address bus, 32-bit registers, and a 64-bit data bus. It also has 16K of L1 cache that is divided into two 8K channels. One channel is for data cache and the other for application code cache.

Before the Pentium came along, processors used one instruction pipeline. This meant that when an application executed, it would run each stage of the application job one step after the other. For example, if an application has three lines of code, as seen in Figure 2-9, each line of code can be processed only after the previous line of code is fully completed. This creates a delay, or wait time, which slows performance.

FIGURE 2-9: Single-instruction pipelined processor executing application code.

The Pentium processor introduced a feature called superscalar design. The processor has two instruction pipelines, named U and V. Having two instruction pipelines enables the processor to execute two instructions at the same time. Thus, the three lines of program code, shown in Figure 2-10, can be quickly executed on a Pentium processor because lines 1 and 2 are processed at the same time, causing line 3 to be processed that much sooner. Notice that lines 1 and 2 execute parallel to one another; therefore, parallel processing is taking place.

FIGURE 2-10: Dual-instruction pipelined processor processing application code.

An application has to be designed to take advantage of two instruction pipelines. These applications are often labeled something like Pentium Aware or Pentium Ready.

Pentium Pro

In 1995, Intel released the Pentium Pro chip, which added a new level of performance to the Pentium processor. The Pentium Pro had all the characteristics of the Pentium processor — such as a 64-bit data bus and 32-bit registers — but it increased the address bus to 36 bits, which means that the Pentium Pro could access 64GB of RAM. The speed of the Pentium Pro ranges from 120 MHz to around 200 MHz.

The Pentium Pro included two additional features on its chip that helped it outperform the original Pentium. First, the Pentium Pro chip is really a two-chip team. One chip was the actual processor (with 16K of L1 cache, like the Pentium chip), but the other chip holds an extra 256K of cache memory. Because this cache memory is physically outside of the CPU, it is considered L2 cache.

The second feature that led to the performance gain of the Pentium Pro is dynamic execution, which has three stages: multiple branch prediction, dataflow analysis, and speculative execution.

Multiple branch prediction is the idea that the processor will look ahead and predict a number of instructions that might be needed in the very near future.
Dataflow analysis occurs when the processor looks at the instructions it has predicted will be needed next and then assigns them a logical order of execution.
Speculative execution is the actual execution of a given instruction based on the prediction and the order of execution assigned.

The Pentium Pro chip, shown in Figure 2-11, was implemented as a PGA chip that was placed in Socket 8.

FIGURE 2-11: The Intel Pentium Pro processor.

Pentium II

In 1997, Intel produced the Pentium II, which was really just an enhanced Pentium Pro with speeds ranging from 233 MHz to 450 MHz. The Pentium II had a 64-bit data bus, a 36-bit address bus (64GB of RAM), and 32-bit registers; and supported features such as MMX.

The Pentium II increased the amount of L1 cache that was integrated into the CPU to 32K, as opposed to 16K. The 32K of L1 cache was still divided into two equal channels: one 16K channel for data and one 16K channel for application code.

Intel packaged the Pentium II in the Single Edge Contact (SEC) — sometimes also referred to as the Single Edge Contact Connector (SECC) — that fits into Slot 1 on the motherboard. The SEC is a module enclosed in a casing or shell with two chips inside: one chip being the processor, and the other chip being the 512K of L2 cache. Refer to Figure 2-9 to see what a Pentium II processor, which uses the SEC, looks like.

Another enhancement that accompanied the Pentium II was single instruction multiple data (SIMD). To visualize how SIMD works, imagine five toddlers in a playroom, and that these toddlers are at the entertaining age of two — the age, of course, when the toddlers are preparing for their teen years by answering “no” to everything you say. You walk into the playroom and see that the five toddlers have found your box of darts and are throwing them at the walls. You are faced with a choice: You can either walk around to each child and explain why throwing darts at your walls is not a good idea (which means you will have to explain the same thing five different times), or you can have a good scream at the top of your lungs, which means that all the children will stop immediately and listen. SIMD works on the same basic principle. With SIMD, the processor gives the instruction to multiple processes at once instead of having to give the same instruction multiple times. Thus, the processor saves time and creates a much more efficient way to work with information.

Celeron

The Pentium II processor performed very well, and with all that cache memory, it should! Unfortunately, that performance came with a price. If you were not willing to pay that price, Intel created a chip for you: the Celeron chip!

The Celeron chip is nothing more than a less-expensive version of the Pentium II processor with the built-in L2 cache either removed entirely or reduced. The first-generation Celeron chip was code-named the Covington; it has no L2 cache memory on it. The second-generation Celeron was code-named the Mendocino, and it contains 128K of L2 cache. Although this version of the Celeron does have L2 cache, it is dramatically reduced from the Pentium II’s 512K so that it can be sold at a lower price.

The original Celeron shipped in an SEC package but also had a version that was packaged as a PGA, as shown in Figure 2-12.

FIGURE 2-12: Intel’s Celeron processor was first implemented as an SEC package, but later had a PGA chip that was placed in Socket 370.

Pentium III

The Pentium III processor shares many of the Pentium II characteristics. It supports dynamic execution (as the Pentium Pro also did) and MMX technology, has 32K of L1 cache, and has either 256K or 512K of L2 cache. The Pentium III runs at a speed of 450 MHz to 1000 MHz (1 GHz).

The Pentium III chip offers 70 additional instructions that are integrated into the chip, enhancing the user’s experience with 3-D graphic applications. The Pentium III chip also supports a number of low-power states to help conserve energy when the system is not in use. This processor is designed to run on either 100 MHz or 133 MHz motherboards.

Also note that there is a Pentium III version of the Celeron chip that runs as fast as the Pentium III processor but again has the L2 cache memory reduced. So now there are multiple versions of the Celeron chip — the PII version and the PIII version.

The Pentium III processor shipped in the SEC2 package (as shown in Figure 2-13) originally, but was then packaged as a PGA chip. The SEC2 goes in Slot 1, and the PGA chip is inserted into Socket 370.

FIGURE 2-13: The Pentium III processor in the SEC2 package that lives in Slot 1.

Xeon

The Xeon processor is built on the Pentium II and Pentium III architecture — meaning that, like the Celeron, there is a PII version and PIII version of the Xeon. The Xeon chip is designed for higher-end systems, such as server-class systems, and contains more cache memory than the typical PII and PIII. The Xeon comes in flavors of 512K, 1MB, and 2MB of L2 cache.

The Xeon can also address 64GB of RAM and is designed for multiprocessing systems: that is, computers with a motherboard that supports multiple CPUs. The Xeon processor has been designed to coexist with two, four, or eight CPUs.

The Pentium II Xeon and Pentium III Xeon chips were originally packaged as an SEC (as shown in Figure 2-14) that was placed in Slot 2, but later versions use the PGA and are placed in Socket 603. The Xeon chip also contains a thermal sensor that shuts the processor down if it starts to overheat.

FIGURE 2-14: The Pentium II Xeon processor was originally delivered in the SEC package.

The Celeron is a scaled-down version of the Pentium II or III processor, and the Xeon is a step up from the Pentium II or III. There are also PIV (Pentium 4) versions of the processors:

PIV Xeon: Designed to work with a multiprocessing system that uses one or two processors
PIV Xeon MP: Designed to work with a multiprocessing system that uses four or eight processors

Pentium 4

The Pentium 4 processor runs between 2 GHz and 4 GHz. The Pentium processor has 20K of L1 cache and 512K of L2 cache. The processor is shipped as a 423-pin or 478-pin PGA package, which means that the chip will be placed in Socket 423 or Socket 478 (as shown in Figure 2-15).

The Pentium 4 processor gets a huge performance benefit by being able to perform four data transfers in one clock cycle along the front side bus (FSB), which is the bus that connects the processor to system memory. (See Chapter 1 of this minibook.)

Most Pentium 4 processors today are multicore processors and enjoy the performance benefits that multicore brings.

Itanium and Itanium II

Intel created its first 64-bit processor in the Itanium and Itanium II processors and was marketed for server class–system or high-end PCs. Although the Itanium is a 64-bit processor and designed to run 64-bit software such as the 64-bit version of Windows, the Itanium can run some 32-bit code with the use of an emulator, but the code will run slower than if it were on a 32-bit processor. Special 64-bit editions of Windows can run on the Itanium processor, which enables you to take advantage of the 64-bit architecture.

The original Itanium processor used a special packaging known as the pin array cartridge (PAC), which uses 418 pins. The Itanium II was packaged in organic land grid array (OLGA) — a variation of the PGA — but the chip is located on a processor card (a circuit board that holds the processor). The OLGA fits into Socket 611.

The Itanium processor runs at around 1 GHz and contains a large block of cache memory: 32K of L1 cache, 96K of L2 cache, and 2MB or 4MB of L3 cache. The L3 cache is an additional block of cache memory located in the chip packaging.

Moving from 32-bit processors and applications to 64-bit versions would truly benefit any user using applications that are memory intensive or calculation intensive. For example, a user who works a lot with multimedia-type applications would see an improvement in performance.

Pentium M

For years, laptop manufacturers have been asking for smaller processors to place in laptop systems, and they finally have their wish. A number of processors have come out with the M version, which stands for mobile. The mobile versions of the processors are smaller than the processors that go in desktop systems, so they fit better and also use a lot less power. The benefit of using less power also means that they run much cooler.

Because the mobile versions of the processors use less power, they also run a little slower than their desktop counterparts.

Some popular brands of mobile processors are the Intel Pentium III M and the Pentium M. Intel’s big competitor, AMD, also has mobile versions of their processors: Athlon XP M and Mobile Duron. (Some manufacturers put the word mobile in the name of the processor instead of the letter M.) The next sections discuss more about AMD processors.

Intel Core 2

Intel designed the Intel Core 2 to be its 64-bit, multicore processor. The Core 2 comes in three flavors:

Core 2 Solo: Single-core processor
Core 2 Duo: Dual-core processor on the one chip
Core 2 Quad: Actually two chips, with two cores per chip, packaged in a multichip module

Core 2 processors range in speed from 1 GHz to around 3 GHz and fit into an LGA 775 socket, also known as Socket T. Core 2 processors come with either 2MB of L2 cache or 4MB of L2 cache.

Intel Atom

Intel has created a processor to run on the now-popular Netbooks. A Netbook is a laptop-like computer — but much smaller — that is used primarily for Internet usage. The Netbook is much smaller and cheaper than a regular laptop system and is marketed for email and web browsing features.

The Intel Atom has many characteristics of normal processors — it runs between 1 GHz and 2 GHz and contains 32K of L1 cache and 512K of L2 cache. Originally the Atom processor had one instruction pipeline, but there are versions with two instruction pipelines. There are also single core and dual core versions of the Atom processor.

Intel Core i3/i5/i7 processors

Intel has created a family of processors to satisfy lightweight system use or heavy system use that needs a powerful processor. The following are three common Intel processors found in new systems today:

Intel Core i3: The Intel Core i3 processor is the least powerful of the three processors and ships with an integrated GPU and two processor cores. The i3 typically has a 64KB L1 cache, 256KB of L2 per core, and a shared block of L3 cache of approximately 3MB.
Intel Core i5: The Intel Core i5 is the next step up with similar specs but with more power. The i5 processor can contain between two and four processor cores, with L3 cache memory ranging from 3MB to 8MB.
Intel Core i7: Intel’s highest-level processor is the Intel Core i7, which ships in different flavors as well. Depending on the model, the i7 ships with four or eight cores and an L3 cache ranging from 6MB to 12MB.

Note: The Intel Core i3, Intel Core i5, and Intel Core i7 processors are on their 6th generation now. When purchasing a system, be sure to know the generation of these processors. The newer generations have improvements to the architecture to make the system run faster. For example, the third generation Intel Core i7 uses Intel Turbo Boost Technology 2.0, which allows the processor to dynamically increase the processor speed when it needs it. Also note that the third generation of these processors integrates the graphics engine right into the processor chip to improve visual performance when gaming.

When upgrading a system, knowing the socket type can help you identify the generation of the Intel Core processor you are dealing with. Also knowing the generation of the processor can help you identify the socket type. For example, a Sandy Bridge Core i3 processor uses LGA 1155 but also does a Sandy Bridge Core i5 processor. This means if you have a motherboard with LGA 1155, you can upgrade the i3 to an i5 processor. Refer back to Table 2-1 and have a look at the Intel Core i3/i5/i7 sockets. Also note that Sandy Bridge is the code name for the second generation of the Intel Core processors. Each generation has a code name.

Don’t Forget Non-Intel Chips

One of Intel’s major competitors is Advanced Micro Devices (AMD). AMD has developed a family of processors that compete with Pentium-class processors. In this section, I provide an overview of some of the characteristics of the AMD processors.

K6

The AMD K6 processor was designed to compete with the original Intel Pentium. The K6 has 64K of L1 cache, supports MMX technology, and has built-in branch prediction techniques. This processor has 321 pins, which means that it will fit into a Socket 7–supported motherboard.

K6-2

The K6-2 processor was designed to compete with the Pentium II chip. It has 64K of L1 cache and 256K of L2 cache. The K6-2 also supports dynamic execution, MMX technology, and superscalar design.

The K6-2 has added 3DNow! technology, comprising a number of additional instructions integrated into the chip to improve 3-D graphics applications. The K6-2 chip also uses a 100 MHz motherboard speed, which is a big improvement over the 60/66 MHz motherboard speed that the original Pentiums used.

The K6-2 has 321 pins, which means that it will fit into a Socket 7–supported motherboard.

K6-III

The K6-III processor is designed to compete with the Pentium III chip. This chip shares many of the features of the K6-2, including a 100 MHz system bus. One of its features was the tri-level cache. Not only can it take advantage of an L1 and L2 cache but also an L3 cache that can be included on the motherboard.

Athlon

The AMD Athlon chip has 128K of L1 cache and 512K of L2 cache. It supports improved dynamic execution, MMX technology, and 3DNow! technology. The Athlon chip runs at speeds of up to 1.2 GHz and is designed to run on a 200 MHz system bus speed.

Unlike the K6-2 and K6-III, the Athlon is not a PGA-packaged chip that supports Socket 7. It uses its own socket type — Slot A, so called because the processor is packaged as an SEC. The Slot A socket is not compatible with Intel’s Slot 1, which means that users have to purchase a motherboard designed for the Athlon chip.

Later versions of the Athlon moved to the PGA package that has 462 pins. These PGA chips are placed in Socket A.

Athlon XP

After the Athlon chip was produced, Intel created the Pentium 4 chip. So AMD wanted to create a competing chip for the Pentium 4: namely, the Athlon XP. The Athlon XP is packaged as a PGA with 462 pins and is placed in Socket A. The Athlon XP runs at 2 GHz or more and contains 128K of L1 cache and 512K of L2 cache.

AMD markets these processors a little differently. Instead of labeling the processor with its speed, AMD labels it with its competitor’s speed. For example, the Athlon XP 1800+ is rated at 1.6 GHz but runs as fast as the Intel 1.8 GHz processor.

Duron

AMD wanted to create a processor that competed with each version of the Intel processors. So, if the Athlon XP competes with the Pentium 4, what competes with the Celeron? You guessed it — the Duron.

The Duron has 128K of L1 cache and 64K of L2 cache. This processor is packaged as a PGA with 462 pins, which means that it, too, goes into Socket A.

Opteron

Just like the Duron was built to compete with the Intel Celeron, AMD created the Opteron to compete with Intel 64-bit Itanium processors. The Opteron runs at about 1.8 GHz and contains 128K of L1 cache and 1MB of L2 cache.

The Opteron is packaged with a micro-PGA, which is made up of 940 pins and is placed in Socket 940. One of the major differences between the Opteron and the Itanium is that the Itanium cannot run 32-bit applications. AMD decided that the Opteron would run in a 32-bit or 64-bit mode, thus allowing it to run 32-bit applications.

Athlon 64 and Athlon 64 X2

The Athlon 64 — the successor to the Athlon XP — is the AMD 64-bit processor for desktop systems. The Athlon 64, with 128K of L1 cache and at least 512K of L2 cache built into the processor, is designed to compete with the Pentium 4.

Although the Athlon 64 is a 64-bit processor, it has been designed to be backward compatible; it can run 32-bit code. The Athlon 64 X2 is the dual-core version of the Athlon 64. The Athlon 64 family processors fit into a number of sockets: Socket 754, Socket 940, and the AM2 socket.

Know that the Sempron processor is a low-end version of the Athlon 64 and has replaced the Duron as the AMD low-end processor. The Sempron was originally a 32-bit processor with reduced cache size, but current versions are 64-bit with the cache size reduced.

Phenom and Phenom II

The Phenom and Phenom II pick up after the Athlon 64 and are the AMD triple- and quad-core processors. The Phenom, designed for desktop systems, uses the code names Toliman for the triple-core version and Agena for the quad-core version.

The Phenom processor, which comes with 128K of L1 cache and 512KB of L2 cache per core, also has 2MB of shared L3 cache. The processor fits in the AMD AM2+ socket and runs between approximately 1.8 GHz and 2.6 GHz.

The Phenom II also sits in the AM2+ socket and increases the shared L3 cache to 6MB! There were issues with the Phenom running on Windows Vista, but those bugs have been fixed for the Phenom II. The Phenom II has a triple-core version, code-named Heka, that runs between 2.6 GHz and 2.8 GHz. The quad-core version that is named Deneb runs between 2.5 GHz and 3.0 GHz.

Turion 64 and Turion 64 X2

The Turion 64 processor is the AMD 64-bit mobile processor for use in laptop computers. The Turion processor used to be called the Athlon Mobile 64, but AMD has moved to the new label of Turion 64 for its mobile processors. The Turion 64 is a single-core processor, and the Turion 64 X2 is the dual-core processor.

Turion 64 processors come with 128K of L1 cache and either 512K or 1024K of L2 cache. Both the Turion 64 and the Turion 64 X2 fit into Socket 754, and the newer chips fit in AMD Socket S7.

Be sure to know the multicore processors for the A+ Exams. The Athlon 64 X2 is the AMD dual-core processor, and the Phenom/Phenom II have triple-core and quad-core versions. Also be familiar with the Turion being the AMD 64-bit processor for laptop computers.

1. Which of the following best describes superscalar design?

(A) The processor is designed using only 3.1 transistors.

(B) The processor predicts the next few instructions to be executed and then determines the optimal order for the execution of these instructions.

(C) The processor has two instruction pipelines, which enables multiple instructions to execute at the same time.

(D) The processor works twice as fast as the motherboard.

2. You are upgrading the processor on one of the workstations for your customer. You remove a second generation Intel Core i3 processor. Which of the following can you upgrade to?

(A) Intel Core i5 (Lynnfield)

(B) Athlon X4

(C) AMD Opteron

(D) Intel Core i5 (Sandy Bridge)

3. You are upgrading an older system for your customer and notice that the system has Socket A on the motherboard. Which of the following CPUs are placed in Socket A? (Select all that apply.)

box (A) Pentium III

box (B) Athlon XP

box (C) Celeron

box (D) Pentium 4

box (E) Duron

box (F) Itanium

4. What chip type was the original Pentium processor packaged in?

(A) SEC

(B) PGA

(C) DIP

(D) LGA

5. You have a processor that contains four cores. Which of the following best describes the L3 cache with this processor?

(A) There is a block of L3 cache shared by all cores.

(B) Each core has its own block of L3 cache.

(C) The odd number cores share a block of L3 cache, while the even number cores share a different block of L3 cache.

(D) The first core has exclusive use of the L3 cache.

6. Which of the following acts as a storage container for information that will be processed by the processor?

(A) Data bus

(B) Address bus

(C) Registers

(D) Math co-processor

7. You have a new system running an Intel Core i5 processor. You are studying for the A+ Exams and need a few virtual machines to test the configuration of different operating systems. You install Hyper-V to create some virtual machines but then receive an error when you go to start a virtual machine. What could be the problem?

(A) EDB needs to be disabled in the BIOS.

(B) Create the virtual machines on a different drive than the operating system.

(C) Disable the integrated network card.

(D) Virtualization support needs to be enabled in the BIOS.

8. Which statement best describes the purpose of a math co-processor?

(A) Performs all logic functions on behalf of the processor

(B) Performs floating-point calculations on behalf of the processor

(C) Runs all applications, and a processor runs the operating system in a multitasking environment

(D) Allows for communication between devices

9. You are upgrading the processor on a system and notice that the socket is a Socket R. What socket type is this also known as?

(A) Socket 1155

(B) Socket 1150

(C) Socket 1156

(D) LGA 2011

10. You have replaced the processor in an old system, and you are now ready to finalize the upgrade. Which of the following is the first item to use to help draw heat away from the processor?

(A) Heat sink

(B) Thermal paste

(C) Fan

(D) Liquid cooling

11. You are looking to add a Pentium 4 processor to one of the systems in your office. What sockets do Pentium 4 processors fit into? (Choose two.)

box (A) Socket 370

box (B) Socket 423

box (C) Socket 478

box (D) Socket 920

12. What is the major difference between a Celeron processor and a Pentium 4 processor?

(A) The Celeron has more L1 cache memory.

(B) The Celeron has less L1 cache memory.

(C) The Celeron has more L2 cache memory.

(D) The Celeron has less L2 cache memory.

13. Which of the following CPU characteristics determines how much total memory the system can access?