Install and Upgrade CPUs

Integrated GPU

. Most computer systems provide some sort of built-in graphics adapter. Initially, an integrated GPU would be implemented as part of the motherboard chipset; Intel's Graphics Media Accelerator, for instance. Nowadays, it is more likely that an integrated GPU, or Integrated Graphics Processor (IGP), will be part of the CPU (Intel HD Graphics, for example). Apart from cost, an IGP is more power-efficient than a dedicated card. Some laptop systems with both an IGP and a dedicated card are capable of switching automatically between them (nVIDIA Optimus and ATI Hybrid Graphics technologies), depending on whether an application requires advanced 3D performance or not, to conserve battery life.

Heat Pipes and Spreaders

A heat pipe is a sealed tube containing some type of coolant (water or ethanol). The liquid close to the heat source evaporates then condenses at a cooler point in the pipe and flows back towards the heat source. The cool parts of the pipe are kept so by convection. This mechanism is more effective than a simple heat sink and fan assembly. It is necessary for a CPU that runs particularly hot or where there is not much space for airflow within the chassis. A dual heat pipe has two tubes, providing better cooling. A heat spreader uses the same design but is a flat container rather than a pipe. This design is better suited to portable computers. If used without fans, heat pipes and spreaders are classed as passive cooling.

Heat Sinks and Thermal Paste

A heat sink is a block of copper or aluminum with fins. As the fins expose a larger surface area to the air around the component, a greater cooling effect by convection is achieved. The heat sink is "glued" to the surface of the chip using thermal paste (also referred to as thermal grease or thermal compound) to ensure the best transfer of heat. At the microscopic level, when two solids touch, there are actually air gaps between them that act as insulation; the liquid thermally conductive compound gel fills these gaps to permit a more efficient transference of heat from the processor to the heat sink. A heat sink is a passive cooling device. Passive cooling means that it does not require extra energy (electricity) to work. In order to work well, a heat sink requires good airflow around the PC. It is important to try to keep "cable clutter" to a minimum.There are various mechanisms for clamping a CPU heat sink to the motherboard. There may be a retaining clip or push pins. Push pins can be released and reset for insertion by making a half turn with a screwdriver.

CPU Installation Considerations

Before you replace a processor, you need to make sure you select a processor that matches the type of socket on the system board. Not all processors that use a particular socket will be compatible with your system; this is just one of several items you will need to check for compatibility. Also, when it comes to removing the CPU, there are several cooling device designs and socket types to deal with. If you are upgrading the CPU, check that the new model is supported by the motherboard. Use the motherboard manufacturer's website to get up-to-date information (for example, to find out about CPU models that were released after the motherboard's documentation was written). Also check for any BIOS updates. Note: Just because a motherboard has the correct socket type does not mean that a CPU model will be compatible. The motherboard must have a compatible chipset and voltage regulators, too. Pin 1 on the processor MUST match pin 1 on the processor socket. Both the processor and the socket carry distinguishing markings to indicate pin 1. On a processor, this may be: A beveled corner or a white dot printed in one corner of the processor. A square, rather than round, joint where one of the pins is connected to the underside of the processor. A "spur" on one corner of the gold patch on the underside of the processor. On a processor socket, this may be: A difference in the pattern of pin holes in one corner. A "1" printed on the motherboard next to one corner.

Superpipelining.

CPUs process multiple instructions at the same time (for example, while one instruction is fetched, another is being decoded, another is being executed, and another is being written back to memory). This is referred to as a superscalar architecture, as multiple execution units are required. Superscalar architectures also feature longer pipelines with multiple stages but shorter actions (micro-ops) at each stage, referred to as superpipelining. The original Pentium had a 5-stage pipeline; by contrast, the Pentium 4 has up to 31 stages (NetBurst architecture). NetBurst actually proved relatively inefficient in terms of power and thermal performance, so Intel reverted to a modified form of the P6 architecture it used in Pentium IIs and IIIs for its "Core" brand CPUs (with around 14 stages).

Clock Speed

Despite the architectural features just discussed, the speed at which the CPU runs is generally seen as a key indicator of performance. This is certainly true when comparing CPUs with the same architecture but is not necessarily the case otherwise. Intel Core 2 CPUs run slower than Pentium 4s, but deliver better performance. Budget and low power models will work at around 1-2 GHz while premium models will run at 3-4 GHz. The core clock speed is the speed at which the CPU runs internal processes and accesses L1 and L2 cache (L2 cache access speed actually depends on the CPU architecture, but full-speed access to L2 cache has been standard for some time). The Front Side Bus (FSB) speed is the interface between the CPU and system memory. Despite the architectural features just discussed, the speed at which the CPU runs is generally seen as a key indicator of performance. This is certainly true when comparing CPUs with the same architecture but is not necessarily the case otherwise. Intel Core 2 CPUs run slower than Pentium 4s, but deliver better performance. Budget and low power models will work at around 1-2 GHz while premium models will run at 3-4 GHz. The core clock speed is the speed at which the CPU runs internal processes and accesses L1 and L2 cache (L2 cache access speed actually depends on the CPU architecture, but full-speed access to L2 cache has been standard for some time). The Front Side Bus (FSB) speed is the interface between the CPU and system memory.

Chip Level Multiprocessing (CMP).

However, improvements in CPU manufacturing techniques have led to another solution: dual-core CPUs, or Chip Level Multiprocessing (CMP). A dual-core CPU is essentially two processors combined on the same die. The market has quickly moved beyond dual-core CPUs to multicore packages with 3, 4, 8, or more processors.

Install and Upgrade CPUs OBJECTIVE COVERED Given a scenario, install and configure motherboards, CPUs, and add-on cards.

In this topic, you will examine the types and features of CPUs and cooling systems. Much like the motherboard, the CPU is another important component of the computer system that actually carries out all the tasks requested by the applications installed in the computer. The CPU is a heat generator, so part of understanding the CPU includes understanding how to manage heat inside the computer case by managing the airflow and temperature. Keeping the system cool is an easy but important way to maintain or even increase its productivity. A computer that runs too hot risks damaging its own components. As an A+ technician, you need to be familiar with these essential components of the computer system.

Intel CPU Ranges and Socket Types

Intel uses Land Grid Array (LGA) form factor CPUs that use notches for easy installation. In LGA, the pins that connect the CPU and socket are located on the socket. This reduces the likelihood of damage to the CPU but increases the chance of damaging the motherboard.

Fans

Many PCs have components that generate more heat than can be removed by passive cooling. A fan improves air flow and so helps to dissipate heat. Fans are used for the power supply and chassis exhaust points. The fan system will be designed to draw cool air from vents in the front of the case over the motherboard and expel warmed air from the back of the case. Note: A common implementation is to include air vents near the bottom of the front of the case and to place a fan near the top of the rear of the case to pull cooler air through the system. Typically, the speed of the fan is varied according to the temperature, and sensors are used to detect whether a fan has failed. Smaller fans may be used to improve the performance of the heat sink on the CPU, GPUs, and even hard disks.A fan is an active cooling device. It requires power to run. The main problem with fans, especially at the lower end of the market, is that they generate noise. A fan also needs to be matched to the CPU model to ensure that it is powerful enough to cope with the processor's thermal output. Most CPU fans are designed to be removed without the use of tools. Usually the fan assembly will have clips and a power connector. Some chassis designs incorporate a plastic shroud or system of baffles to cover the CPU and channel the flow of air. The shroud is usually attached to the case using plastic clips.

CPU Manufacturing Process

Note: This information is provided for reference; it is not part of the exam objectives. A microprocessor is a programmable integrated circuit (IC). An IC is a silicon chip embedded on a ceramic plate. A silicon chip is a wafer of purified silicon doped with a metal oxide (typically copper or aluminum). This doping process creates millions of transistors and signal pathways within an area called the die. These transistors provide the electrical on/off states that are the basis of binary computer systems. The process used to create the transistors is referred to as an n-micron or n-nanometer (nm) process, reflecting the size of the features (a transistor for instance) that can be created. A micron is a millionth of a meter; a nanometer is a billionth of a meter). This process has developed from 1 micron (80486) to 0.014 micron (or 14 nm). Scaling down the process allows reduced voltages and therefore more speed with less heat. It also allows more components to be added to the same package, which has enabled innovations such as on-die cache, multicore CPUs, and on-die graphics processors.

AMD CPU Ranges and Socket Types

Older AMD brands such as Athlon, Phenom, Sempron, and Turion have been phased out over the last few years. The following brands represent the company's Zen microarchitecture in different segments: Ryzen/Threadripper and Ryzen Mobile—this brand now represents AMD's pitch for the high-end enthusiast segment, replacing older AMD FX chips. Epyc—AMD's server-class CPU brand, replacing its long-standing Opteron series of chips. AMD uses Pin Grid Array (PGA) form factor chips, designed to fit in a Zero Insertion Force (ZIF) socket on the motherboard. As the name suggests, a PGA chip has a number of pins on the underside of the processor. These plug into corresponding holes in the socket. Care must be taken to orient the CPU correctly with the socket and to insert it so as not to bend or break any of the pins.

Hyperthreading

One way to make instruction execution more efficient is to improve the way the pipeline works. The basic approach is to do the most amount of work possible in a single clock cycle (multitasking). There are various ways to achieve this goal, though.

Liquid-Based Cooling Systems

PCs used for high-end gaming (those with twin graphics cards, for instance) and with overclocked components may generate more heat than basic thermal management can cope with. PCs used where the ambient temperature is very high may also require exceptional cooling measures. Figure: A liquid-cooled PC. Liquid-based cooling refers to a system of pumping water around the chassis. Water is a much more effective coolant than air convection and a good pump can run more quietly than numerous fans. On the downside, liquid cooling makes maintenance and upgrades more difficult, requires comparatively more power to run, and is costly. Liquid cooling is an active cooling technology as the pump requires power to run.

Fans and Power

Power is supplied to a CPU or case fan by connecting its power connector to an appropriate header on the motherboard (make sure you plug the CPU fan into the header marked "CPU Fan" to ensure that the chipset can run the fan at an appropriate speed). Power connectors and headers for fans are 3-pin or 4-pin. 3-pin models control fan speed by varying the voltage. 4-pin models control fan speed by switching the voltage on and off (using a Pulse Width Module [PWM] signal carried by the fourth wire). This gives better control over fan speed. Fans with a 3-pin connector can usually be used with 4-pin headers but the system may not be able to vary the fan speed (or may need special configuration to be able to do so). A fan with a 4-pin connector will usually work with a 3-pin header but will not be able to use PWM.

Power Management (Throttling)

Rising energy costs and environmental legislation are placing power efficiency at the top of the agenda for IT buyers. In terms of CPU performance, more speed means greater power consumption and heat production. To deal with these issues, CPUs can implement power management to enter lower power states, referred to as throttling. Another aspect of power management is protection for the CPU. If a processor runs too hot, the system can become unstable or damage can occur. CPUs provide routines to reduce performance to protect against overheating.

CPU

The Central Processing Unit (CPU), or simply the processor, executes program instruction code, performs mathematical and logical calculations, and controls Input/Output (I/O) functions. The CPU is commonly described as the "brains" of a computer; in fact, it is better thought of as a very efficient sorting office. The CPU cannot think, but it can process simple instructions very, very quickly and efficiently. A computer is only as "clever" as its software. PC processors are produced by Intel or other manufacturers who use the Intel instruction set and whose processors are therefore IBM PC (or x86) compatible. Currently only AMD (Advanced Micro Devices) falls into this category.

Pentium

The Pentium used to be Intel's premium 32-bit CPU brand and you may still find Pentium 4-based computers in use. The Pentium brand has been reintroduced to represent "mid-range" CPU models based on the Core microarchitecture

CPU Packaging and Compatibility

There have been numerous CPU architectures, and within each architecture, a number of different models, and for each set of models, a brand to position them within a particular market segment. CPU packaging refers to the CPU's form factor and how it is connected to the motherboard. Intel and AMD use different socket types so you will never be able to install an AMD CPU in a motherboard designed for an Intel CPU (and vice versa). Additionally, within Intel's and AMD's own ranges, a given CPU socket type will only be compatible with a fairly limited number of CPU models. The following tables summarize some of the various CPU models and socket types that have been used over the years. Note that the supported desktop processors and memory are illustrative rather than definitive. For more up-to-date information, visit a site such as CPU World, Tom's Hardware, or AnandTech.

atom

This brand designates chips designed for low-power portable devices (smartphones and tablets).

Xeon

This brand is aimed at the server/workstation market. Current Xeons are often differentiated from their Core counterparts by supporting n-way multiprocessing and ECC memory and coming with larger caches.

Celeron

This has long been Intel's budget brand.

Core

This is Intel's flagship desktop and mobile CPU series. The earliest models (Core Solo and Core Duo) were laptop-only chips. The Core 2 series introduced desktop versions plus 64-bit and multicore support. The current range is divided into Core i3, i5, and i7 brands, with i7 representing the best performing models. The Core iX range has been based on successive generations of microarchitectures, named Nehalem, Sandy Bridge, Ivy Bridge, Haswell, Broadwell, and Skylake.

Other CPU Features

Two other features of modern CPUs need to be covered here. These support the use of virtualization and power-efficient graphics capability.

Virtualization extensions.

Virtualization software allows a single computer to run multiple operating systems or Virtual Machines (VM). Intel's Virtualization Technology (VT) and AMD's AMD-V provide processor extensions to support virtualization, also referred to as hardware-assisted virtualization. This makes the VMs run much more quickly. These extensions are usually features of premium models in a given processor range. There is also a second generation of virtualization extensions to support Second Level Address Translation (SLAT), a feature of virtualization software designed to improve the management of virtual (paged) memory. These extensions are referred to as Extended Page Table (EPT) by Intel and Rapid Virtualization Indexing (RVI) by AMD.

Overclocking

When a manufacturer releases a new chip, it sets an optimum clock speed based on systems testing. This clock speed will be set at a level where damage to the chip is not likely to occur during normal operation. Increasing this speed (overclocking) is done using the system setup firmware program by adjusting the CPU Speed or Advanced Chipset Features properties. You can either increase the core clock speed (multiplier) or the FSB speed (overclocking the memory chips) or both. Increasing the clock speed requires more power and generates more heat. Therefore, an overclocked system must have a suitable power supply and sufficient cooling. The operating environment (the warmth of the room and build-up of dust) must also be quite carefully controlled. Overclocking is generally performed by hobbyists and games enthusiasts but it is also a means to build a PC more cheaply by specifying lower cost components, then boosting their performance. Without cooling, overclocking increases the risk of thermal damage to components and may increase the frequency of system lockups. It also invalidates the warranty. Original Equipment Manufacturers (OEM) generally try to prevent overclocking in their PC systems by disabling custom settings in the computer's system setup program. A CPU may also run at a lower actual speed than it is capable of if it is put in a power saving mode.

Symmetric Multiprocessing (SMP).

Yet another approach to making a computer system faster is to use two or more physical CPUs, referred to as Symmetric Multiprocessing (SMP). An SMP-aware OS can then make efficient use of the processing resources available to run application processes on whichever CPU is "available." This approach is not dependent on software applications being multithreaded to deliver performance benefits. Traditionally, SMP was provided by physically installing two or more CPUs in a multi-socket motherboard. Obviously, this adds significantly to the cost and so is implemented more often on servers and high-end workstations.