Introduction to PC Hardware

You don’t need to know how a PC works in order to use it to browse the Web and update your Facebook page. However, the computer is a core part of modern life, whether you are at work or arranging your family photos. Those who are curious and want to learn more about technology should have a place to get started. This PCLT article assumes no technical background, but explains in simple language how the CPU, memory, disks, and other components work and how they have improved over time.

Build Confidence


Every PC contains chips, boards, and cards based on current technology and industry standards. None of these components are actually made by the company that sells you the machine. You get the same technology and approximately the same quality from every vendor. If you can get the vendor to repair anything that comes broken, there are few moving parts to wear out and the ones that do ($5 electric fans for example) are standard parts easily replaced. It is easy to learn how each component works and to understand enough about the technology to make intelligent decisions.

The Real Problems are Simple


Disks are one of the few moving mechanical parts of a computer. They will all fail eventually and some will break sooner. Back up all your personal data files regularly. Before sending your computer in for service, back up all your personal data files on a DVD, USB memory stick, or if you have a high speed network connection copy your data to an Internet backup service.

If you visit a web page and instead something pops up that claims that your system contains viruses or that they can make your machine run faster, it is a fraud. Web pages cannot read the disk or access the system to find out such things. Do not let any of these fraudulent sites or ads install any software on your machine, no matter what the promise you or what they threaten you with.

Who Made It?


Your disk came from Thailand or Malaysia. Your memory from Korea. Your case from China. Your mainboard from Taiwan. The CPU was made in Germany. All these components came from companies the consumer never hear about. The only thing that actually has the name of the company that sold you the computer may be the cardboard box it came in.

All you need is a Philips Screwdriver


There are a dozen components that are combined to make a desktop computer. Given the parts, you can assemble one in a couple of hours with just a Philips screwdriver. With practice, you can get it down to 20 minutes and still do a careful job.

Convenience or Performance?


You don’t buy a minivan based on how fast it can go 0 to 60, and you don’t need to measure a computer just based on its speed. You may find that the most important upgrade is a better keyboard and mouse so your hands feel better after several hours of use. Today, for much less than a computer monitor you can get a much bigger and higher resolution 4K “TV set” at Best Buy or Costco, and then all your documents will display in a much larger form that may be easier to read.

The Q Bridge


In New Haven, the speed of a morning commuter is determined by the backup to cross the river on the I-95 bridge and not how fast your car can go if you floor the accelerator. Internally, your computer has a different choke point depending on what you are doing, so there are different ways to speed up any task. To boot up quickly, get a high speed Solid State Disk. To edit video files, get a second hard drive. In many cases the cheapest performance improvement is more memory. Unless you are playing video games, the CPU is almost never the choke point.

Circuit Size, Voltage, and Heat


Every two years the size of a transistor on an integrated circuit is cut in half. Computer circuits are 1 or 0, “on” or “off”, full or empty. It is faster and requires less work to fill a shot glass with water than to fill a bath tub, and by analogy it requires less time, less power, and generates less heat to change the state of smaller circuits. Circuits also run faster at higher voltage, but that increases heat. Filling in the details of this relationship explains how chip manufacturing technology improves.

Clocks and Cycles


Each circuit performs some function. The design of the chip says how fast it should work, but small imperfections may slow down any individual circuit. A “clock” signal is distributed throughout the entire chip to tell the circuits when to start performing their next operation. For the chip to work reliably, the clock has to be slow enough to allow the slowest individual circuit to complete its operation. The only way to know how fast the entire chip can run is to extensively test all the circuits to find a speed at which everything works properly. Then to provide some protection, the chip is rated and sold to run 1/3 slower than the speed at which it passed the tests.

CPU Instructions


The programs you run (browser, word processor, games) are stored in memory. Each instruction is a string of bytes that identifies an operation (add, subtract, move, compare) and maybe the address in memory of some data the operation works on. For the first 15 years of personal computers, new generations of CPU chips executed programs faster by assigning specialized circuits to decode the instruction, fetch the data from memory, perform the operation, and store the result. Eventually, the computer was able to complete a typical instruction in a single clock tick. However, the delay between the time that the computer asks for data and the time that the memory can respond has not changed much. If the data is not already in the CPU chip cache, then getting data can require an amount of time equal to hundreds of internal CPU clock cycles. So during the second 15 year period, CPUs were designed to reorder the sequence of operations and to have a backup secondary “thread” of operations to which it can switch when one operation blocks waiting for memory.

Hyperthreading and Multi-Core


Up to the Pentium IV, CPU makers tried to make chips that ran faster. Then problems in the materials from which chips are made limited the maximum clock speed to around 3 Megahertz and that ceiling stuck for about a decade. If they could not make the CPU run faster, then (at least for some applications) the alternative was to add more CPU cores so the computer could do two, three, or four things simultaneously.

Memory and “Burst”


Each memory location has a big number called its address. The CPU sends memory an address value, waits for a period of time, and then reads the data that the memory chip sends back. The delay between the request and the response is called “latency” and it really hasn’t changed much in the last couple of decades. However, memory is not designed to fetch individual bytes. When you send memory an address, it really fetches a block of (say) 4096 bytes of data. Since after reading data at one particular address there is a high probability that the CPU will shortly need data next to that address, the CPU actually reads a burst of 32 or 64 bytes of data at a time. The speed of this burst of data transfer has been made faster.

Memory is advertised by how fast it can transfer data during the burst. However, during normal operation the CPU will jump around from location to location, and so the much slower latency number determines how fast the memory actually responds.

The Mainboard (Motherboard)


The Mainboard contains a socket for the CPU chip (Intel or AMD), PCI/PCIe slots for adapter cards, and connectors for the USB, Firewire, audio, and SATA disks.

Most of the function in the mainboard is provided by two large chips that are collectively called “the chipset”. The chipset and socket have to match the brand (AMD or Intel) and the generation of CPU chip you are using and the type of memory (DDR2 or DDR3) you plug in.

The least expensive mainboards typically have an integrated video controller that, on new systems, is perfectly adequate for business applications, browsing the Web, and even watching videos and movie disks. Match it with a simple dual core CPU chip. You will have to buy a more expensive mainboard to get all the electric power needed to drive systems with four CPU cores or more than one video card.

Hard Disks and CD Drives


The hard disk is a mechanical device that rotates and has moving arms. Its performance is thus limited by physical rather than electronic considerations. Although the disk is designed to randomly access files anywhere on its surface, the arms have to jump back and forth when you are reading from one file in one place and writing to a second file on another part of the disk. New flash memory Solid State Disks (SSDs) don’t have moving parts and provide faster access at a supportable price to 32 to 64G of randomly accessed, read-mostly data. Put the operating system here. Then one or maybe two big conventional disks for bulk data storage, such as your video files.

On a laptop, the single slow speed disk may be the performance choke point. Unfortunately, unlike a desktop there is no room to add a second disk. In places where you frequently use your laptop, like home or work, you can add an external conventional disk.

Disk Partitions

A hard disk can be divided into partitions that hold different operating systems or data with different performance or recovery characteristics. When planning the use of multiple disks or migrating your system to a newer disk, it is important to understand the different types of disk partitioning to make the right choices.

Video and Monitors


A computer monitor is basically the same technology as your big flat screen HD TV set, but it is optimized to be sitting two feet away instead of across the room. Because of the common technology, modern computer monitors tend to have resolutions close to, and maybe slightly greater than, the two types, 1280x720 (720p) and 1920x1080 (1080p), of HDTV sets.

In the last few years the integrated video on the mainboard of inexpensive systems has become perfectly adquate for Web browsing and even Blu-Ray movie viewing. Expensive video cards are only useful for gaming.

The Bus


Inside the computer the CPU has to connect to the memory, disks, LAN, and adapter cards. Then outside the computer it has to talk to the keyboard, mouse, and external disks. In the first PC there was a single bus that connected everything together. Over time, specialized circuits and cables have developed to connect different types of devices at different speeds.



Ethernet on the mainboard or through a plug in card connects the computer to other computers and to the internet. Wireless Ethernet is convenient but has a lower maximum speed than wired.

External Devices: USB or e-SATA 


USB connects slow speed (keyboard, mouse, printer, scanner) and medium speed (DVD, TV tuner, Web cam) devices to your desktop or laptop. External SATA allows a SATA disk to be stored in a cabinet external to the computer and yet operate at the full 300 MB/s speed of an internal disk. The USB 2.0 standard in wide use today hits a maximum data transfer rate at about 40 MB/s, but a new USB 3.0 standard is appearing that will allow external disks to connect at full speed. However, since you can already do this with e-SATA, you do not have to wait for USB 3.0 to become widely available.

Duo to Quad


A Dual Core laptop or desktop is perfectly adequate for business and ordinary home use. A third or fourth core only helps out when you are doing specific tasks, like encoding video. Servers support hundreds of concurrent users, and so they can use as many cores as they are given. However, a laptop has only one user and most applications can only use one core.

The GPU (Specialized Processors)


The CPU is a general purpose processor that can run any kind of program and perform any function. You need one CPU and maybe can use a second on a desktop or laptop. Beyond that, the only need for additional processing power is during specialize tasks (video games, encoding recorded HDTV programs) where there are a small number of intense operations that must be continouously repeated. Instead of adding additonal general purpose processors, Intel and AMD can add specialized processors that just perform the specific functions needed to perform the few applications in which an ordinary home user can require extra power. Because specialized processors can be 10 or 20 times as powerful as a general purpose processor at performing these specific functions, this is a resonable design.