Hardware: What Parts Does A Computer Need?

Anytime we build something as a condition-based cause-and-effect, we’ve made a “computer”, though using household items usually never represents more than a few “bits” of memory, and it can’t be used for a broad range of purposes.

If someone can physically observe or measure something, they can use a computer to magnify their purposes. Technologists tend to go too far and veer into making computers a religion, but it’s irrefutably true that computers have and can improve any aspect of daily living that can distilled into math.

All computers are simply electronic math machines, built up from logic using two-based numbering and various forms of programming to accomplish specific purposes.


Computers are much older than people realize. The oldest one we know of is the Antikythera mechanism, made ~150 BC, but adding machines have been around since the 1800s, with the most famous one being the difference machine.

Computers had been mechanical up until World War II, where they became electronic to keep pace with the need to decrypt German military messages. The first electronic computer was named Colossus, and was dramatically faster than the mechanical implementations that came before it.

The mathematician David Hilbert asked a simple question in 1928, “Is it possible to have a method/procedure that will clearly answer all math questions?” Alan Turing had a theory called the Universal Turing machine, which was designed to answer that question.

Now, most everyday computers use the Von Neumann Architecture, based on the Universal Turing machine theory, and it was first implemented by Konrad Zuse’s Z3 in 1941:

  • Input – something goes into it
  • Processing – it does logical, conditional stuff with that input
  • Memory – it stores and retrieves information
  • Output – it yields some final results from all that processing

Across the years, computer “abstractions” typically stay about the same while its “implementations” trend insanely fast through various technologies:

  • Right now, computers use electricity through silicon (0s and 1s represented as positively and negatively charged electricity).
  • At one time, the 0s and 1s were electric charge held within vacuum tubes.
  • Originally, computers were mechanical plates that read holes punched into old 35mm film stock.
  • In decades, it may be something else, such as quantum superstates or DNA.

In fact, most of the standard computer needs were demonstrated already in a 1968 demo by Doug Engelbart’s team.

Design: Hardware

Since most computers run at the speed of electricity, the input-to-output in modern computers feels like it transfers information instantly. However, on the back end, computers are doing many types of calculations based on logical conditions which were configured in advance by engineers.

Technically, a computer could be assembled with a vast agglutination of wires and transistors, but that’s messy. To make it simpler, a “printed circuit board” (PCB) will contain all the wires in pre-configured arrangements. Then, by using a standardized plug or solder point, just about anything can be easily attached to the PCB.

Around the core wiring/logic, the rest is largely protective. Padding, a housing, screws, and metal shielding protect it from getting broken, wet, picking up static or (in some cases) interference from radio waves. For higher-power needs (e.g., a PSU), they can fill the rest of the space with a brown epoxy-like substance called “potting” around the wires to provide protection from electric shock, vibration, or moisture.

In practice, this means a hacker can cobble together just about anything that’s conductive and attach it to anything. With the exception of circuitry that’s too small to work with (e.g., cell phones), most low-grade electronics are very accessible (such as the Raspberry Pi).

Design: Software

With the exception of logic gate primitives, anything can be assembled as software on logic gates, and further hardware design simply speeds up how fast the computer will perform.

In simple terms, computers are vast collections of logic gates that build into ever-more complicated things. Combined together, it makes working with them extremely trivial.

Designers are constantly improving computers, but the gains in improvement are starting to become incremental (similar to the autos that came before it). Even though they can become much faster, they’re not really adding much to the experience anymore for a typical user, and even a low-tier desktop computer is more than enough for most non-gaming/design computer needs until VR headsets or AI become a popular trend to everyday people.


Inputs can be pretty much anything you can stick a sensor to and measure. That sensor must then do one of two things:

  1. Activate something after reaching certain thresholds (aka “analog signal”)
  2. Convert that information into a string of 0s and 1s (aka “digital signal”)

Conventionally, we’re used to keyboards and mice, but pretty much anything can be an input:

  • Barcode readers
  • Touch pads
  • Body gestures
  • Speaking
  • Heat or moisture


The memory holds “programs” that tell the processor what to do.

It also holds the results of that processing for further use.

If that information is moving to the processor faster than it can process (which very often happens) then it’ll offload the information in its memory.

This memory takes a wide variety of forms, from USB sticks to hard drives to the memory built into the CPU chip. Each type of memory has pros and cons.

The memory to store processed information has gone through many iterations of implementation. Vinyl records, magnetic tape, compact discs, magnetic disk drives, and solid-state memory are all examples of long-term storage.


The processor is where the magic happens. By using a set of very clear rules, a processor can convert the input data in endless ways.

A processor cycles the information between a control unit and math/logic unit.


In effect, output is pretty much anywhere that the processor sends information to:

  • It could send it to a computer screen or a printer, which is what most people think of.
  • It may output to audio or activate lights.
  • It might store it to the hard drive for long-term storage between power cycles.
  • It may send the information across an electrical cable to be another computer’s input.

General-Purpose Computers

Many random geeks engineered what we now call a computer. But, we tend to use that word in a more limited scope than reality.

The cheapest alarm clocks and refrigerators now use computers, but those computers are specific-purpose. It’s usually something small like activating an alarm when the time matches the alarm time or sending an electrical signal that turns a compressor on or off.

On the other hand, a “general-purpose” computer can do nearly anything based on logic or math. If you want, you could program your desktop computer to control your thermostat or run a timer. Sometimes it won’t be as reliable, but it can do just about anything you want it to do.

Because they’re so useful, manufacturers have adopted a relatively standard “modular” format for selling general-purpose computers.


The biggest enemy of computers is dust, for several reasons:

  1. A layer of dust is like a blanket on the computer, and a case keeps most dust away.
  2. Computers, specifically the processors, generate a lot of heat, so you need something to reliably hold the cooling system.
  3. Since computers operate on electricity-based logic, a static charge can ruin a computer or fry the system, and dust often builds up static electricity.

Thus, a case is critical to fight dust, though it doesn’t need to be fancy. It also should be big enough to hold everything you need on it. Typically, there’s room to feed cables toward the back to prevent them from blocking air flow.

Apparently, building a PC has become cool, so tempered glass and lighting is an accented up-sell.

To make the device waterproof, standard waterproofing engineering applies for most of the device (e.g., seals, gaskets), but the ports themselves typically have a sensor that kills all electrical power from running through them. No electricity means the electronics are safe as long as no power runs through it before it’s dry again.

Power Supply

Computers process with DC (“direct current”) instead of AC (“alternating current”). But, your typical household plug runs on AC, which tends to fluctuate all over the place.

A “power supply” serves two roles:

  1. Convert AC power to DC.
  2. regulate the flow of electricity for what the computer needs (“continuous wattage”), measured as a wattage rating.

Conveniently, batteries are designed to hold DC power already. Depending on how the circuit runs into the computer, batteries are separate power supplies because they regulate the wattage.

The power supply can sometimes be fragile, especially with events like a lightning storm. It’s a really good idea to have a power strip or a UPS (“uninterruptible power supply”) with a backup battery to keep a steady flow of electricity.

An average computer uses 200-300W (i.e., 0.2-0.3 kWh/hr), though it can go as high as necessary. It’s worth noting that more electricity generates more heat and loses its efficiency, meaning that a standard computer needs about 20% more wattage than you think it’ll need (i.e., get a 350W power supply for a standard computer just to be safe).

Power supplies can be non-modular, semi-modular, and fully-modular. Non-modular has all the cables you’ll need, semi-modular has only the cables you’ll need with options to plug in more, and fully-modular is a box without cables. This is almost strictly a cosmetic thing if you’re not trying to do something elaborate, but many people building PCs will attach their identity to how their computer looks.

Most power cables for laptops and other smaller tech have a “stepdown transformer” to lower the voltage that’d normally come from a household power plug (120-240V, depending on your country), since they don’t need much (never more than 12V).

Amusingly, if you use a power plug adapter for your car to power your laptop, you’re converting from AC (to crank the car’s starter) to DC, back to AC inside the computer.


The CPU and power supply generate the most heat, so they need cooling.

Conventionally, both the CPU and power supply have had a fan to send heat away:

  • Positive air flow is pulling (theoretically) cooler air into the case.
  • Negative air flow is pulling warmer air out of the case.
  • Neutral air flow is a combination of both positive and negative air flow.

The fans that usually come with a preassembled computer are sufficient for most purposes.

  • Adding more can’t hurt, assuming you can maintain a healthy air flow and don’t overload your power supply.
  • However, more fans guarantee more noise.
  • The shape of the fan’s grille pattern also makes a difference in noise. In general, wire and mesh grilles attached to the housing are the best for noise, and a swirl pattern directly engraved into the housing comes as a close second.

The more air that runs across a surface, the more friction it creates, and it ends up with the engineering constraint of a boundary layer that moves rather slowly near a surface. The only solutions that really work to offset it are to either blast the air directly onto the surface, or blast it rapidly near the bottom with an extreme aerodynamically-enhanced surface.

The latest in cooling technology involves running water through pipes over the CPU and fan, a bit like a car’s radiator.

  • In fact, some of the most elaborate and advanced computers involve running water inside the CPU’s chipset.

If you’re particularly bold to try it, and the motherboard supports it, you can overclock the computer in the BIOS settings. This makes the computer run faster, but increases heat output everywhere (especially the CPU), so make sure you have enough cooling beforehand and know exactly what you’re doing or the heat will literally melt the processor.

On the other hand, if the BIOS supports changing the clock settings, you can also underclock the computer, which might be useful if you want to prolong the life of your computer and don’t need a computer to be relatively fast.


A motherboard is a “printed circuit board” that connects all the parts together on a really small scale. It also usually has a “clock” with a “CMOS battery” to track time and a “BIOS” chip (“basic input-output system”) that runs before the computer “boots” up. They use a variety of connectors with various voltages (mostly 12V, but also 3.3V and 5V), and connect large buses via the 20+4 pin connector (which permits both 20-pin and 24-pin buses to connect).

Motherboards are usually the most complicated-looking and largest part of a computer, but are often much less expensive and complicated compared to the CPU, though they can often have many components built-in for increased functionality or reliability.

Each motherboard has specific limits to its “form factor”. Specifically, the bigger the motherboard, the more chips in its “chipset”, the more stuff you can put on it, and the bigger your case has to be. In order of size:

  • Nano/Pico/Mobile ITX (best for small embedded systems)
  • Micro ATX (standard for many generic PCs)
  • ATX
  • EATX (Extended ATX)
  • XL ATX
  • (they get larger for enterprise-grade needs)

Motherboards have a variety of “rear ports” to plug other things in with the case closed. You can usually see these on the back of your computer, but they often stick a few through the front for convenience:

  • USB (“universal serial bus”) plugs, which are pretty much the universal standard right now, with different versions that have different speeds, and a few types of plugs.
  • Display ports for the screen, which can include HDMI (“high definition multimedia interface”), VGA (“video graphics array”), and DVI (“digital video interface”).
  • “PS/2” port (from IBM’s “Personal System 2”) for keyboard/mouse, though this has largely been replaced by USB.
  • Digital/analog sound cables. Unless you’re doing something fancy, lime-green is for audio out and pink is for your microphone.

When you open the case, the motherboard has a wide variety of connectors:

  • “CPU socket”, which holds a CPU (absolutely necessary, and must match the motherboard)
  • “SATA” connectors (only as necessary as the things you need, though “hard drives” have been popular for a few decades)
  • The “RAM slots”, which hold short-term memory (really necessary to have at least 1 if you don’t want The World’s Slowest Computer).
  • Expansion slots, which permit you to expand what you can do (only as necessary as you need).
  • Connectors that go to the “power supply” (one for the CPU, one for the motherboard), the front panel (separately for USB plugs and the power button), and fans.

Motherboard sizes generally use the ITX/ATX standard (in order from smallest to largest: Micro-ITX, Mini-ITX, Mini-ATX, Micro-ATX, ATX, EATX). Make sure you have the correct case for your motherboard: if the motherboard is too large it may not fit, and if it’s too small you may need to get creative with cable ties.

Further, each motherboard has specific size requirements for all the parts that go on it. The pieces snap on like “LEGO” if they’re the right size, but if they don’t fit you’ll need extreme tech skills to get it to run (though it’s rarely impossible if the components themselves work fine).

There’s no substitute for experience to tell why a motherboard won’t detect a component:

  • Sometimes the motherboard and component don’t mesh because of how much newer one of them is, even if it fits fine.
  • Sometimes the component or that specific port on the motherboard was manufactured incorrectly.
  • The computer may be running on low power or have an issue with power draw.
  • Some other weird thing that happens 1 in 1,000 times, but only someone who works with the troubled computers would have experience with.


The CPU is where the computer processes information. It fits into a “CPU socket” on the motherboard. Because it runs so hit, it needs a fan sitting right on top of it, and it also needs a dab of heat-conductive “thermal paste” to get the heat over to the CPU fan.

In a modern computer not designed for cheapness, that CPU has multiple “cores”. Those cores are all CPUs in their own right, but are part of the singular chipset.

While there are a ton of various features (e.g., hyper-threading to help with VMs), most of them aren’t particularly relevant for you unless you know exactly what you’re getting into. However, make absolutely sure you’ve got the right socket. CPU sockets are constantly changing every 2-3 years, and a CPU can not work in a different socket, so buy precisely what you need.


The RAM is the “working memory” of the computer. They usually slide into the motherboard into conspicuously nearby slots to the CPU. Abstractly, it functions the exact same as a storage drive, but is much faster and gets wiped as soon as you kill the power to the computer (which is often why rebooting can often fix your computer).

Generally, RAM is the easiest (and most benefit-per-dollar) upgrade, though it’s a good idea to keep an eye on which RAM slot numbers on the motherboard you’re using, since it may require a specific order for using RAM slots.

The only thing to watch out for is that you need the same exact speeds and latency timing of RAM together. The RAM will often work with two different capacities, but not different speeds.

Hard Drive

The hard drive contains the memory that stays around when you turn the computer off. It usually sits plugged into the motherboard on the side, but off in a separate part of the case.

It’s worth noting that the hard drive doesn’t necessarily need to fit in a “slot” in the case, especially with the newer “solid state drives”. It just needs to be somewhere where air can easily flow by.

External Media/Networking

You can often use a variety of external media to store and migrate information, such as CDs, tape drives, and USB/flash drives.

But, the more popular way to send information around is via the internet, which rquires a network card. With the internet you can send and receive information, including updates. This makes the computer less of an information box and more of a connection device.


The modular design of general-purpose computers lets you build computers from all the parts or upgrade parts as you need. This is crucial since new computer parts blow the old ones out of the water. In 5 years, a fancy $4,000 computer will be worth $1,000 before anything breaks on it that costs more than $500 to fix.

Things that fit in the expansion slots are never necessary to run a computer. These days, a computer has everything it needs to run everyday sound and video. However, upgrading the graphics card and sound card can dramatically improve your gaming/productivity.

When buying anything, the computers run benchmarks that measure how well something performs. Keep in mind that benchmarks for any computer part only apply for when the computer operates in a vacuum on the dark side of the moon. Real-life performance is often a fraction of the advertised number.

When you’re working on a computer, make sure you have anti-static mats and an anti-static wrist strap. Electrostatic discharge can often damage circuitry or wipe drive space. Also, never drop the parts. The parts are so delicate that even bumping some of them at the wrong angle can permanently break them.

Bear in mind that upgrading anything often requires more electrical power and more space. Noting your power draw/capacity and physical space dimensions can really save on Next-Day Delivery shipping charges.

Of course, stuff does break, and a modular design means you can replace the part instead of the whole thing. Generally, once you’re all setup, hard drives are often the most unreliable part in a computer.

Laptops and mobile devices are a weird exception to modularity. Their form factor is so small that most of their parts don’t have standard sizes. To use something “standard”, you’re forced into a few possible options:

  1. Get creative to hook it up, while will likely remove the “mobile” part of the mobile device.
  2. Order another part of that exact type from the manufacturer (which is rare because they want you to buy a new device every 2-5 years)
  3. If it was popular enough, buy an aftermarket part from a part vendor that specializes in it.
  4. Pull apart a similar or exact model elsewhere to replace that component.
  5. Accept defeat and throw it away.

I/O Chains

Computers frequently output things without considering where that input will go. Then, someone can connect that “output” anywhere:

  • A peripheral, such as the speakers or screen
  • Long-term storage
  • Another computer
  • Absolutely nowhere

Physically, components typically connect with “ports” to other peripherals (e.g., mouse, keyboard, extra storage). They’re highly modular, and abide by standardized “form factors”:

  • The design communicates what’s supposed to connect and where, as well as any constraints on what can’t or shouldn’t connect.
  • However, bad design can sometimes mean multiple types of devices can create conflicts by using the exact same form factor (e.g., most USB cables). This defies the IETF principle to be precise in what you send, and flexible in what you receive.
  • Further, it doesn’t mean it’s not possible to use alternative cabling. By separating out the wires and carefully attaching them to a different form factor plug, anything can technically plug into anything else, and may only need some programming (or software) to fix any information transfer problems.

This limited information transfer is a huge reason why operating systems are elaborate while also allowing many unrelated people to work with them at once. Each person can focus exclusively on what they understand.

We can also create some fascinating combinations by stapling the input and output together onto the same object.

  • The touch screen, for example, is both an input (digitizer) and an output (LCD screen).
  • A network router takes raw information, then routes the information somewhere. It’s the basis of most networks, but focuses on finding where something is supposed to go, then sending it there.

By sending the output to another computer, we can string them together:

  • If they have different purposes, multiple computers talking with each other creates a “network“. The largest network we have is the internet.
  • If that network is tight enough and the computers are sharing resources, they’re known as a “distributed system“.

Other Computers

There are a lot of other computers beyond your average household computer. Technically, you can do anything with a general-purpose computer, but it’s often far more inefficient than using that special-use computer.

Since computers have gotten so cheap, so easy to work with, and so powerful, most manufacturers stick them in everything that isn’t disposable or edible.

You want a remote control? Computer. Maybe you want to entertain your drool factory one-year-old? Computer. How about showing what song is playing? Computer. You want to track the time? Clearly, you’ll want a complex assembly of cogs, sprockets, and springs.

If you have a modern office workspace, you may be surprised how many separate processors are in it:

  1. The CPU on the motherboard, naturally, though it may have a split processing/graphics setup with an APU.
  2. If you have a graphics card or sound card, it has a CPU on it, though graphics cards call it a GPU.
  3. Inside the computer, it has specific-function chips called “device controllers” that regulate peripherals.
  4. If you have a printer/scanner, it typically has its own computer to regulate its jobs.
  5. Most fancier computer monitors have small processors that regulate information or tweak visual settings.
  6. Computer mice and most keyboards have small processors inside them that convert the signals into digital information before sending.
  7. Modems, which are usually built right onto the motherboard, are computers designed to send and receive information.
  8. Network routers and adapters, all along the string between computers with screens, are each their own computer. These computers are most of the internet.


By sticking multiple parts together in a big metal box, such as hard drives or CPUs, they can all work together as if they’re all one “supercomputer”. This is how big companies run their computers, since they need a lot of computing power.

The problem with big supercomputers, though, is that there are a lot more parts, so more things can break:

  • To make sure the computer is always on, they need redundant power supplies.
  • To ensure the data is safe, the computers are typically on a RAID array.
  • Several separate processors run inside the device at once, just in case one goes out.


On the extremely cheap end, some general-purpose computers like the Raspberry Pi and Arduino are designed as “single board computers”. They’re meant to be so affordable that you can give it to children (~$40).

In fact, if you’re tenacious enough, you can even build a computer in your own garage.

Also, like stated above, many computers can be specific-purpose. NAS (network attached storage), for example, is just a normal computer that does nothing but send information back-and-forth to other computers on a network. You can even make one with a Raspberry Pi!

Computers are so fast now that these computers can pretty much do anything the average person could imagine using it for, so there are a ton of projects available for them. These can range from automatic pet feeders to home security systems to running websites!


Computers require consistent maintenance. This pretty much guarantees job security for everyone working with computers, even with automation.


The biggest enemy of computers is dust, for several reasons:

  1. Dust creates a blanket-like effect on the parts, which keeps heat in. Left unchecked, it can melt the computer, especially the CPU.
  2. Static electricity builds up in dust, which can short-circuit the parts. Since most things in a computer are electrical, this is bad.

Thus, computers need regular dusting. Once a month is ideal for home users, but don’t let it go for more than 6-12 months, especially if you’re in a dusty area.

When dusting, an air compressor works much better than canned air, since canned air has chemicals that can damage the circuitry.


Computers always need updating. It can be the instructions for input/output devices (“drivers”), instructions for how the system that keeps it operating (“operating system“) or specific instructions for software.

There’s often downtime during updates, which isn’t that big a deal if you’re browsing the web, but becomes an issue if you run a business.

Also, updates might not work right. Always, always, always play it safe:

  1. Keep a copy of everything (“backup”) that you can restore everything from.
  2. Avoid running more stuff than you have to while it’s updating, especially if it’s a BIOS update. Call of Duty can wait.
  3. Only run the updates when you don’t plan to use your computer much. Overnight on weekends is usually best because you get a few days if you really screwed up.

The reason it becomes an issue is because sometimes two “processes” on the operating system can use the exact same “file” and screw up what they’re both doing with it.


Of course, sometimes that computer needs new parts. Other times, it’s been long enough that it’s a cheaper and more economical solution to simply buy a whole new computer.

Typically, upgrading computers requires understanding the specific marketing of each OEM (“original equipment manufacturer”). Each of them has their own compatibility requirements, and often make agreements with other OEMs about how to interconnect their parts.

This interconnected nature means that if a part is old enough, other OEMs might stop designing their hardware to connect to it.


Not all computers are the same. Consumer-grade technology is relatively flimsy, with the “planned obsolescence” of at least one major component in it set for about a month after the extended warranty expires, presuming home use (i.e., no more than 2-6 hours of moderate use a day). Commercial-grade technology, on the other hand, is designed to last for about the same amount of time, but under heavy use 24 hours a day. This is a portion of why enterprise-grade computers are far more expensive.

The actual quality of any given computer is a competing, advanced calculation of economics and technological development. When something has become a hot trend, manufacturers will spend more effort and money on developing that component, up until people stop caring and the trend dies. As boring technology starts becoming a commodity instead of a novelty, a Big Tech company will pick up most operations in that space (and not innovate much more), all the way until something better comes along.

To that end, buying the best computer is nearly impossible because you’re paying for an experimental technology (which may break on you) or a boring technology (which won’t be as fast).


Computers will always be a big career-maker as long as the following chain of events happens:

  1. Computers get used for more and more things.
  2. Computers need increasingly more complicated instructions to deal with the many varieties of things they can get used for.
  3. Complicated instructions create horrifically complex situations.
  4. More elaborate systems mean more links in the chain that can break.
  5. If anything breaks, it’s so complicated that replacing it isn’t very easy.
  6. Only people who are patient enough to slog through encyclopedias of documentation can work with complicated computer issues.

In other words, if you’re patient with computers, you have job security.

Further Reading

Effects of Grill Patterns on Fan Performance/Noise