What Makes A Computer Fast? [Performance Defined]

What makes a Computer Fast?

Everyone likes to talk about fast PCs, but how are we actually defining “fast” and, more importantly, how should you be prioritizing components inside your PC to maximize that speed?

Do other factors besides your basic components impact your PC’s speed?

I’ll be tackling all these questions and a few more in the article below.

What Makes a Computer Fast? Understanding PC Performance

First, let’s just define “fast” as “responsive” in a PC context.

Older generations of PCs with much slower hardware still managed to feel “fast” when running operating systems and software built around those limitations, but not quite to the extent of systems today.

If I had to choose a rough order of components by how much they impact your PC’s perceived speed, I would choose the following:

1. Storage
2. CPU
3. RAM
4. GPU (and Display)
5. Networking Hardware

If you aren’t sure how each of these components earned its place in that hierarchy, keep reading!

I’ll dive into my rationale for each, and more importantly, give you actionable advice for building a faster PC or upgrading your existing PC.

How Storage Makes a Computer Fast

The biggest thing that will make a computer feel “fast” with modern hardware is your storage. Specifically, the kind of storage you have (not the amount).

What you may not know is that a common bottleneck on modern machines has actually been a staple of PCs for decades: mechanical hard disk drives, also called HDDs.

So, how do HDDs bottleneck a PC’s performance?

Essentially, the raw throughput of an HDD tops out at around ~150 MB/s in sequential transfer speeds. That isn’t bad by any means- it’s actually still more than enough for playing back modern 4K HDR video files, provided the rest of your PC can keep up.

When it comes to actually booting up your PC and loading programs (which means randomly reading and writing many little files), though, it is a lot slower.

Now, taking a glance at the chart above, it’s important to make a quick clarification that sequential read speeds are very specific (loading/compressing large files), and that the metric more likely to impact your OS and application loading times will be your random read speeds.

Even so, random read speeds are exponentially higher on even a basic SATA SSD when compared to a mechanical HDD, and while they do see a further increase on NVMe, it isn’t nearly as huge as the sequential read speed boost is.

Think about it: the slowest part of using your PC, historically, has always been the boot-up process. If you don’t have an SSD, it still is.

Compared to even the best HDDs, SSDs drastically cut down on the time it takes for you to boot into your operating system and launch programs.

Even on really weak and “slow” PCs, a SATA SSD upgrade can provide a shocking boost in speed and responsiveness, though it won’t magically improve CPU/GPU performance.

Even when it comes to HDDs, there are meaningful differences in performance depending on the supported RPM (Rotations Per Minute) of the drive-in question.

Since HDDs are basically metal boxes with spinning disks inside, RPM serves as a pretty direct measure of speed. The “HDD” we’re referring to in the chart above would be a market-standard 7200 RPM HDD.

The problem is, especially if you’ve used an older laptop, not all HDDs run at 7200 RPM.

Some are instead limited to 5400 (or lower) RPM, and this is usually a result of making an HDD smaller and drawing less power to suit a 2.5-inch SATA form factor rather than the 3.5-inch SATA form factor used by desktop hard drives.

Unfortunately, this makes an already-kinda-slow storage technology even slower, and if you’re trying to have fast boot times or have a generally-responsive machine, a 5400 RPM HDD will definitely be a problem.

Past standard 7200 RPM HDDs, higher-RPM HDDs, and SSHDs (Solid State Hard Drives) exist as well.

While neither of these options really compete with SSDs in terms of raw throughput, especially not NVMe SSDs, they can be a nice compromise between speed and price-per-gigabyte.

The big catch of SSDs compared to HDDs is that all that extra speed comes at a huge cost-per-gigabyte, which makes having a primary SSD (OS/apps) and secondary HDD (media storage) a very popular choice.

To conclude this section: Your Storage can bottleneck your entire PC even if the rest of your components are blazingly fast. An OS and Applications have to be loaded from storage, and only a fast (SATA) SSD or NVMe SSD will truly make your PC feel fast and not bottleneck any of the other components.

My recommendation for a fast PC: Get an NVMe SSD.

How a CPU Makes a Computer Fast

Many people probably expected me to put CPU (Processor) first in this hierarchy, and they aren’t wrong, strictly speaking.

While storage will have the biggest impact on your loading times and a general feeling of OS responsiveness, fast storage doesn’t matter if your CPU isn’t fast enough to do what you need it to.

CPU performance is directly tied to every workload that you’re going to be doing on your PC, and while modern CPUs outstripped the requirements for basic use a long time ago, heavy-duty workloads are still very CPU-dependent.

Things like video editing, file compression, gaming, and professional rendering tasks are all heavily reliant on raw CPU performance.

Even in the case of something like gaming, which is commonly understood to rely more on GPU than CPU power, CPU power will still determine the maximum possible in-game framerate, regardless of GPU or graphics settings changes.

With that general point aside, how can one actually expect the technical specifications of a CPU to reflect its real-world performance?

There is some nuance to this, as differing CPU architectures in the same or following generations can have the same basic specs (4 cores, 3 GHz, for example) but perform massively differently.

However, these specs are still worth looking at, especially for comparing performance within the same CPU architecture (ie 10th Gen Intel Core i3 vs 10th Gen Intel Core i5, where those core/GHz distinctions are working off the same baseline architecture).

To paint a rough picture of how CPU performance works, I need to start by painting a rough picture of the CPU itself.

A CPU is composed of one or more processing “cores”– just one of these is what we used to understand as a CPU, but as we managed to slap more of them onto the same chip, we took to calling them “cores” instead.

There’s other stuff in there too, like memory controllers and cache, but we’re just gonna focus on cores for now.

A CPU core is a processing unit that communicates with your operating system and the rest of your PC.

A CPU core’s speed can be increased by increasing its “clock speed”, measured in Megahertz or Gigahertz. This makes CPUs with higher clock speeds favored when compared to CPUs of the same architecture, and even results in the practice of “overclocking”, where the user increases that speed themselves.

Past individual CPU cores and their clock speeds, you have the number of cores themselves, and their corresponding “threads”.

Think of a “thread” as a virtual representation of your core seen by your operating system.

For CPUs without SMT support, 1 thread = 1 core…but when you introduce SMT into the mix, that rule changes.

SMT, or Simultaneous Multi-Threading, enables a CPU core to be read as two rather than one thread by the operating system.

This doesn’t necessarily double CPU power, especially for real-time workloads like gaming, but it does make a CPU much better at multitasking and non-real-time workloads (like video rendering).

Multi-core CPUs in general already benefit from multitasking as you increase core count, but SMT support on top of that can provide a further boost in productivity and multitasking.

Source: Intel

A notable exception to the 1 core = 2 threads rule can be found in newer Intel CPUs, which are divided into P-Cores and E-Cores.

P-Cores still follow the SMT rule, but E-Cores do not support SMT, and so are still read as 1 core = 1 thread by the OS.

This has paid off pretty well for Intel’s 12th Gen CPUs, but AMD may not end up adopting this architecture. Being pretty good at multi-core CPU architecture in general, AMD may not ever feel the need to.

To conclude this section: The CPU (Processor) is a crucial performance factor in almost any task that you execute on your PC. The More Cores and Threads your CPU has, the more tasks it can run in parallel. The more modern its architecture, the larger its cache, the higher its IPC (instructions per cycle), and the higher its clock speed, the faster those tasks will be executed.

My recommendation for a fast PC: Get a high-clocking, modern CPU, with at least 6 Cores.

How RAM Makes a Computer Fast

So, how does RAM (= Your PC’s main Memory) make a computer fast? Of the components on your PC, your RAM is most directly tied to your CPU.

Like your CPU, it’s responsible for a lot of the general heavy-lifting that goes on in your system, and here’s why:

RAM is needed to hold all of the tasks being actively managed by your CPU. It’s the working memory of a PC.

In general, though, RAM capacity matters more than RAM speed for getting the desired results.

The reason why RAM capacity matters more than RAM speed is that your PC won’t just die if it runs out of available memory. Instead, it’ll fall back on a paging file on your storage drive.

Unfortunately, even a fast SSD will be a lot slower than your actual RAM, so running out of RAM and being forced to use a page file results in a significant performance loss.

This is even worse on an HDD, as outlined in the Storage section earlier in the article.

That being said, you generally don’t need that much RAM until you start opting into higher-end workloads.

With heavy-duty workloads like professional editing, professional rendering, or gaming, 16 GB is more of the baseline where you want to start, with high-end needs breaching 32 GB.

High RAM capacity is integral to professional workloads, where you want as much of your project files as possible present in your memory, not falling back onto your hard drive.

Past RAM capacity, there is also RAM speed and RAM latency. The performance these specifications have on your PC performance is much harder to quantify since the impact is usually small.

I’ve written a more in-depth guide on how RAM speed impacts different workloads for those interested, but the general takeaway is that most performance gains are around 5% or under, if present.

For gaming, at least, RAM speed does matter a bit more. Gamers may not notice much of an increase in average FPS with faster RAM, particularly with older games not suited to utilize it, but fast RAM is great for improving framerate consistency.

That is, even if it doesn’t increase your average FPS as it may in modern games, it will still increase your minimum FPS, resulting in a less perceived loss in fluidity whenever performance dips in intensive scenes.

To conclude this section: More RAM will make your PC faster up to a certain point. If you already have enough, adding more will do nothing towards increasing performance. Higher DDR Generation RAM, higher clocking, higher channel, and lower latency will increase your RAM performance.

My recommendation: Get at least DDR4 or DDR5 RAM with high clock speeds and low latencies that run at dual channel (with two+ modules).

How a GPU Makes a Computer Fast

While the GPU (short for Graphics Card) isn’t particularly taxed by most things you’ll be doing with your PC (even GPU-accelerated media consumption isn’t very demanding on modern cards and iGPUs), the tasks that are tied to GPU performance are extremely tied to it.

The main tasks that are tied to graphics performance are gaming and professional rendering (3D or video).

While your CPU will still serve as an overall limiter, your GPU still matters a great deal in these scenarios, especially if you’re trying to push higher resolutions, texture detail, and graphical effects.

The better your GPU, the more you’ll be able to reduce render times and latency, increase framerates, and improve visual quality.

Source: NVIDIA

Besides raw graphical power, GPUs can also be used to accelerate non-graphics workloads.

Rather infamously, cryptocurrency mining is one of those non-graphics workloads, and the combination of a crypto boom and chip shortage in the past few years made GPUs extremely expensive, like 2-4x MSRP expensive.

Fortunately at the time of writing this trend has started to fall with a dip in crypto and improved supply, but it can still be an issue sometimes.

My recommendation: Check our GPU Lists that can be ordered by performance to find one that best fits your needs and budget. AMD List here, Nvidia List here.

How Networking Makes a Computer Fast

Last but not least, let’s talk about how networking makes your computer fast.

Obviously, the plan and Internet speeds you pay for from your Internet Service Provider (ISP) will have a big impact on how you use your PC, especially when web browsing or downloading/uploading/streaming pretty much anything. But that isn’t actually what I mean.

In this case, I mean your onboard networking hardware, and whether you’re using Wi-Fi or Ethernet.

Most onboard networking hardware is perfectly fine for Gigabit Ethernet cables, and if it doesn’t support a recent high-speed Wi-Fi standard, an adapter or expansion card can do the job just as well.

However, out-of-box networking solutions might not be right for everyone.

In an enterprise or business environment, for instance, getting a dedicated networking card that supports multiple Gigabits per second for a server PC or a workstation can help significantly accelerate file transfers across the same network.

A fun example can actually be seen in the LinusTechTips video embedded above- since they’re a hardware-centric YouTube channel with daily uploads across multiple channels, making use of a 100 Gigabit Network Switch actually makes a lot of sense.

Remember: there are also multiple staff members doing footage recording, editing, etc on separate PCs in their office space for their consistent large-scale video projects.

I have a more generally-applicable solution for most users, though. Generally speaking, you’ll want to use an Ethernet cable for your networking whenever possible, especially if you’re doing any kind of latency-sensitive work or gaming.

While modern Wi-Fi standards continue to improve their speeds, Ethernet doesn’t suffer from the inconsistency and interference that comes as an inherent part of a Wi-Fi connection, and thus is ideal for stabilizing network performance and latency.

FAQ

Before we wrap things up completely, there are a few things worth talking about that don’t necessarily make a computer fast but still have a pretty big impact on performance.

Let’s talk about your motherboard and your cooling real quick.

Does a Motherboard affect performance?

Yes and no.

By itself, the motherboard doesn’t really have any processing power to speak of. So you’d think it doesn’t affect performance, but you’d actually be wrong.

Think of your motherboard as the spinal cord of your PC. It’s integral to your PC’s functioning since everything has to connect to it and communicate through it.

More importantly, your motherboard will also determine a lot of other factors, including what hardware is actually compatible with your PC and how fast that hardware can run within its constraints.

A weak but functional motherboard won’t be a major bottleneck, but it will prevent you from doing things like overclocking or multi-GPU or multi-NVMe SSD setups. If you want high-end features like these, you’ll want a high-end board to match.

To learn more about how a motherboard affects your performance, consider Alex’s article on the topic.

Does PC cooling affect performance?

Oh, absolutely.

See, as it turns out, pretty much anything powered by electricity is going to generate and exhaust heat. That includes every component in your PC, but especially your CPU and your GPU.

In fact, these components can very easily reach and exceed 95 degrees Celsius when under heavy load.

This by itself wouldn’t be a big deal except for the fact that if your temperatures are too high, your hardware will begin to degrade in functionality- and can even be damaged in extreme scenarios.

Fortunately, this thing called thermal throttling exists in basically all modern hardware to prevent permanent damage from an overheating system.

Unfortunately, thermal throttling only works by reducing power and clock speeds to your CPU or GPU, which results in minor-to-major performance drops whenever it kicks in.

Thermal throttling is especially problematic for hardcore gamers and professionals who are regularly pushing these components to their limits.

For an optimal positive pressure airflow configuration, always make sure you have at least one more intake fan than an exhaust fan.

Also, fortunately, most cases of thermal throttling can be fixed by simply making the correct adjustments to your cooling setup, or making some software adjustments if that isn’t an option for whatever reason.

I’ve covered various ways to alleviate thermal throttling in extensive detail in my Thermal Throttling Guide, so head there if you want a detailed rundown on this.

Otherwise, my basic advice of keeping your PC dust-free and having enough fans for a good positive pressure airflow configuration should be a good starter.

Over to You

And that’s it, for now! I hope this article helped provide a clearer picture of PC performance and how it actually works across your various components.

If you have any lingering questions not answered by this article, feel free to ask them in the comments section below or our forums! Both are monitored by myself and other hardware experts on the CGDirector team, and we’ll be happy to help where we can.

Until then or until next time, have a good one! And remember to at least get a SATA SSD for your operating system. Even if everything else is on a slower HDD, that step alone will do wonders for the perceived speed and fluidity of your PC usage experience.