Misconceptions about Modern PC Hardware

In this article, we’re going to take a look back at some of the most common concepts that people get wrong when talking about PC hardware. In each of them we are going to reveal the common fallacy and why it is wrong. That said, let’s go.

You can compare two CPUs by their cores and speed

If you’ve been in technology for a while, you may have heard someone make the following comparison at some point:

• CPU “A” has 4 cores and runs at 4 GHz, while CPU “B” has 6 cores and runs at 3 GHz. Since 4 × 4 = 16 is less than 6 × 3 = 18, CPU “B “is better than” A “.

This is one of the capital sins of computer hardware and not only of PC, but of any device. There are so many variations and parameters that it is impossible to compare two processors in this way.

Taken on its own and all other things being equal, a six-core processor will be more powerful than the same four-core design. Similarly, a 4 GHz processor will be faster than a 3 GHz processor. However, if we add to these claims the complexity of modern processors such as lithography, threads, cache memory or instruction sets , to this day they have already lost consciousness.

Also, keep in mind that there are workloads that benefit more from a higher operating frequency than from a higher number of cores and vice versa, so of course when it is said that one CPU “is better” than another, there are To answer “in what”, since depending on what the perception is used for it will change.

Speed is the most important performance indicator

Based on the first misconception, it’s important to understand that a processor’s clock speed isn’t everything. Two CPUs in the same price category running at the same frequency can have a big difference in performance.

It is true that operating speed has a direct impact on performance, but once you reach a certain point you have to take into account all the other parameters that play their own role in the same concept (“performance”), such as memory cache or architecture.

Imagine for example an AMD FX-8350 that works at 4 GHz. If we compare it with a Ryzen 7 3700X that works at 3.6 GHz as a base, we have that in reality the Ryzen is much more modern and much more powerful than the FX, much more old.

Processor is everything in PC hardware

This is something that used to be an absolute truth many years ago, but is becoming less true every day. We tend to bundle a lot of functionality under the word “CPU” or “processor” but the reality is just one part of a much larger ecosystem. The current trend, known as heterogeneous computing, involves combining many elements of computing on a single chip.

Generally speaking, the processor has a huge influence on the performance but it is not everything, since the performance also depends on the RAM, storage and graphics card when we talk about a desktop or laptop PC. If we refer to a SoC of a smartphone, for example, things do change a lot because everything is included in the same chip.

Lithography is useful for comparing two chips

Recently, there have been many rumors about Intel‘s delay in rolling out its new manufacturing node. When a chipmaker such as Intel or AMD designs a product, it is manufactured using a specific technological process known as lithography, and the most common metric to measure this is the size of the small transistors that the processors have inside.

This measurement is made in nanometers, and the most common sizes will sound like 32 nm, 14 nm, 10 nm, 7 nm, etc. It would make sense that we could place two 7nm transistors in the space that one 14nm would occupy, but the reality is quite different. There is a lot of internal overhead and interconnections, so the number of transistors and therefore the processing power does not exactly match the size of the lithography used to make them.

In addition to this, as we have already spoken before, it must be borne in mind that there is no standard system to measure this. All large companies used to measure in the same way but now each one does it in their own way, and for this reason it is not uncommon to see analysts comparing, for example, Intel’s 10nm with AMD’s 7nm.

The number of cores in a GPU to determine its performance

When comparing CPUs and GPUs, the biggest difference is the number of cores that one and the other have. CPUs have few cores but very powerful and above all, very versatile, while GPUs have very many cores (on the order of thousands) much less powerful, but dedicated only for one purpose, so they are designed to work in parallel.

In the same way that a 4-core processor from Intel can have a very different performance than a 4-core processor from AMD, the same thing happens with GPUs. There is no one correct way to compare the core counts of a GPU between different architectures or manufacturers, as each uses its own architecture and this makes measuring performance by the number of cores meaningless.

For example, one manufacturer may choose to add fewer cores but with higher functionality, while another may prefer more cores, each with reduced functionality. In short, the number of cores in a GPU cannot be used to determine its performance.

FLOPs in PC hardware

Today we use FLOPs (or more specifically TFLOPs) to measure the graphics performance of a GPU, although CPUs are also capable of performing these operations. When a new high-performance chip is released, one of the first things they advertise is the FLOPs they can generate, and this is nothing more than the number of floating point operations per second that it is capable of performing.

This seems straightforward enough, but of course vendors can play with the numbers to make their product look better than it actually is. For example, calculating 1.0 + 1.0 is much easier for a processor than calculating 1234.5678 + 8765.4321. Companies can mess with the type of calculations and their associated precision to inflate these numbers.

Also, looking at FLOPs only measures raw CPU / GPU computing performance and does not take into account other very important factors when calculating overall performance, such as memory bandwidth. Companies also optimize the benchmarks they run to unfairly favor their products, so ultimately, it is not a value that we can trust for sure.

ARM as a PC hardware and chip maker

Almost all embedded and low-power systems run on some type of ARM processor. However, it should be noted that ARM is not a chip maker as such and, in fact, does not make physical chips. Instead, what ARM does is just design, and then it sells the license for other companies to make them.

For example, the A13 SoC in iPhones uses ARM architecture but is manufactured by Apple. By licensing their intellectual property, ARM allows Apple, Qualcomm, Samsung and many others to make their own chips using ARM designs but tailoring them to their needs, of course by charging them for it.

CPUs will keep increasing their speed

One of the most famous representations in the industry is Moore’s Law. It’s an observation that the number of transistors on a chip doubles roughly every two years, and while it has been fairly accurate for the past 40 years, manufacturers have been in serious trouble for quite some time now, and in fact It seems that many have thrown in the towel because they are not able to keep reducing manufacturing processes.

If we can’t add more transistors to the chips, one solution would be to make them bigger. The limitation here is getting enough power for the processor and then being able to dissipate the heat it generates. Modern chips draw hundreds of amps of current and generate hundreds of watts of heat, so having this on a larger scale would actually be a pretty big problem in PC hardware.

The power supply and cooling systems we have today are already struggling to keep modern processors running, and many manufacturers say they are near the limit of what can be achieved with the technology they have. If the chips were larger, a greater number of transistors would be integrated, but the power and dissipation requirements would be unaffordable.

Therefore, and although little by little processors are delivering more and more performance, manufacturers are opting for a more efficient approach (higher performance per watt), and although the speed of the processors was increasing year after year for a long time. Time, it seems that we are at a tipping point where 5 GHz has established itself as the barrier at which they have stagnated, and it does not seem that we are going to go much higher from there.