CPU and GPU might be familiar terms for even the most novice of computer users, but the APU might be a relatively novel concept.
Truthfully, that might’ve been the case by design, but APUs have certainly come a long way from their modest beginnings.
Before we dive into what an APU actually is, let’s first make sure we have clear definitions of both CPU and GPU.
Table of ContentsShow
Central Processing Unit – CPU
Also simply referred to as the processor, the CPU is the main controlling chip in what we consider a PC. At its core, it executes basic arithmetic, logic, controlling, and I/O (input/output) operations given to it by the program. On a granular level, it’s a very simple piece of electronic circuitry which consists of the arithmetic logic unit (ALU) and control unit (CU).
The ALU is in charge of executing arithmetic and logic operations, or simply Boolean algebra, while the CU fetches data from memory and basically gives the instructions to the ALU. Below is a simple example of how an OR gate functions. When people say that computers are ones and zeroes, this is what they mean by that.
The CU gets data that software is in charge of sending and it determines which operations the ALU needs to do in order to return the desired result. The ALU then uses data stored in registers and then compares them and produces an output which the CU then sends through to the appropriate location. Of course, this is a simplified representation, as fully explaining what these things do would be steering too far away from our topic.
Modern CPUs have multiple cores and multiple threads and this allows them to simultaneously perform tasks that older, single-core CPUs could not. These modern CPUs often have four or more cores (with extreme cases with 16 cores) which can be divided into virtual or logical cores with hyper-threading or multi-threading.
The processor’s clock speed is generally accepted as the metric of the speed of operations executed which pretty much is true, but let’s give an example. The unit of frequency is hertz (Hz), which indicates a number of cycles executed per second. Modern computing has gotten so impressive that CPU speed is measured in gigahertz (GHz) which means that a given CPU can execute utterly mind-boggling several billions of operations in a single second.
And even if say 3 GHz is not enough for you, most modern CPUs can be overclocked to get even more computing power. Although, it’s crucial to note that even the most powerful CPUs out there cannot go past a certain point. The current Guinness World Record is at 8.79433 GHz and that was set all the way back in 2012, so it’s safe to assume we probably won’t be going too far from there until we get to quantum computing.
Gamers shouldn’t be too concerned about getting those super high clock rates, as even mid-range CPUs will be able to avoid bottlenecking when paired with a high-end or even enthusiast-class GPU.
The vast majority of mainstream CPUs are produced by AMD and Intel, where the latter was considered king for a long time until the former introduced the world to its Ryzen CPUs and fundamentally altered the world of technology. Although these two are best known, and generally accepted as the best CPU manufacturers out there, it’s worth mentioning that other companies such as IBM and Apple are also producing processors.
Graphics Processing Unit – GPU
Fun fact: the term GPU was coined by Sony when they were launching their PlayStation console back in 1994.
Another fun little fact about the GPU is that due to its parallel design structure, it’s actually more efficient than the CPU when it comes to algorithms that process large blocks of data in parallel. Of course, CPUs are more general-purpose, while GPUs are optimized to work with specific data.
There are two main forms of GPUs and they are called dedicated and integrated GPUs.
With a dedicated graphics card, the chip comes on a separate video card that’s removable and upgradeable. It also comes with its own dedicated RAM which is customized to work better with graphics-related processes. Dedicated GPUs interface with the motherboard via the PCIe slot.
Integrated GPUs are what the name suggests – a part of the motherboard. It uses system RAM for its operation and as such is often a lot slower than a dedicated GPU.
There are also external GPUs which are what they sound like- a GPU that’s outside the case. They connect to the PC or laptop via a mini PCIe port, ExpressCard, or a Thunderbolt port, and they’re certainly worth mentioning as they’ve begun gaining traction recently. Most often they come with their own PSU, as most laptop PSUs don’t have enough to support an external GPU, which in itself is a huge energy consumer.
Much like the CPU market, there are two competing GPU manufacturers and while AMD is also competitive here, they’re largely outshined by their rivals Nvidia. And just like CPUs, GPUs can get overclocked if their base clock is too slow for you.
Accelerated Processing Unit – APU
So, what is an APU then?
Actually known as AMD APU, it’s simply a marketing term for microprocessors that have both the CPU and the GPU on the same die. This might sound a little bit too similar to an integrated GPU, but there are distinctions between the two terms.
The original idea for an APU started all the way back in 2006 when AMD ran a project named Fusion, but the first APU wasn’t released until 2011.
The interesting thing about the name APU is that it’s an HSA-compliant which is the reason why Intel can’t call their integrated graphics chips that. HSA stands for Heterogeneous System Architecture which is a set of cross-vendor sets of specifications that allow for the integration of CPUs and GPUs on the same bus with shared memory and tasks. HSA is defined by the HSA Foundation which was founded by, you guessed it, AMD and several other vendors, Intel not included, evidently.
Intel does have an integrated GPU line, simply known as Intel HD Graphics, but as previously mentioned and largely due to its rivalry with AMD, it can’t be classified as an APU as it does not offer HSA features. Because HD Graphics technically is a combination of CPU and GPU on the same die, they deserve to be mentioned here, but due to the technical difference, they can’t and aren’t considered APUs.
What makes an APU a really interesting piece of technology is the unique way in which the CPU and the GPU are combined on a single die. As such, they are able to share the same resources (including RAM) which allows them to be more effective in the way they take advantage of those resources.
Another upside is that in the long run, there is a noticeable power efficiency increase, all stemming from that resource sharing idea. Another product of that idea is the performance speed, as well as the decreased cost of manufacturing which makes APUs have a great dollar value to its name.
Just like the CPU and the GPU, APUs are able to get overclocked. This comes in rather handy as APUs can be limited in performance out of the box, although it’s fair to note that you shouldn’t expect a stark improvement.
Unlike the CPU and GPU, APU’s market is a bit harder to define as AMD pretty much holds all the cards in regards to the APU patent. It’s interesting to note that APUs are most often compared to Intel’s integrated graphics line because of the similarity in the sense of GPU and CPU being on the same die, but APUs easily trump anything Intel has to offer, performance-wise.
Even though APUs originally hit the market back in 2011, which in technology years might’ve as well been a century ago, and they were slowly waning from a novelty combination to obscurity, AMD has brilliantly adapted their Ryzen processors and their latest release Zen 2 garnered a positive reception. AMD will look to build on that momentum with the next scheduled Zen 3 release coming in 2020.
APU Or CPU/GPU Combo?
As always, when figuring out your next tech purchases, it’s best to fairly and precisely assess your needs and figure out the intended use for the hardware.
If you’re looking for a work computer and are looking to game casually with eSports games or indie titles, then APU is a great choice. What makes APU an excellent option is its ability to be upgraded. In other words, you can get an APU and then down the line when you decide you want to play some more graphics-intensive games, you can obtain a dedicated GPU.
However, dedicated CPUs and GPUs are a better option in almost any other situation. It’s just too hard for a single-die combo to compare against the separate CPU and GPU. Simply put, APU is the jack of all trades, but master of none, while both CPU and GPU are polar opposites.