SoC: Quick breakdown of “System on a Chip”
What is a SoC?
To understand the concept of a SoC, think of the anatomy of a human. The body comprises of multiple organs that perform separate functions. These organs communicate with the brain through the body’s nervous system. Imagine, hypothetically, all these organs were miniaturized and fit into one single organ, in a way that optimizes communication between them making it faster and more efficient. From a biological standpoint this seems bizarre and impossible, but from a technological standpoint, something similar has been done. Let me explain.
Taking a regular computer for instance, on its PCB (printed circuit board) you would have separate components such as a CPU, GPU, RAM, and a plethora of other circuitry mounted in sockets/slots within the Motherboard. These are integrated to work seamlessly with each other using circuits called Buses. The CPU is a microprocessor that performs all the mathematical computations or the processing in a computer, the GPU does all the heavy lifting on the graphics side, the RAM is the computer’s memory and so on and so forth. That is the general gist of it.
Now, take all the required electronic circuitry of those components and combine it onto one single integrated chip or circuit (IC). Instead of a PCB where several components and chips are assembled, all these are fabricated onto one unit. That is simply an SoC or System on a chip.
It is a complete electronic substrate system with analog, digital, mixed-signal, and radio frequency functionality. Signal processing, wireless communication, artificial intelligence, and other operations can be performed depending on the type of system that has been shrunk to the size of a chip.
The greatest advancements in SoC technology occurred in the 1980s, when the personal computing revolution took off. With the debut of the AMD286ZX/LX family of SoCs by AMD in 1991, this technology flourished even further. ARM Holdings began licensing its fabless processor designs to other companies in the late 1990s.
The tight integration of components within SoCs was made possible due to the adoption of the Arm CPU architecture by manufacturers, considering that integration with x86 CPUs would have been challenging due to their higher power consumption and high TDP (Thermal design power). Because SoCs are designed to maximize power efficiency, Arm CPUs were ideal since they outperform x86 CPUs in terms of performance per watt.
Arm is based on RISC (Reduced Instruction Set Computing), whereas x86 is based on CISC (Complex Instruction Set Computing). Arms’ CPU instructions are atomic (no further simplification or division required), with a close relationship between the number of instructions and the number of micro-ops. CISC, on the other hand, has a far larger set of instructions, many of which perform numerous tasks (such as optimized math and data movement). Decoding these complex instructions consumes more power, resulting in improved speed. Even though this was the case several years ago, modern Arm CPUs have proven to be more powerful, faster while being more power efficient than x86 based CPUs. Hence adoption of Arm architecture has rapidly grown among manufacturers, leading to a higher SoC production.
Advantages of SoCs
· Power efficiency due to maximized performance per watt. This results in drastically improved battery life compared to conventional architectures.
· SoCs can have CPUs with much lower TDP without sacrificing performance. This results in better thermal management enabling SoCs to be passively cooled without having an active cooling system.
· SoCs have smaller components which means the chip itself can be the same size or even smaller than a traditional CPU, therefore they can be used in products with limited physical space. On the other hand, devices using SoCs can be shrunk to much smaller sizes.
· At the hardware and firmware levels, a SoC delivers more security.
· Because of the tight integration between the high-speed processor and memory, a SoC allows for speedier execution of tasks.
· They reduce or eliminate the need for cabling within the PCB.
· When it comes to components like RAM, factors like type and bus speed will differ from one to another, therefore for general consumers this might be a confusing process when upgrading or swapping out RAM. Should you go with SODIMM or DDR? What speed should you choose? 2400MHz? 3200MHz? Does it matter? Having an SoC eliminates these confusions among general consumers. The only concern this will raise is that they will need to purchase a product configuration that has adequate RAM since upgradability down the line is not a possibility in SoCs.
Disadvantages of SoCs
· The initial design and development costs are significantly high therefore the cost per SoC will be indirectly proportional to the number of SoCs produced. This is of course a disadvantage for manufacturers and not general consumers.
· Even a single transistor or system failure can be exceedingly costly because the entire board must be replaced. SoC maintenance and reparability is very difficult and pricey.
· The complexity of integrating all systems on a single chip is high. It is not ideal for applications that require a lot of power such as video games. This is why gaming CPUs still utilize the x86 architecture.
· No upgradability of components like RAM.
Real world use cases
In an era where PC like performance is expected on portable devices such as smartphones and tablets, it is impossible to have separate components assembled within a PCB inside devices with such limited space. This is where SoCs come into play and tackle this issue head on. An SoC is the heart of any smartphone and most tablets. If you didn’t already know, manufacturers like Qualcomm, Samsung and TSMC mass produce SoCs specifically for smartphones. Samsung Exynos, Qualcomm Snapdragon, Apple A-series chips are prime examples of mobile SoCs. Google’s upcoming (at the time of writing) Tensor is said to be a revolutionary SoC that will drastically improve AI and Machine learning tasks in their Pixel phones.
SoC technology has recently been utilized in smaller sized PCs and laptops to minimize battery consumption and increase the performance by employing a single chip to manage all of the system’s numerous elements. Apple’s in house chip, the M1 is the best example of a desktop level SoC currently on the market. It is powerful enough that they are confident with using it not only on their MacBooks but their 24” iMac Desktop computer as well.
Conclusion
Higher prototyping and architectural expenses, more complex debugging, and lower IC yields are among the challenges of a SoC. IC is expensive to produce and requires a long time to do so. However, as the technology develops and is implemented, this is likely to change. SoC technology is a new field that’s getting a lot of attention as a way to support high-speed computing. There are numerous concerns in this subject that necessitate a thorough investigation. After these challenges are resolved, SoC technology can be significantly improved for future computer systems.
