Cameron ByersIn the near future, we expect to see a potential disruption in computing due to a switch in computer architectures as traditional architecture begins to reach its upper limits, and newer alternate architectures currently used in mobile phones and low-power devices become more prominent and powerful.
Computing has been dominated by the x86-64 architecture for the better part of the last 40 years. It is used in the majority of desktops, laptops and servers and was designed to use a large number of complex instruction sets and to execute many actions in one instruction. This is referred to as complex instruction set computing (CISC), great for handling complex workloads and providing lots of raw computing power, but this comes at a cost in the form of high energy usage and heat. Thus, efficiency is secondary to compute performance and may result in limitations of particular spaces where energy efficiency is a necessity, such as mobile devices.
In contrast, the ARM architecture has seen its usage increase over the last decade as it has been heavily leveraged in the mobile space in cell phones, tablets and hobby computers such as the RaspberryPi. It was designed to use fewer and simpler instructions and will only perform a single action per instruction. This is referred to as reduced instruction set computing (RISC). This is great for efficiency and low power usage which is why it has been adopted in mobile devices and hobby computers.
X86-64 has not seen the rapid increases that it had seen in the past. It has plateaued, as Moore’s Law’s (the expectation that the speed and performance of computers will increase two-fold every two years while halving in cost) end is on the horizon. Intel and AMD have been pioneering innovations that have increased the power of x86-64 CPUs such as adding additional compute cores and the addition of hyperthreading. These have been used to mitigate the fact that single core performance has not been improving at the rate that was once common. There is also the physical limit at which transistors cannot shrink without additional complications such as dealing with electrons skipping transistor gates via quantum tunneling.
ARM system-on-a-chip (SOC) processors have gotten more powerful and are beginning to rival x86-64 processors in compute capabilities. It has introduced a blend in the mobile and desktop platforms. In the last decade, the computing space has changed tremendously. ARM SOCs have grown in popularity as they are highly customizable and can be modified to fit the needs of the manufacturer. Tech giants like Apple and Microsoft have either developed or are developing ARM SOCs that will fit their needs. While it is an expensive process, it does allow for powerful processors that will run cooler and use less energy than current x86-64 processors.
As generalized computing becomes more difficult (with Moore’s Law becoming a barrier) there has been a shift to optimizing workloads with specialized computing. Leveraging the customizability of the ARM architecture and its ability to address the desires of manufacturers and clients, it can be just as powerful if not more so than its x86 counterpart. Just as different machines will have different needs, different SOCs can be designed to fit the use case, making them more efficient. IoT, mobile, laptop and server chips will all be custom built to fit their particular use cases and data center servers will require multiple, high-performance cores while a laptop SOC will require a mix of high and low-performance cores with LTE built-in for connectivity. Because of the patented technology designs of ARM SOCs, general computing will begin to be less prioritized in favor of specialized computing needs. If a function requires specific cores for AI functionality, then chip manufacturers will build a SOC that will have more AI compute cores and not generalized compute cores. The introduction of ARM processors to the server environment has the potential to save corporations money on both chip costs as well as heating and cooling. This can be seen with Amazon’s Graviton2 processors which have a better price to performance and the potential to change the server computing market as ARM processors begin to gain traction and move within the marketplace.
App compatibility and cross-platform support have been the largest obstructions to additional integration of ARM in the server, laptop and desktop spaces. This is beginning to be addressed by major companies like Microsoft and Apple creating x86 emulation technologies that run on ARM-based SoCs. Apple has produced a line of laptops and desktops that are leveraging ARM-based SoCs which can run x86-64 code via an emulation layer. Microsoft is also developing an in-house SoC, designed by ARM and an SoC designed by Qualcomm in one of the machines in their surface line of computers. Microsoft has been working to port their Windows 10 operating system to work natively on ARM devices as well as leveraging emulation technology to run x86-64 applications on ARM systems.
As manufacturers build scalable and specialized systems, they will begin to move away from the common generalized computing model and create more specialized systems that fit the business need.
To learn more about our technology consulting solutions, contact us.
Warren Fish
Jon Medina
Nishi Prasad
Julia Artzi
