Will your next Apple MacBook be powered by Intel or ARM? The ultimate showdown has already begun.
Forget Mac vs. PC, or Android versus iPhone. The real showdown has already started, between Intel Core CPUs, and upcoming ARM-based Cortex A-series processors, for supremacy of the laptop market.
ARM chief architect Mike Filippo isn’t pulling any punches, as he stated that, not only, ARM Cortex A76 will be able to go toe-to-toe with the Intel Core i5 on its best day: he’s also confident that ARM architecture will be able to compete with Core i7 CPUs, with a 2019 target release.
The reason that makes this information newsworthy lies in the widely accepted notion that silicon has reached its peak when it comes to speed, at roughly 4.5GHz, and all that has remained in terms of improvements is stability, energy efficiency, and number of cores, the latter of which is the only factor that can barely provide any illusion of an increase in execution speed, not because the processor would work faster, but because more cores make it for a higher number of parallel operations.
By this token, there is a finite number of operations that can be performed, which does not guarantee any increase in performance, at least in some cases, simply because most software, as complex as it may be, only requires a limited number of tasks to be processed at one time.
For instance, when transferring files via FTP, over a network, there is only one process that takes care of transferring files, which gets repeated times the number of concurrent connections. No matter how many cores your processor has, each operation only uses the resources it needs to complete the task, no matter the number of cores available. This will not speed up how fast files are transferred, nor how fast a connection is made.
Much like when transferring files, processing video in software like Adobe Premiere, or After Effects, requires a finite number of CPU requests to be processed, over a lengthy period of time. This means that even with 20+ cores available, processing footage will not occur any faster than the top speed of the processor itself, which can be, at best, 4.5GHz to 5.0Ghz.
By claiming that ARM can beat Intel at powering MacBooks and Windows 10 laptops, Filippo is challenging a concept that CPU manufacturers have long trusted, according to Gordon Moore’s assertion that processor capabilities and miniaturization will hit a hard limit by 2025.
There is only one problem with this proposition, and that is that Moore’s law is far from being an established law of physics, and more of a product of predictive research, which was last updated in 2003, with Moore’s “second law”, pushing its original claims from a mere decade, with a hard limit up to 1975, which clearly did not happen, to 2025, in 2003, which still did not consider the true potential of researching new materials and molecular properties, down to atomic levels of manipulation and storage of electronic data.
By this token, one could theorize, or at least consider the possibility that the discovery of new materials and manufacturing processes, could potentially unleash a new category of processors that do not have to obey Moore’s law. By that assumption, ARM CPUs could have a shot at challenging the traditional mold of X86 architecture, and reach more powerful devices, including high-end laptops and workstations.
Will ARM manage to accomplish this extraordinary feat by 2019? The deadline is short, but the boldness of Filippo’s claims seems to be a hint that ARM may already have reached important breakthroughs in that direction.
The biggest question, however, is whether these new ARM CPUs are anything like the ARM processors we already are familiar with, as they power 50% of worldwide’s mobile devices like smartphones, tablets, streaming devices, like Apple TV, and more.
Chances are that ARM processors designed for future laptops like the MacBook, will have little in common with their current architecture, as their design and technology will most likely scale across selected ranges of devices, and split in classes, similarly to how X86 CPUs are already organized.
Apple’s current adoption of ARM processors is the best example. T-class processors are specifically designed to handle deep functions within Apple iMac Pro computers, while A-series ARM processors are relegated to powering iPhone and iPad devices. Finally, S-series CPUs are SoC’s designed to power much smaller devices, like the Apple Watch.
To understand the real potential of ARM, means to take a hard look at ARM architecture, as something that, unlike what some hope and wish, will not be a “one-size-fits-all” silver bullet for the tech industry, especially for manufacturers.
ARM’s best characteristics are in regard to power management, and efficiency in execution, rather than speed, but the future of computing relies on the optimization of code to be leaner and optimized for speedier processing on slower processors, to lower the cost of devices, and create easier entry points into the technology, for more consumers. The need for speed relies now on a niche market that includes creative professionals, and power-users requiring hardware that satisfies very specific needs, like industrial design, high-end video processing, or crypto-mining, which is something that most consumers aren’t expected to develop enough of an interest in, to invest their lives savings.
Of course there will always be a market for high-end gamers, and that represents much more than a niche category, but the fact is that GPUs have already taken over much of the responsibility to make high-end games perform well, and render massively detailed virtual environments, for 2D screens, and VR. ARM CPUs may not yet handle the intense workload from a session of Fortnite at full specs in desktop mode, but thankfully, they aren't likely to have to be responsible for that particular job, at least not entirely, as far as anything that can, and should use graphical acceleration.
In conclusion, ARM architecture has more than a few reasons to flourish and become worthy competition of both Intel and AMD desktop CPUs. As far as whether ARM will, that’s something we’ll have to wait next year to find out.