As more and more new late 2020 MacBook Airs and MacBook Pros land in users’ hands, don’t be surprised to see the pile-up of news and reports about their impressive performance. That has always been Apple’s assurance anyway for those uncertain about the switch to a new Apple Silicon. Much of the company’s messaging has revolved around the usual culprits of office work and video editing but now it is trying to appeal to a particular class of users, specifically developers need to do some AI and machine learning on their M1-powered Macs.
MacBooks, particularly MacBook Pros, have always been a favorite of some developers but they might have felt a bit held back when it came to machine learning work, particularly training AI models. The Google-developed TensorFlow AI and machine learning environment is one of the most popular tools in that industry but it has so far been limited to using only the CPUs in Macs. Given that most Macs only have an Intel CPU with an integrated GPU, that’s not exactly surprising.
Things have changed with the introduction of the Apple Silicon M1, of course, that boasts of an octa-core CPU and an octa-core GPU. Synthetic benchmarks have shown how this combination is running circles around Intel-powered MacBooks as well as older NVIDIA and AMD desktop graphics. Now Apple is offering that power to AI developers on the new M1 Macs.
Apple created a fork, that is their own version, of TensorFlow that is specifically optimized for macOS Big Sur on M1 processors. The difference with the regular TensorFlow on older Macs is that this utilizes both the CPU and the GPU of the late 2020 MacBooks, yielding significant improvements not only in speed and performance but also in power efficiency.
Most end users won’t probably care about this breakthrough but it’s an important message that Apple wants to get across. Not only does it show off the power of its new Silicon M1 chip and its new ML Compute framework, it is also pretty much selling the new MacBook Pros as something even developers with heavy AI and machine learning requirements will want to use, expanding its customer base beyond office users and multimedia creatives.