New technologies keep changing the way we live, and tech fans always wait to see what comes next. One of the most recent innovations comes from Apple. The company announced a new processor that will power its Macs.
The Apple-built M1 chip is the company’s first in-house PC chip. It is the first time since 2006 that new Macs will run on Apple’s own chipset instead of Intel processors. After launching the iPhone 12, Apple held another event to show its new hardware. The event covered updates to the MacBook series and the M1 chip, and it debuted the latest range of macOS devices.
The new Macs advertise incredible specs. The M1 chip brings CPU speeds 3.5 times faster than the previous series and GPU speeds 5 times faster. Its Neural Engine should boost machine learning about 15 times, and Apple advertises up to 20 hours of battery life. Apple says the M1 will accelerate machine learning drastically, and here is how.
Apple M1 Chip Architecture and Specs
Apple says the M1 is the first PC chip built with 5-nanometer process technology. The 8-core chip has over 16 billion transistors, and Apple’s Neural Engine enhances machine learning. Architecturally, the chip packs 1 GPU, 1 CPU, 1 Neural Engine, and more DRAM units into one processor.
The 8-core CPU has 4 high-performance cores for tasks and data processing. This setup makes the chip more efficient and lowers power consumption. Apple does not advertise any speeds for the processor on its website. However, some cross-platform benchmarks say the base frequency is 3.2GHz, which is higher than the Intel Core i9-9880H in the 16-inch MacBook Pro.
The M1 has an integrated 8-core GPU that executes about 25,000 threads. Apple claims it is currently the fastest integrated GPU for PCs on the market. With 2.6 TFLOPS (FP32), it is certainly impressive in overall performance.
Data science values for GPU performance suggest it would probably work well with video and image processing, gaming, and Apple Arcade. However, the GPU will not work well for machine learning. Still, the secret might lie in the M1 architecture. It comes with a Neural Engine that Apple designed explicitly for machine learning.
Apple’s Neural Engine
Think about how much computing power office work needs. A low-powered Chromebook would probably be more than enough. Serious machine learning programmers are different, because they need all the processing power a CPU can give them.
Most of the hype around the M1 chip revolves around the Neural Engine. Apple specifically designed it and built it into the chip to tackle machine learning. But can the M1 compete against the most reliable Intel processors for machine learning known today? If Apple’s claims are true, it could be a game-changer in the industry.
For the new 16-core Neural Engine, Apple claims about 11 trillion operations per second (TOPS). Data science might indicate that the M1 adds up to 6 times faster GPU performance and almost 4 times faster CPU performance. The same figures point to machine learning that is 15 times faster than other chips.
For ML compute, the Neural Engine is the component that can accelerate machine learning up to 15 times. But since these are only claims, let’s look at the data science from tests on the pre-release ML platform and a MacBook system with the M1 chip.
Intel Processors vs. Apple M1 in ML Compute
Everyone in the ML and artificial intelligence world has used TensorFlow, the most popular open-source ML platform, owned by Google. One of TensorFlow’s leading product marketers found that a 13-inch MacBook Pro with an Apple M1 chip, 256GB SSD, and 16GB of RAM saw significant performance jumps. The system ran the optimized version of TensorFlow 2.4, and the jumps appeared when compared with Intel-based MacBooks.
The M1-powered MacBook Pro had an 8-core GPU, an 8-core CPU, and a 16-core Neural Engine. It delivered better ML performance than a 1.7 quad-core Intel Core i7-based MacBook Pro. A newer model beating an older one is expected. The real eye-opener came when the M1 MacBook outperformed even the best Intel Mac Pro systems.
M1 MacBook Pro vs. Intel Mac Pro
The 13-inch MacBook Pro had better ML compute capabilities than a 3.2GHz 16-core Intel Xeon W-based Mac Pro. That Mac Pro also had AMD Radeon Pro Vega 2 Duo graphics, a 256 GB SSD, and 32GB of RAM. All the tests used pre-release macOS Big Sur and TensorFlow versions 2.3 and 2.4.
The improved ML compute over the Intel competitor came from a much faster M1 chip and an optimized version of TensorFlow 2.4 for it. As a result, after testing on improved hardware and software, the system runs 4 times faster than previous system versions on old software.
The M1’s speed, integrated memory for reduced latency, and 20-hour battery life could make this system a game-changer in data science and machine learning. Many people anticipate healthy competition between the HP Z6 G4 Workstation and the M1 MacBook Pro.
However, ML developers who never considered buying a Mac for ML compute might want to consider the M1-powered Mac after its release.
Other New Releases From Apple
Apple’s big announcement was the release of the all-new M1 chip, which will replace Intel processors in the new MacBook series. Apple also mentioned updates for each model and the latest version of macOS, called Big Sur.
MacBook Air
The MacBook Air has been the best-selling 13-inch laptop on the market for some time. Apple plans to keep it that way with the new, improved model. With the M1 chip, the latest MacBook Air will be 3.5 times faster than its predecessor. Apple says the 8-core GPU will enhance video performance and process graphics 5 times faster. The most significant boost is, of course, ML compute, which is also allegedly 9 times faster than the previous version.
MacBook Pro
The MacBook Pro is not far behind, and Apple claims its specs meet more specialized demands. The added functions should improve graphic design and audio and video editing, so it is a fine addition for students in those fields. The M1’s image signal processor also promises better image quality and clearer audio for remote workers on video calls.
Mac Mini
Nothing compares to the Mac Mini for maximum portability, especially with the improved M1 version. Its specs promise to make it 3 times faster than its predecessor. The Mac Mini also features 6 times better graphics performance and 15 times better ML compute than the previous version. All of that fits in the smallest possible unit. Apple also claims its advanced thermal design will keep it cool for longer.
Volico Apple M1 Bare Metal Servers
On top of that, Volico Apple M1 Bare Metal Servers come in two configurations. With the single-tenant dedicated server, you are the only user, so you can access all its resources. The unit benefits from the DDoS Threat Detection system, and it is one of the most powerful processing units in a single-tenant environment.
Volico Apple M1 bare-metal servers come with the latest version of macOS Big Sur. With a 100% uptime SLA, the Cisco Powered Resilient Network keeps the servers running around the clock. The M1 bare-metal offering comes at an extremely affordable price, and you get personalized support from Volico for all hardware-related issues.
Conclusion
News that Intel processors no longer power Apple’s devices was shocking enough. Powering the new MacBooks with an in-house chip made it even more surprising. The M1’s advertised performance proves to be a strong competitor against some of the best chipsets on the market at the moment.
Can Apple’s new M1 chip outperform some of the best processors known so far? Is it able to deliver faster machine learning than other Intel-based systems? We will have to wait and see!
