Now that the dust has settled from Apple’s third and final fall media event, we’ve made it through all of the company’s annual iPhone, Apple Watch, iPad, and Mac refreshes, toplined this year by the debut of four 5G-equipped iPhone 12 models and the first Macs powered by an Apple-developed M1 processor. Yet despite all the potential excitement that might have come from new Apple chips and designs, most of the new products weren’t just iterative and safe — they were unexpectedly boring.
We’ve moved past Apple Watches that launched with one new feature and missing features, respectively; new iPads that were basically just old iPads; and larger yet shorter-running and unpredictably performant iPhones. This week was the Mac’s turn to shine, and though Apple made good on its promise to reveal the first Macs with Apple-developed CPUs, many observers hoped the ARM-based chips would redefine Apple’s computers, not just retrofit older machines with faster processors. In other words, this week’s event could have been a huge, transformational-class moment for the Mac platform, introducing touchscreens, Face ID cameras, or any other iPhone/iPad-inspired innovation to its long-stagnant laptops and desktops — a chance to reposition Macs within the competitive PC world.
Instead, all three of the new M1 Macs are powered by the same basic chip, they’re all effectively entry-level models, and they’re just like last year’s models — but faster. In some ways. Customers can technically choose between consumer “Air” and professional “Pro” MacBook laptops, as well as a “mini” desktop Mac that straddles the consumer/professional line, but now they share the same CPU. Each M1 machine has only two USB-C ports and a maximum 16GB of RAM, and remarkably, Apple wound up keeping older Intel CPU-based models in its lineup to give “Pro” users extra ports and memory. That was a subtle sign that Apple’s chips still aren’t ready to fully displace Intel’s as of late 2020.
Apple is pitching all three new Macs as roughly three times faster than the Intel machines they replace, but it used a slick marketing trick to achieve that significant multiplier. In recent years, the company began offering entry-level Macs with conspicuously underspecced CPUs, such that the prior Intel MacBook Air and Mac mini started with Core i3 chips, while the MacBook Pro had a 1.4GHz Core i5. When it came time to make the M1-to-Intel comparison — something Apple has been planning for years — the marketing didn’t need to benchmark M1’s performance versus Intel’s fastest chips, and instead focused on some of its weakest ones.
Moreover, the M1’s performance levels weren’t exactly surprising. Credible rumors have hinted that the M1 is effectively just a rebadged A14X — the as-yet-unannounced sequel to the A14 Bionic chip that’s already powering iPhone 12 phones and iPad Air tablets, which normally would have debuted in next-generation iPad Pro tablets. The A14 launched in September with a 5-nanometer manufacturing process and 11.8 billion transistors; two months later, the M1 uses the same 5-nanometer process and 16 billion transistors, the same type of change Apple offered two years ago in the 7-nanometer A12 and A12X chips.
To Apple’s credit, the M1’s higher-performance CPU cores are claimed to be “the world’s fastest,” and there’s every reason to believe that this will prove to be accurate by comparison with other low-power chip cores. That said, Apple is touting the chip’s overall performance as “faster than 98% of PC laptops sold in the last year,” a soft metric that isn’t so much about performance as sales volume. Cheap, underpowered machines tend to sell better than expensive ones with tremendous horsepower, so if 90% of laptops sell for under $500, it wouldn’t be a surprise that Apple computers would outperform them at twice the price.
The M1 replaces Intel’s CPU, but there’s more to the change than just the central processor: Back in June, Apple chip lead Johny Srouji suggested that moving from Intel to Apple-designed chips would give the Mac access to “many custom technologies” that were missing from Intel’s CPUs. Because of those technologies, which Apple has been perfecting over years of iPhone and iPad chip development (to say nothing of the years it has prototyped ARM technology-based Macs), M1 Macs could have been different, not just more. Look over the list of what Apple promised for M1 back in June, and judge what happened for yourself:
- A Secure Enclave for “best in class security.” Apple has offered these in iPhones and iPads for years, and during the event claimed M1 has a new version, without providing much in the way of specifics.
- A high-performance GPU for “a whole new level of graphics performance” on every Mac, suited to both pro apps and games. Like the CPU design, M1 will make low-end Mac GPU performance better, but doesn’t set a new overall Mac standard.
- Neural Engines for machine learning. Apple’s 16-core Neural Engine appears to be the same one previously introduced in the A14 Bionic, with the same promise of 11 TOPS performance.
- A video display engine that could potentially improve Mac screen performance. Apple said almost nothing about the M1’s capabilities here.
- An image processing engine to improve camera performance. Apple said little more than that the engine exists and is “lightning fast,” while promising noise reduction and color rendition improvements for the MacBooks’ FaceTime cameras, but not much else.
One item was missing from that list. During the M1 reveal, Srouji made a big deal about the chip’s unified memory architecture, which unquestionably speeds up the machine and reduces power consumption. It’s a non-trivial step forward for Macs, just as the continued weaving together of multiple formerly discrete chips (think modems) will continue to be over the next several years, but “making everything faster” isn’t as obviously user-facing as a new feature.
Apple’s graphics story was similar to the CPU story. It appeared to be dunking on Intel’s integrated graphics chips by promising five to six times faster speeds from the M1’s integrated GPU, but like Intel’s basic Core i3 and i5 CPUs, its Iris Plus and UHD Graphics weren’t exactly trailblazers, or the best that modern PCs offer. So while the M1’s published raw GPU numbers are great by integrated graphics standards — 41 gigapixels per second, 82 gigatexels per second, 2.6 teraflops of processing power, and 128 execution units, each fractionally higher than Intel’s brand new Iris Xe Max graphics chip, which claims theoretical peaks of 39.6 gigapixels per second, 79.2 gigatexels per second, 2.5 teraflops of FP32 performance, and 96 execution units — neither higher-end Macs nor PCs rely on Intel integrated graphics.
Once again, it’s to Apple’s credit that those Intel numbers are for a discrete 10-nanometer processor, while Apple’s are for a 5-nanometer GPU that’s inside the same chip package as the CPU. Intel may be pitching the Iris Xe Max at “mainstream thin light notebooks,” but Apple’s solution is simultaneously more powerful, considerably smaller, and likely to consume less energy.
It’s easy to complain that the M1 Macs don’t look any different than the Intel machines they replace. The real problem is that despite all of those “custom technologies” Srouji mentioned back in June, the Macs don’t actually bring over the big features that have made iPads and iPhones leaders in their respective categories: their intuitive multi-touch displays, advanced photo/video/depth cameras, or atypically light and thin bodies. After at least two years of preparation, Apple launched M1 Macs that are just like prior Intel Macs, only with different chips inside, and — for now — apparently focused more on raising the performance floor than the ceiling for Apple’s computers.
Having replaced three of its most affordable computers in one day, and even lowered the price on one of them, Apple will have no issue selling M1-based Macs during this holiday season. That’s the safe, predictable outcome we’ve come to expect from Apple over the past decade, and one that will ultimately satisfy Wall Street; between cost savings and deeper component-level integrations, the switch from Intel to Apple chips will surely yield dividends for years to come.
Unfortunately, however, user-facing Mac innovations will apparently have to wait until 2021 or later, giving chip- and device-making rivals such as Qualcomm, Samsung, and Lenovo another opportunity to introduce ARM-based PCs with new features that make the latest Macs look like laggards rather than leaders. Apple may well have taken the lead with ultra low-powered chips, but the past 15 years of Intel Macs have proved that gaining market share in the PC world isn’t just about performance per watt, and faster clock speeds alone aren’t enough to keep users from using more versatile and convenient devices instead.
How startups are scaling communication:
The pandemic is making startups take a close look at ramping up their communication solutions. Learn how