The Apple M1 Pro and M1 Max’s Power Efficiency Should Rattle Intel, AMD

Reviews of Apple’s M1 Pro and M1 Max landed yesterday. While most articles focused on the overall performance of the laptops, we’re more interested in the comparative performance between Apple’s larger M1 CPUs and the x86 CPUs it competes with from Intel and AMD. The competitive data now available suggests the scaled-up M1 remains a potent threat to Intel and AMD.

The problem for the x86 manufacturers isn’t necessarily raw performance. A number of reviews today, including one at our sister site, PCMag, show that there are various 11th Gen Intel laptops that can more-or-less keep pace with the M1 Pro and M1 Max in CPU workloads. There are also workloads like Cinebench R23, which show Intel’s top-end 11th Generation CPUs beating their MacBook Pro counterparts. Between the two, it might seem like Apple has matched — but not exceeded — what x86 machines are capable of. But look a little deeper, and the wheels start coming off that narrative.

!function(e,i,n,s){var t="InfogramEmbeds",d=e.getElementsByTagName("script")[0];if(window[t]&&window[t].initialized)window[t].process&&window[t].process();else if(!e.getElementById(n)){var o=e.createElement("script");o.async=1,o.id=n,o.src="https://e.infogram.com/js/dist/embed-loader-min.js",d.parentNode.insertBefore(o,d)}}(document,0,"infogram-async");According to Anandtech, the M1 Max draws 39.7W at the wall to deliver a Cinebench R23 score of 12,375. The Core i9-11980HK inside the MSI GE76 Raider beats that score, with a 12,830 — but it draws no less than 106.5W at the wall.

Chart and data by Anandtech

Anandtech’s tests show a clear pattern. x86 CPUs can sometimes match the performance of the M1 Max, but they need to draw far more electricity to deliver the same performance. In some floating point tests, the M1 Max offers more performance than any x86 CPU, courtesy of its mammoth on-chip bandwidth. Anandtech’s tests show that the CPU doesn’t have access to the full 400GB/s that Apple claims, but it still can tap ~200GB/s.

So far we’ve confined ourselves mostly to a discussion of the new CPUs, but the GPU deserves a nod here too. Apple’s new graphics solution looks amazing, if one only consults synthetic tests. Factor in its real-world gaming performance and the results are a bit lackluster. Apple’s overall performance is still up substantially compared to its previous generation of products, however, which tapped relatively low-end AMD mobile GPUs. GPU-centric content creation benchmarks also showed better performance results than gaming.

All that said, the M1 Max and M1 Pro are not automatic, must-have processors for everyone. Apple’s support for gaming is still practically nonexistent, and it’s no longer as easy to run Windows on a Mac as it once was, which may matter for some users. Also, there’s enormous inertia in the PC universe and 10 percent of the PC market isn’t just going to tromp over to macOS in a year, no matter how good the M1 Max is.

Power Efficiency is Just Performance a Company Hasn’t Tapped Yet

The biggest problem for Intel and quite possibly AMD isn’t the M1 Pro / M1 Max’s raw performance. Though it may take current x86 CPUs far more power to match Apple’s newest silicon, high-end chips from both manufacturers can hang with the top-end M1. It’s also possible that AMD’s Zen 3 would compare more favorably to Apple’s power consumption than Intel does.

It’s power efficiency where Apple’s latest systems seem to leave x86 in the dust. Every CPU is more and less efficient at a given clock speed. As CPU clock speeds rise, the amount of power required to increase performance an additional one percent grows. The high clocks Intel and AMD are forced to utilize to offer equivalent performance are not an advantage against the M1.

We don’t know what the M1 family’s power consumption looks like across its entire range of potential operating frequencies, but Apple is having no problem hammering x86 on power efficiency at 3.2GHz. This implies there’s at least some headroom left in the core. Apple could use that headroom to launch a desktop chip running 15 – 25 percent faster than its current laptop processors, or it might spend its power budget scaling the chip’s core count and improving internal parallelism. Likely, it’ll be a mix of both. Either way, this comparison is going to get tougher for the x86 manufacturers once the fight moves to the desktop arena.

CPUs that offer a combination of high power efficiency and high performance tend to succeed in the consumer market. Ryzen was vastly more efficient than the Bulldozer family of products that preceded it. Intel’s Core 2 Duo was far more efficient than the Pentium 4 architectures. Both launches ushered in a new era for their respective companies.

The length of CPU design cycles means challenges like the one Apple is mounting play out in slow motion. Intel’s upcoming Alder Lake launch will provide an updated comparison point and a first look at a high-end x86 hybrid CPU. AMD also has plans to introduce V-Cache for Zen 3 and Zen 4 beyond that, in 2022. There will be responses to this launch.

But make no mistake: If the original M1 CPU was Apple’s warning shot, the M1 Pro and M1 Max are the opening salvo. A lot of factors go into whether or not a system is attractive, including ecosystem support and familiarity, but if Apple continues to deliver better performance than x86, content creators are going to notice. It might not pick up in the gaming community — serious gamers are not well-served by Apple systems at this time — but it could let Apple make inroads in other markets.

Now that we’ve seen what Apple can do in a high-performance mobile form factor, there’s no reason to doubt the company’s ability to introduce a high-end Mac desktop with CPU performance that can match x86. It may not launch such a chip for six months to a year, and the resulting system may be far too expensive to be practical for most buyers, but the M1 Pro and M1 Max prove that Apple’s silicon can scale. It may have taken a decade longer than anyone expected back in 2011, but the long-awaited battle between x86 and ARM is finally happening, one market segment and product launch at a time.

Continue reading

Samsung Stuffs 1.2TFLOP AI Processor Into HBM2 to Boost Efficiency, Speed

Samsung has developed a new type of processor-in-memory, built around HBM2. It's a new achievement for AI offloading and could boost performance by up to 2x while cutting power consumption 71 percent.

New ARMv9 Cortex X-2, A710 CPUs Deliver Major Efficiency Gains

ARM is announcing new CPUs today for its ARMv9 architecture. The Cortex-X2, Cortex-A710, and Cortex-A510 deliver a solid set of performance improvements and efficiency gains.

AMD Wants to Improve AI, HPC Efficiency 30x by 2025

AMD has a new plan for improving energy efficiency and claims it can deliver 30x gains by 2025.

AMD’s Ryzen Mobile 6000 Emphasizes Power Efficiency Over Raw Performance

AMD's new Ryzen 9 6900HS has launched. The CPU isn't quite as fast as Intel's top-end Alder Lake but it wins marks for power efficiency and overall performance.