Ever since Apple launched the M1 it’s been clear that the CPU was going to be trouble for Intel and AMD. Apple has now published its own power consumption figures for the M1-based Mac Mini as compared with the 2018 Intel Mac mini refresh, and the Intel systems don’t compare very well.
Apple’s published figures on its own website look good against data published by independent reviewers. The 39W peak power consumption is higher than what reviewers measured, as is the idle power. Apple, in other words, claims higher figures than measured independently. This strengthens the likelihood that the evaluation was fairly done.
The 2018 Mac mini refresh draws 19.9W idle, according to Apple, and 122W at maximum. The Apple M1-powered system is drawing less than a third of the power of the equivalent Intel rig. That’s not a great look for Intel, and it illustrates the problem M1 poses for both x86 manufacturers. Keep in mind that this is a comparison against a 14nm Intel CPU — Coffee Lake vintage — not Ice Lake or Tiger Lake. We don’t know how a six-core ICL or TGL CPU would compare against the M1, but it would likely be somewhat better on both idle and max power.
A Walk Down Memory Lane
To put this story in context, scroll down Apple’s page and take a look at the minimum and maximum power consumption of the various Mac minis that have existed through the years. The first Intel Mac mini, released in early 2006, idled at 23W and could draw 110W at peak. An updated version later that year upgraded to Core 2 Duo but kept idle and max power consumption identical.
In 2009 idle power dropped again, to 13W, while max remained at 110W. At this point, Apple had cut its idle power from 32W with the 2005 G4-based model to 13W with an Intel CPU. Furthermore, the 2009 update was superior to the 2007 model in terms of RAM speed, HDD capacity, and built-in graphics capability (the 9400M would go on to have issues of its own, but that’s a different story).
From 2009 – 2012, idle power scaling mostly stopped. Apple dropped maximum power instead; systems from this era pull 85W at most. Systems in this era featured quad cores and were upgraded from Core 2 Duo hardware to Core i7 and i5 models with integrated memory controllers and, later, features such as AVX.
In 2014, idle power dropped again, to just 6W, but the number of cores also dropped, from quad-core to dual-core. For the first time, the number of cores in the Mac mini went backward and stayed that way. When the Intel Mac mini refresh shows up in 2018, it delivers far more horsepower at the cost of higher power consumption.
The M1’s greatest strength is not its performance. While it absolutely can outperform x86 CPUs, the M1’s performance varies depending on whether a workload is emulated or native. Comparisons with Tiger Lake as opposed to Ice Lake slice into the lead it claims in certain tests. The M1 is a threat, not a one-shot knockout.
The problem with the M1, from Intel and AMD’s perspective, is that even when it loses to x86 it draws a fraction of the power doing it. And low-power, highly efficient CPUs are typically the ones that have the most room to grow. Part of the reason for the M1’s lauded efficiency is that the CPU is only running at 3.2GHz. Higher clock speeds are inefficient, and each additional MHz costs more power the higher you clock a chip.
We don’t know how well the M series scales above 3.2GHz, but if Apple can scale this design or has further significant IPC improvements ready and waiting, it’s going to get harder for x86 to compete. Data centers are very interested in lowering CPU power consumption, and while Apple probably doesn’t have plans to start selling servers again, Qualcomm just bought Nuvia, a company focused on building ARM server solutions.
For now, software ecosystem issues, user preferences, and Apple’s business decision to only focus on certain parts of the PC market will limit Intel and AMD’s competitive risk. Neither company is saying much about the M1 yet, but both of them are going to have to contend with it in the future. Laptop OEMs such as Dell, HP, and Lenovo operate on the assumption that the chips they buy are the fastest processors in the world. If it turns out that ARM CPUs are faster than x86 CPUs in a way AMD and Intel can’t match over time, somebody is going to fund the development of a competitive ARM chip to sell to companies that aren’t Apple.
Intel and AMD aren’t talking about the M1 much right now. Benchmark performance between the two ISAs, while interesting and indicative of the overall comparison, isn’t the real threat. The real threat is that Apple has more than enough room in its power consumption budget to either add CPU cores, boost clock, or both. The x86 manufacturers are in no near-term danger, but they have no time to waste, either. Both companies have affirmed that they are taking the M1 seriously. We’ll have to wait and see what that means a year or two from now.
Feature image by John Burek/PCMag.
Intel’s Desktop TDPs No Longer Useful to Predict CPU Power Consumption
Intel's higher-end desktop CPU TDPs no longer communicate anything useful about the CPUs power consumption under load.
Review: The Oculus Quest 2 Could Be the Tipping Point for VR Mass Adoption
The Oculus Quest 2 is now available, and it's an improvement over the original in every way that matters. And yet, it's $100 less expensive than the last release. Having spent some time with the Quest 2, I believe we might look back on it as the headset that finally made VR accessible to mainstream consumers.
New Intel Rocket Lake Details: Backwards Compatible, Xe Graphics, Cypress Cove
Intel has released a bit more information about Rocket Lake and its 10nm CPU that's been back-ported to 14nm.
Intel’s Raja Koduri to Present at Samsung Foundry’s Upcoming Conference
Intel's Raja Koduri will speak at a Samsung foundry event this week — and that's not something that would happen if Intel didn't have something to say.