Intel has repeatedly stated that Lunar Lake can destroy the “myth” of Arm's efficiency, but how the company actually plans to do that is unclear. Arm has been praised for its fantastic architecture with long battery life and great performance, with Apple Silicon MacBooks being the prime example of what Arm can do in computing. Since then, the first Snapdragon X laptops have hit the market, which Also have fantastic battery life.
However, it is possible that Intel could Arm in a number of ways. I'm not saying it will happen, but the company seems pretty confident in its abilities, and there are a few viable paths to success that could see Team Blue catch up to the competition.
3 Better manufacturing processes
TSMC N3B could be better than Intel 4, but we don't really know
Intel started using what they call “tiles” with the launch of Meteor Lake, where different parts of the chip can be manufactured separately and assembled into a processor. AMD's equivalent is called a “chiplet” and the whole idea is that instead of making a processor on a single piece of silicon (also called monolithic), you have multiple chips, each containing a part of the CPU.
With Meteor Lake, Intel produced the compute tile along with other core components, while the graphics tile was manufactured at TSMC. This time, with Lunar Lake, the whole thing is being built at TSMC. While Intel 4 has shown promise so far, TSMC's proven manufacturing processes are known to work well. Better yet, rumors suggest Lunar Lake will be manufactured on TSMC's N3B node, an advanced version of TSMC's 3nm manufacturing process.
Currently, we don't have enough data from Intel 4 to judge if this is really a good thing, but we do know that TSMC's N3B performs well. The manufacturing process can go a long way in terms of power efficiency (just look at the Snapdragon 8 Gen 1 compared to the Snapdragon 8 Plus Gen 1 as a real-world example), and if TSMC's N3B is so much better than Intel 4, it's a great first step to boost Intel's efficiency.
2 Architectural changes
Where the real changes take place
Source: Intel
Intel has made some big architectural changes with Lunar Lake that may also lead to massive efficiency gains. Lunar Lake includes four “Lion Cove” P-cores and four “Skymont” E-cores, with a focus on running as much as possible on those E-cores. Intel has increased the size of the micro-operation queue while improving the out-of-order execution engine. Other changes, such as improvements to buffering and queuing functions, will also mean the cores can run more efficiently. These are all small changes, but they add up.
To get an idea of how Skymont compares to Meteor Lake's E-cores, Intel says you'll get almost three times more performance out of Skymont while using only a third of the power. These are big, big claims, but they'll immediately stand out if they hint in any way at real-world performance. There are also four E-cores on Lunar Lake, instead of two on Meteor Lake, so it should better also the overall performance.
With improvements to the Thread Director and power management controllers, there's a very real chance that Intel has found a way to reduce the power consumption of the x86 architecture. We're not far from testing devices either, but if there are any real efficiency gains, they're very likely to come as a result of these architectural changes.
1 Pure performance
If you can't reduce power consumption, increase performance
If Intel cannot reduce idle or very light power consumption, it could also try to squeeze more performance out of higher wattages. Meteor Lake was fine for battery life, and if you could squeeze significantly more performance out of the higher power targets, that could convince power users to switch to Lunar Lake. x86 is still a powerful architecture, even if it consumes a lot of power, and it could be something Intel is banking on.
To be clear, I don’t really think that will happen, given how much the company is focusing on efficiency instead, but if Lunar Lake proves to be the most powerful by far If there were a larger number of laptop CPUs, this would be somewhat forgivable, especially if it did not come at the expense of increased power consumption.