When Apple announced it was abandoning Intel in favor of its own ARM-based laptop processors, the big question was how fast computers with these new chips would be. In the last episode of our tutorial, Upscaled, we took a look at how the new Macbook Pro with Apple’s M1 technology performs compared to a 16 ”Macbook Pro with Intel technology and the Windows ARM-based Surface Pro X. It turns out that the M1 is very fast.
The next question is how did Apple engineers do this? Processors are not magic. Each project is a balancing act between a dozen variables. A larger cache may contain more data, but it is slower to access. Higher clock speeds increase performance, but consume battery life. This is further complicated by the fact that designing a chip can take years, but despite all the models and simulations available to engineers, it can be difficult to predict exactly how a chip will work until you actually manufacture it.
The key to Apple’s success seems to be the M1’s incredibly “wide” design. This refers to how many instructions the chip can process each cycle. Although the M1 is single-threaded, which means that each core can process only one instruction stream at a time (unlike Intel and AMD’s multi-threaded design), it can process up to 8 instructions per cycle. This is almost double that of modern designs. This still allows it to do a lot of work, even when running at a lower clock speed, which can help save energy.
This is just one of a series of smart decisions Apple made, all resulting in an impressive processor. Some see the M1’s advantage as an ARM chip, unlike the x86 like Intel or AMD, but many of Apple’s design decisions are things that rival companies could adopt (and are starting). Add the fact that the M1 is built on a cutting edge 5 nm design from TSMC, and Apple’s success is not magical.
Check out the full video for even more details about Apple’s M1 design and how other companies are working to update it.
You can see our list of fonts for this piece here.