Somewhere over the rainbow there’s a MacBook Pro 16-inch winging its way to an unboxing appointment here later this week. Let me explain why it’s coming with an M3 Pro chip and not an M3 Max.

Early in the transition to Apple silicon, I decided to continue my practice with Intel Macs, of using a desktop system for production, and keeping a notebook for travelling and as a fallback. Until Apple drops support for Intel Macs, that requires my trusty old iMac Pro and an Intel MacBook Pro to cover old models, and a pair of Apple silicon Macs.

When the original M1 chip first became available, the latter were a Mac mini M1 and a MacBook Pro 13-inch M1. Once the M1 Pro and Max started shipping, they were replaced by my current MacBook Pro 16-inch M1 Pro and a Mac Studio M1 Max with its gorgeous matching Studio Display.

It wasn’t so much that I was unimpressed by the M2 range, just that the distance between M1 and M2 versions of those products didn’t justify such rapid replacement. That has changed with the M3 series announced at the Scary Fast event almost a fortnight ago. The difference now is the distance between the basic M3, M3 Pro and M3 Max and their M1 predecessors.

My M1 Pro has eight CPU P cores, two E cores, and a GPU with 16 cores, making it 8P+2E+16G; the M1 Max is the same apart from having 24 cores in its GPU. When I’m looking at CPU cores for articles here, that’s convenient, as the differences between the two are so small as to be barely noticeable. In that initial design cycle, Apple only had two underlying designs, for the M1 of 4P+4E+8G, and the Pro/Max, which is then doubled for the Ultra at 16P+4E+48G. What few noticed at the time was how the Pro, Max and Ultra favoured multi-core benchmarks, with their four-to-one ratio of P to E cores. This was a compromise towards performance rather than energy efficiency, and it paid off with reviewers.

Apple’s second cycle brought all-round increases in GPU cores, and the M2 Pro/Max gained two more E cores to 8P+4E, again doubled in the Ultra at 16P+8E. This widened the performance difference between the basic M2 chip and its more costly siblings, but didn’t do much for multi-core benchmarks as there was no change in numbers of P cores.

It’s now in this third cycle that Apple is diversifying the range, with

• M3 basic 4P+4E+8G
• M3 Pro 6P+6E+18G
• M3 Max 12P+4E+40G
• M3 Ultra most likely to be up to twice a Max, at a monstrous 24P+8E+80G.

With those three underlying designs comes greater diversity, and a better choice for users.

As I’ve cautioned before, simply counting cores and throwing in multi-core benchmarks contributes little to deciding which is most appropriate. Although there has been relatively little change in core architecture, M3 E cores can now run at a maximum frequency of 2750 MHz, and P cores at 4050 MHz, compared to the M1’s 2064 and 3228 MHz (Pro and above), increasing their maximum throughput by a third and a quarter relative to those of the M1, although at increased power consumption.

Running tight in-core code loops, M1 E cores can complete instructions at just over 60% the rate of P cores, but in doing so each E core uses 10% of the power of a P core at maximum frequency. Performing the same task, you can expect each E core to accomplish it using around 20% of the total energy used by a P core. Light up the whole of an M1 Pro or Max, and its CPU cores are likely to burn power at just under 8 W, an M3 Pro at between 7-8 W, and an M3 Max about double those, at 15 W or more. Differences are likely to be even greater for their GPUs. It’s thus not surprising that Apple’s claimed battery life of “up to 22 hours” was measured on the MacBook Pro 16-inch M3 Pro, rather than the M3 Max.

If early reports are to be believed, M3 Pro and higher variants bring a major internal change to the CPU cores. In M1 and M2 chips, CPU cores have so far been grouped into clusters of two or four, even in the Ultra with its 16 P cores. All the cores in a cluster share L2 cache and are run at the same frequency. While that has its advantages, it has been proposed that maximum cluster size in M3 chips has risen to six, which conveniently fits the new core numbers. That would in turn change power draw under different loads, and may well lead to different frequency management strategies on the part of macOS.

One striking feature of M1 Pro and Max chips has been their tendency to run E cores at maximum frequency in situations where basic M1 chips kept them to half maximum. It will be of practical significance to discover how willing macOS is to run all six E cores in an M3 Pro chip at higher frequencies, and the effects of that strategy on both power use and performance.

So I’m hoping that the arrival of my M3 Pro will enable me to tell you exactly what the Wizard of Oz has been up to.