From Sean Hollister’s “How fast are Apple’s new ARM Mac chips? It’s hard to tell” posted Tuesday on The Verge:
The company’s press release says very specifically that Apple’s new chips will “give the Mac industry-leading performance per watt,” and that’s a very deliberate turn of phrase. Apple’s arguing that by building the most efficient kind of chips it can — “the highest performance with the lowest power consumption” — it can achieve more raw performance by tipping the scales of that performance-per-watt formula toward more watts.
In other words, if you build a MacBook Pro-sized chip with a MacBook Pro-sized heatsink and enclosure, plus a MacBook-sized battery, your iPhone-esque processor theoretically has room to do a heck of a lot more work. But it’s almost always been true that ARM-based processors are more efficient than the competition, and the scales don’t tip on their own. Speeding up a chip isn’t just a simple matter of giving it more juice — you’ve got to design a beefy enough processor (or, say, the world’s fastest supercomputer) around that efficient architecture, and Apple isn’t bragging that it’s actually done that yet.
My take: Smart. I never think about the heatsink until my laptop is too hot for my lap.
P.S. “… the scales don’t tip on their own.” Words to live by.
I’m sure there’s a lot of work going on inside Apple on tuning Mac OS X (11!), on looking at ways the hardware could change to provide better (and cooler) performance, etc.
Of course, understanding the kinds of system optimizations between hardware and software (both system software/OS and applications) is difficult. Explaining those trade-offs won’t get bloggers many clicks.
Sure you can get better elsewhere, but rarely at a cheaper price and likely with limited use cases.
Firstly, and obviously, the A12Z Mac mini, is part of the “Developer Transition Kit”, as specifically labeled by Apple. Apple and their developers have a lot of work to do before both release production A-Series Macs running production 3rd party software.
Second: Just my guess, Apple targets the first production A??Macs at the lower end of the market who use the high volume applications. The high end applications and high end Macs, come later. At the high end performance matters. This takes more time for Apple and the developers. And it’s lower unit volumes at higher gross margins.
In classic Clayton “disruption from below”, Apple helps iOS developers become Mac developers. It helps legacy Mac developers too.
This particular type of Apple CPU chip would then be designed potentially with a larger die, heat spreader, more robust connections, and adjacent memory can be separately cooled and managed. Scaling up an A14 chip from a mobile device to a desktop device could allow heat dissipation in the 25-40 watt range. Considering Intel 10th Gen Core i9-10900K itself consumes up to 250-300 watts using the 14nm process using all 10 cores at maximum boost frequency of 4.9Ghz whereas it normally runs at 125 watts. AMD chips using 7nm process also pull 200-300 watts in bursts. So Apple’s work on producing very efficient low power processors that could eventually approach Intel and AMD chip processing levels at lower power could prove very beneficial, as will their lead in moving over to 5nm process via TSMC already.