This article is nearly bad enough to go on AppleInsider. :-(
This guide contains everything that we know about Apple's upcoming M3 chips, and it will be updated over time as we learn more about them.
You *really* should stop using that phrase "what we know". Yes, I know it's popular, but it's also a *lie*. In most cases when you say that what you should be saying is "what we think will happen".
Like the M2 chip, the M3 chip will feature an 8-core CPU and a 10-core GPU, but we will see more
notable performance improvements when it comes to the higher-end chips.
This completely misses the point. We are likely to see *extremely* notable performance improvements even on the baseline M3: IPC gains of 20-30% even with no clock speed increase. If they do drive up clocks, then that much faster- though I expect that they won't, at least in the laptops. Just about everything in the M3 incorporates two generations of design work, not one, due to the "last-minute" changes in the M2 (and A16).
Current M-series chips are using TSMC's 5-nanometer technology, but the M3 chips will use TSMC's newest 3-nanometer chip technology. A smaller node size equates to more transistor density, which improves both efficiency and performance. 3nm chips could offer up to 35 percent better efficiency, which would allow for longer battery life for M-series Macs.
This mixes things up and also gets basic numbers wrong. TSMC claims "up to 30%" better power efficiency (not 35%), which you only get if you keep clocks the same. Boost clocks and the efficiency gain disappears.
It's sort-of wrong to say that density improves efficiency and performance. Rather, increased density is another result of smaller feature sizes, just as efficiency and performance are. Correlation as opposed to causation.
Finally, since you're not TSMC's marketing department, consider calling the processes by their actual names (N5, N3), rather than parroting the marketing lies - there is nothing in 3N that is 3nm. I too used to just say "5nm", "3nm", etc. but I was being lazy.
Apple supplier TSMC is one of the only chip companies that is able to make 3nm chips, and rumors suggest that even
TSMC's yield rates are just above 55 percent right now because the technology is so new. Apple's shift to 3nm will mark the first node update since the 5nm M1 chip came out in 2020, and it will bring a bigger performance update than we saw with the M2.
Strictly speaking, nobody's making 3nm chips. But even allowing marketing speak, no end users have 3nm-class chips yet. When Apple's A17/M3 chips ship, TSMC will be the ONLY company shipping product reaching end users. Not "one of the only".
There is no such thing as a "yield rate" for a process. There's a defect rate for a process, which results in a yield rate for specific chip designs being manufactured. That's because yield is a direct function of defect rate and chip area. Thus on any given process, yields for small chips will be much better than for large chips. That is likely to be a major gating factor on how soon we see M3 Pro, Max, and Ultra chips.
So far, Apple has used standard "M1" and "M2" chips in its lower-end MacBook Pro and MacBook Air machines, while higher-end MacBook Pro machines use "Pro" and "Max" chips. The Mac Studio and Mac Pro use Apple's "Ultra" chips.
You left out iMacs and Minis. And Mac Studio uses Ultra or Max.
Macs Expected to Use M3 Max Chip
- 16-inch MacBook Pro
- Mac Studio
14-inch MBP also uses Max.