According to Apple, the M1 Ultra, which will debut in the new Mac Studio, cleverly connects the die of two M1 chips for some pretty stellar improvements to performance. Here are all the key stats you need to know...
Apple claims the M1 Ultra has a 64-core GPU which is eight times that of the original M1. That feat is impressive in its own right, but even more so when you consider that the M1 is still relatively nascent in the chip world. Clearly, Apple’s engineering efforts are paying off.
The M1 Ultra manages to increase performance drastically while still using 100W less power than a 16-core PC desktop. In its footnote, Apple doesn’t mention which PC the Ultra was tested against, but last time it made M1 performance claims of this nature, they lived up to the hype. Only time and independent benchmarks will tell.
Apple says it managed to fit a mind-bending 114 billion transistors in the M1 Ultra making it the “most ever in a personal computer chip.” Admittedly, from that standpoint, the Ultra makes Intel’s 2nm architecture with 50 billion transistors look pretty quaint. Moore’s Law who?
The M1 Ultra is connected by what Apple is calling “UltraFusion” — an architecture that uses a silicon interposer to translate 10,000 signals between the two M1 chips. In all, that gives the M1 Ultra 2.5TB of “low-latency, inter-processor bandwidth” which Apple claims is more than four times the bandwidth of leading (again unnamed) competitors that use chip-interconnect technology.
One of the biggest benefits of Apple’s M1 chips isn’t just that they greatly increase performance, it’s that they do so while also using less power. That equates to better battery and, theoretically, they’re also easier on the environment. Over the course of a year, Apple claims a Mac Studio using an M1 Ultra consumes 1,000KWh less energy per year than other high-end PCs.