[The Exiled]

Apple announces M1 Ultra: Combining two M1 Maxes for Workstation Performance

Apple unveiled the fourth and final member of the M1 family of Apple Silicon SoCs, the M1 Ultra. Aimed squarely at desktops – specifically, Apple’s new Mac Studio – the M1 Ultra finds Apple once again upping the ante in terms of SoC performance for both CPU and GPU workloads. And in the process, Apple has thrown the industry a fresh curveball by not just combining two M1 Max dies into a single chip package, but by making the two dies present themselves as a single, monolithic GPU, marking yet another first for the chipmaking industry. Apple has bonded two M1 Max dies together on to a single chip, with all of the performance benefits doubling their hardware would entail.

At the heart of the new M1 Ultra is something a bit older: the M1 Max. Specifically, Apple is using two M1 Max dies here, and then bonding them together to form a massive amalgamation of 114B transistors. On the CPU front, this means Apple now offers a total of 20 CPU cores. This is comprised of 16 of their performance-focused Firestorm cores, and 4 of their efficiency-focused Icestorm cores. With 32 GPU cores, M1 Max was already setting records for a monolithic, integrated GPU. And now Apple has doubled things to 64 GPU cores on a single chip.

The secret ingredient that makes this all possible – and which Apple has been keeping under wraps until today – is that M1 Max has a very high speed interface along one of its edges. An interface that, with the help of a silicon interposer, allows two M1 Max dies to be linked up. Apple calls this packaging architecture UltraFusion, and it’s the latest example in the industry of 2.5D chip packaging. Thanks to UltraFusion, Apple has become the first vendor to ship a chip that transparently combines two otherwise separate GPUs. In particular, the company is touting that the M1 Ultra’s GPU performance exceeds that of NVIDIA’s GeForce RTX 3090, which at the moment is the single fastest video card on the market. And furthermore, that they’re able to do so while consuming a bit over 100 Watts, or 200 Watts less than the RTX 3090.

Izvor: AnandTech