What’s the Fuss With Apple’s M1 Ultra?
March 9, 2022During its 8th March Keynote, Apple introduced the M1 Ultra. It’s the company’s latest in-house mobile processor meant to put high-end PCs to shame. The latest addition sits at the top of the current line-up, which also consists of the M1, M1 Pro and M1 Max. Cupertino is making bold claims about it, going so far as to say that it’s the only chip that can play back up to 18 streams of 8K ProRes 422 video.
But the M1 Ultra isn’t an all-new design. In fact, it’s made up of two M1 Max chips that have been spliced together through the use of UltraFusion. This is Apple’s packing architecture that interconnects the die of two M1 Max chips. It uses a silicon interposer that connects the chips across more than 10,000 signals, providing 2.5TB/s of low latency, inter-processor bandwidth.
UltraFusion works behind the scenes to make sure that the Software sees these two M1 Max chips as a single M1 Ultra processor. It’s a more efficient way of combining chiplet-style integration to allow for more processing power. As such, the Ultra consists of a 20-core CPU with 16 high-performance cores and four high-efficiency cores. It also has a 64-core GPU.
The M1 Max can be configured to have up to 64GB RAM but the M1 Ultra has a ceiling limit of 128GB of high-bandwidth, low-latency unified memory. Apple claims that the M1 Ultra can deliver 90 percent higher multi-threaded performance than the fastest available 16-core PC desktop of similar capabilities. Yet, the former can function at peak performance using 100 fewer watts than the latter.
Apple’s new Mac Studio is the first machine in the company’s computer line up to be powered by the M1 Ultra but we should start seeing more of the higher-end ones adopt the new architecture soon.
