The star announcement of Apple's presentation yesterday was the staging of the M1 Ultra chip. The M1 family of chips designed in-house by Apple to unify its devices has grown with a chipset that, simply put, joins two M1 Max chips to combine their power.
Apple claims that will rival Nvidia's RTX 3090 graphics card, the fastest GPU on the market right now. Apple's new M1 Ultra chip is a combination of two M1 Max chips, fused together to create a single powerful chip.
Apple's most powerful chip to date
The M1 Max has a high-speed interface that allows Apple to combine two M1 Max chips into one. The result is an M1 Ultra chip that has double the CPU cores, double the memory, double the memory bandwidth, and double the GPU cores.
Apple calls this combination UltraFusion, and is effectively Apple's proprietary 2.5D chip packaging implementation. They claim that this technology allows them a bandwidth of 2,5 TB/s between the two M1 Max chips. This is a huge jump in bandwidth from what Nvidia offers with NV Link for SLI or AMD with Infinity Fabric, which are used as high-speed links between GPUs.
Apple's high-speed link for the M1 Ultra means that the two separate M1 Max GPUs will show up as a single GPU on macOS. This allows applications to easily harness the combined power of both GPUs in a unified manner. This should mean that apps and games don't have to do anything special to use the full power of the M1 Ultra chip.
The opposite happens with the games on the other platforms. Many pieces of software have had to be updated to natively support Nvidia's SLI implementation on Windows in order to see any performance improvements. Also, it's not a foolproof formula given that Nvidia has dropped multi-GPU support with its RTX 30 series, and combining two RTX 3090s for productivity or gaming rigs has very mixed results. The union of two graphic processing units could be big news for gamers and professionals who might have a lot of output ports and a lot of graphics power.
Source: The Verge
