As we witness Apple's generational updates of the M series of chips, the highly anticipated SKU of the 3rd generation of Apple M series yet-to-be-announced top-of-the-line M3 Ultra chip is growing speculations from industry insiders. The latest round of reports suggests that the M3 Ultra might step away from its predecessor's design, potentially adopting a monolithic architecture without the UltraFusion interconnect technology. In the past, Apple has relied on a dual-chip design for its Ultra variants, using the UltraFusion interconnect to combine two M series Max chips. For example, the second generation M Ultra chip, M2 Ultra, boasts 134 billion transistors across two 510 mm² chips. However, die-shots of the M3 Max have sparked discussions about the absence of dedicated chip space for the UltraFusion interconnect.

While the absence of visible interconnect space on early die-shots is not conclusive evidence, as seen with the M1 Max not having visible UltraFusion interconnect and still being a part of M1 Ultra with UltraFusion, industry has led the speculation that the M3 Ultra may indeed feature a monolithic design. Considering that the M3 Max has 92 billion transistors and is estimated to have a die size between 600 and 700 mm², going Ultra with these chips may be pushing the manufacturing limit. Considering the maximum die size limit of 848 mm² for the TSMC N3B process used by Apple, there may not be sufficient space for a dual-chip M3 Ultra design. The potential shift to a monolithic design for the M3 Ultra raises questions about how Apple will scale the chip's performance without the UltraFusion interconnect. Competing solutions, such as NVIDIA's Blackwell GPU, use a high-bandwidth C2C interface to connect two 104 billion transistor chips, achieving a bandwidth of 10 TB/s. In comparison, the M2 Ultra's UltraFusion interconnect provided a bandwidth of 2.5 TB/s.

Max Tech performed the first detailed teardown of the Apple Mac Studio, the most powerful Mac since Apple dumped Intel for processors in favor of its own silicon based around high-performance Arm chips built from the ground-up for its own software ecosystem. The M1 Ultra SoC powering the Mac Studio is its most striking piece of technology, with Apple attaching some very tall performance claims not just for its CPU compute performance, but also graphics rendering performance.

The M1 Ultra SoC is physically huge, with roughly similar package size to an AMD EPYC processor in the SP3 package. An integrated heatspreader (IHS) covers almost the entire topside of the package. Things get interesting under the hood. The M1 Ultra is a multi-chip module of two M1 Max dies connected on package using Apple UltraFusion, a coherent fabric interconnect that allows the various components of the two M1 Max dies to access memory controlled by the other die. Speaking of memory, The M1 Ultra features up to 128 GB of LPDDR5 memory that's on-package, This memory is used for the CPU, GPU, as well as the neural processor, and has a combined memory bandwidth of 800 GB/s. The M1 Ultra features up to 20 CPU cores, up to 32 Neural cores, and up to 64 GPU cores (8,192 programmable shaders).

Apple yesterday announced its M1 Ultra processor. It is designed to be one of the most powerful solutions ever envisioned for desktop users, and it leverages some of the already existing technologies. Essentially, the M1 Ultra chip combines two monolithic dies containing M1 Max designs. They are stitched together to create one massive chip behaving in a rather exciting way. To pair the two M1 Max dies together, Apple has designed a package called UltraFusion, which is a die-to-die interposer with more than 10,000 signals. It provides 2.5 TB/s low latency inter-processor bandwidth and enables seamless sharing of information across two dies.

What is more interesting is that this approach, called multi-chip module (MCM) design philosophy, allows the software to view these two dies as a single, unified processor. Memory is shared across a vast pool of processor cache and system memory in a single package. This approach is software agnostic and allows hardware to function efficiently with loads of bandwidth. Apple notes that no additional developer optimization is required for the new processor, and the already-existing stack of applications for M1 Max works out-of-the-box. Talking about numbers, the M1 Ultra chip has a potential main memory bandwidth of 800 GB/s, with up to 128 GB of unified system memory. We are yet to see how this design behaves as the first Mac Studio units start shipping, so we have to wait for more tests to check these claims out.