At the second new product launch event this fall, Apple will bring a new generation of 14-inch and 16-inch MacBook Pros for professional users. Compared with the M1 chip launched last year, the two new products are equipped with more powerful M1 Pro and M1 Max chips.
Both M1 Pro and M1 Max use 10-core CPUs with 8 performance cores and 2 efficiency cores. GPUs are different. M1 Pro has 16 cores and M1 Max has 32 cores. Both chips are manufactured using a 5nm process. Recently, Twitter user @VadimYuryev, after researching the Apple M1 Max chip, believes that it has an entire hidden section on the bottom, can support multi-chip MCM packaging, and can provide up to 40 CPU cores and 128 GPU cores. In the future, Apple may continue to expand based on the architecture of the M1 series of chips.
According to this analysis, there is a hidden part at the bottom of the actual M1 Max chip, which does not appear on the rendering of Apple’s promotion of M1 Max. It is likely to be an I/O interface for stacking. If one M1 Max chip is flipped over to another M1 Max chip, through a specific interposer and packaging, theoretically stacking can be achieved. In fact, only the M1 Max chip has extra extensions, not on the M1 Pro chip. But even if this can be done, the estimated cost is a big problem.
A few months ago, there was news that Apple is developing Jade 2C-Die and Jade 4C-Die. The CPU has 20 cores and 40 cores, respectively. It uses 64-core or 128-core Apple’s self-developed GPU, and may adopt a multi-chip module design, it seems that the rumors and specifications are also matched.
In addition, according to DigiTimes, Apple will use TSMC’s N3 process to manufacture chips for Mac and iPhone models in 2023. Mac models may include M3 chips. As for the M2 series SoC, TSMC’s N5 process nodes (N5P, N4, N4P) will still be used.