Apple unveiled its latest generation of iPad Pro tablets last month, much to the surprise of many, but instead of a high-end version of Apple’s latest A13 soc, it uses a soc that apple calls the a12z, apparently indicating that it is based on the same vortex/tempest architecture as the previous A12x Bionic, the processor on the 2018 iPad pro.
Apple’s unusual move raises suspicions that the A12Z may not be a new chip at all, but that it is essentially the A12X Bionic, a claim confirmed today in a techinsights report.
In a brief tweet, the technical analysis and reverse engineering company released a note publishing its findings, along with a side-by-side comparison of the A12Z Bionic and the A12X Bionic processor. In short, the two chips appear to be identical, with each function block on the A12Z being positioned and sized exactly the same as the A12X Bionic, which only opens the A12X’s eighth unturned GPU core.
While TechInsights’ chip screenshot analysis doesn’t bring out more subtle details, such as the chip’s step- – whether the A12Z is newer, the A12Z doesn’t bring anything new when it comes to the chip.
In contrast, the significant change between the two chips is their configuration: the A12X has only seven GPU clusters enabled at the time of shipment, while the A12Z has enabled all eight at the time of shipment. While this is a small change, it makes sense for Apple to enable the 8th cluster for higher performance. The A12X is produced on TSMC’s 7nm production line, and when released in 2018, the A12X is one of the largest 7nm chips and should have a higher yield after 18 months, so it can open the eighth GPU core.
As for why Apple chose to reuse the A12X on the 2020 iPad Pro rather than debug the A13X. We can only guess, but it should be due to cost because the tablet market is very different from the smartphone market. Apple is almost unbeatable in terms of high-performance Arm tablets, and even so, the number of iPads they sell has been unmatched by the number of iPhones.
As a result, there are fewer devices to compete with, and the cost of chip development continues to rise as each generation of lithography technology is updated. In short, at some point, it makes little sense to create new chip designs for mid-range products every year.