The A9X processor used in Apple’s new iPad Pro is a beast. When the professional-grade tablet was announced back in early September, Apple claimed that it was faster than 80 percent of the PCs that have shipped in the past year. Apple also bragged that the SoC is 1.5 times faster than typical x86 processors, while offering up to two times the graphics performance of the A8X SoC found in the 9.7-inch iPad Air 2.
Now, thanks to Chipworks, we’re learning a bit more about what makes the A9X tick. For starters, the A9X forgoes the tri-core architecture of the A8X in favor of a more conventional — for Apple — dual-core arrangement. That puts the A9X more in line with the much less powerful A9 used in the iPhone 6s and iPhone 6s Plus.
The A9X is a relatively large chip built using TSMC’s 16nm process. It measures 147mm2 compared to 128mm2 for the previous generation A8X and 104.5mm2 for the A9 (TSMC version). Some other revelations include the fact that the A9X doesn’t feature any L3 cache onboard. Apple’s older A7 and A8 series SoCs — along with the A9 — make use of the L3 to avoid having to tap into the main memory. However, given that the A9X now uses a 128-bit wide memory bus, it’s likely that the L3 cache was unneeded in this particular application.
We’ve also learned that the A9X makes use of a PowerVR Series7XT GPU with 12-cores — twice the number found in iPhone 6s’ A9. That explains the graphics prowess of the iPad Pro, which is comparable to some of the discrete NVIDIA GeForce GPUs options available in Apple’s MacBook Pros.
Needless to say, the A9X is a potent mobile chip, and has no trouble helping the iPad Pro put up some respectable performance numbers in benchmarks. Even more impressive is that Apple’s “desktop class” processor is capable of such performance while still providing 10 hours of battery life in a 12.9-inch tablet with a 2732x2048 screen resolution.
Now if Apple can just fix those “blackout” issues with the iPad Pro…