> Care to give actual data to substantiate that, Apple?
The chip seems like its designed to beat benchmarks honestly. The A12x has an incredibly HUGE L1 data cache of 128kB, but only has an L2 level of 8MB after that (while your typical Desktop / Laptops have 64kB L1 data cache, 256kB or 512kB L2 cache, and then 8MB L3 after that).
This includes a suite of tests from perl, to GCC, to h264 (CPU-only) encoding. Now, it isn't "official" SPECint, but Anandtech hopefully used the correct compile flags for the test. (We just gotta hope anandtech did everything correctly to make it an apples-to-apples comparison).
> What does that even mean? The footnote is even more obscure than the marketing piece. "Fast" measures what exactly?
Fast graphics generally refers to the ability to run graphic benchmarks well.
So benchmarks like Ice Storm (OpenGL) would run faster on the iPad A12X. GPUs need to calculate where all those triangles go after affine transformations: there are a lot of matrix-multiplications that go into GPUs.
Its a bit of a special case, so this is generally called "graphics performance".
Care to give actual data to substantiate that, Apple?
> A12X Bionic delivers 2x faster graphics. (footnote: compared to the previous generation).
What does that even mean? The footnote is even more obscure than the marketing piece. "Fast" measures what exactly?