This is marketing hype. Apple’s commits to the LLVM (where the shader gets compiled into GPU executable code) states the A11 GPU target is “PowerVR Rogue” ISA.

This is marketing hype. Apple’s commits to the LLVM (where the shader gets compiled into GPU executable code) states the A11 GPU target is “PowerVR Rogue” ISA. Which is identical to how it compiles for the A10. So it’s still using foreign IP for the instruction set. Maybe Apple moved things around/optimized internals? Get better path optimization, or improve IPC? But it’s weird they didn’t add optimizations specifically for that A11. So an A10/A11 GPU just run the same opcodes.