SK Hynix is pushing computation deeper into high-bandwidth memory (HBM), saying a custom HBM architecture that places compute functions in the base die could improve large language model inference performance by up to 5.15 times as data movement becomes a growing constraint on computing performance.