F1: Compute-in-X (CiX): Overcoming the Data Bottleneck in AI Processing

2022 IEEE International Solid- State Circuits Conference (ISSCC) · 2022

As network- and dataset-sizes keep growing, the motion of data - traveling from the various tiers of the memory hierarchy to the computeengines and back - becomes critically important. In the worst-case, such data-motion could start to constrain further improvements in both system-performance and energy-efficiency of future AI processing systems.

Read the paper · More papers on PaperTik