Glossary

Memory bandwidth (AI/ML)—the rate at which data can be transferred between memory and processing units, typically measured in gigabytes per second.

Memory disaggregation — a system architecture that separates memory resources from compute units, allowing flexible allocation and sharing across a network.

The growing disparity between processor speed and memory bandwidth that limits system performance in computing.

Microring resonators — a looped method for guiding light within a small footprint (nanometer scale).

Antenna technology using multiple transmit & receive antennas to exploit multipath propagation, enhancing throughput, capacity, coverage & reliability.

Mixture of experts (MoE) — a machine learning technique that uses multiple specialized models and a routing network to efficiently handle complex tasks in AI systems.

Model dimension — the measure of an AI model's structural complexity, determined by its parameters, features, and architecture that define its learning capacity.

Moore’s Law — Gordon Moore’s observation that the number of transistors on a chip would double every two years.