Glossary

Profitability (AI/ML) — A metric that assesses the economic viability of AI applications by balancing performance with cost and power.

Real-time serving inference — the process of generating predictions from AI/ML models as new input data arrives, without batching.

Remote laser — an optical transceiver that separates the laser from the optical module.

Retimer — a device that reconstructs high-speed digital signals to extend reach and improve signal integrity in data transmission systems.

Scale-out (AI/ML) — A network architecture that interconnects AI compute clusters together, typically using standard networking technologies.

Scale-up (AI/ML) — A network architecture within an AI compute cluster that maximizes intra-cluster communication bandwidth and minimizes latency.

Sequence length — the number of consecutive elements that an AI model processes in sequential data, such as text, images, or time series.

SerDes — a way to send data that converts parallel data into serial data and back for more efficient transmission.