Microsoft: Scalable in field test (IFT) for modern cloud silicon

As cloud silicon grows in scale and complexity, diagnosing silicon issues in datacenters has become increasingly critical. Traditional manufacturing test infrastructure is not available in datacenters, creating a need for scalable, secure, and high-coverage diagnostic capabilities in the field. In-Field Test (IFT) extends structural test methodologies, such as scan-based diagnostics and memory testing beyond the factory into live production systems with minimal service disruption.

This presentation describes our journey designing, deploying, and scaling In-Field Test for in-house developed silicon across Azure datacenters. We discuss the system-level architecture that enables multi-gigabyte test pattern delivery, on-device execution, and reliable extraction of diagnostic results while preserving strict security guarantees. Our implementation leverages Tessent technologies including IJTAG, HB IJTAG, and SSN to align field test execution with manufacturing ATE flows.