The idea
A different way to close the gap between design and deployment.
Every chip is engineered once, for a worst-case scenario, then run that way for its entire life, regardless of how it actually performs once deployed. That is a reasonable approach given what has been possible to measure in the field. But as transistors shrink, the gap between how a chip was designed and how it actually behaves keeps widening, and it gets harder to see.
What we build
A health layer for the chip
A small region beside your design that measures the timing of every logic element while the chip runs, so the chip can report its own condition. How it works
What it changes
Uncertainty becomes a number
How much margin each device has, where it is weak, how fast it is aging and when it will need attention, measured rather than assumed.
Where it runs today
AMD and Altera FPGAs
Families from 28 nm to 10 nm class, integrated through each vendor's own flow, deployed on your site. Device support