Entropy tells us where a design hides information from the tester. Test points go where they buy the most observability.
Signals are read as spectra, not coin flips. Competitors use simple probability theory and no frequency analysis at all.
More accurate testability measures that account for signal correlation, which competitors' measures ignore.
We use a series of 10–50 tournaments to insert test points. It makes better decisions and keeps compute time low.
Reduced area → fabrication line cost saving. The biggest factor.
Reduced number of tests → test equipment saving.
Reduced % untested faults → labor cost saving.
Lowers computer time across the insertion run.
Fewer memory modifications to the chip.
Source: direct chip layout experiments with our test point inserter using Cadence tools.