Reconfigurable TAP Controllers with Embedded Compression for Large Test Data Volume

Sebastian Huhn, Stephan Eggersglüß, Rolf Drechsler

In: 30th IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems (DFT). IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems (DFT-30) 30th October 23-25 Cambridge United Kingdom 2017.


The increasing modularity of state-of-the-art integrated circuit designs leads to new requirements in terms of accessibility during testing and debugging, particularly in postsilicon phases. IEEE 1149.1 Test Access Port (TAP) controllers are typically introduced to the design and certain external hardware equipment is incorporated to enable the required access. However, transferring large data through this TAP causes high costs. Thus, an embedded compression architecture is introduced to the TAP to significantly reduce the test application time and the test data volume. Here, the retargeting of the test data is a crucial task. This work presents a partition-based formal retargeting technique to take advantage of embedded compression while processing even large and high-entropic test data. The proposed technique tackles the shortcomings of previously proposed retargeting approaches, which require an impractical computational effort for large test data volume or cause an adverse impact on the test application time. For evaluating the proposed method, several different test data sets have been processed to determine suitable parameter sets. As shown by the results, this method allows to compress even huge and high-entropic test data in average by 37.3% and to compress functional verification tests for stateof-the-art industrial designs by up to 62.5%. Furthermore, any adverse impact on the test application time is completely avoided and the procedure always finishes within reasonable run-time.

2017_DFT_Reconfiguration_Large_TDV.pdf (pdf, 251 KB)

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz