Data and Test - Ken Butler: The Road to Chiplets is Paved with Data скачать в хорошем качестве

Data and Test - Ken Butler: The Road to Chiplets is Paved with Data

Road to Chiplets - Data and Test The Road to Chiplets is Paved with Data Ken Butler Advantest Semiconductor test, like many industries, is experiencing a data revolution. With unprecedented demand for electronics of ever higher complexity, and in the face of a bend in the curve of Moore’s Law, manufacturers are moving to a chiplet-based paradigm. These products enable us to cram incomprehensible numbers of components into a growing array of products, resulting in the world’s total output of a staggering 1.6x10^21 transistors in 2021 according to VLSI Research. All those components must be tested, and the corresponding amount of test data would meet any reasonable definition of “big data”. Compounding the problem, the sources of data feeding the test process are becoming more diversified, and traditional production electrical data is being augmented with equipment metadata, optical inspection data, on-chip lifecycle management data, and others, limited only by our collective creativity. The challenge is then to collect, cleanse, and integrate all these distinct data sets so that they can be utilized for workflow optimization and automation. In this presentation, we describe data analytics-related trends that we observe in the industry and across our customer base. We show how these trends are leading to an increasing need for advanced analytics to tackle a wide array of problems such as achieving parts per billion quality levels and test time optimization. We elaborate the challenges that we see such as high- performance real-time decision-making in test operations and IP and data security across a disaggregated supply chain. We give some specific examples of collaborative solutions being developed to address these issues including adaptive test solutions at parametric (e-test) and production test and show how they solve real industry problems and result in significant improvements in product decision-making, test cost, and yield. We conclude with observations about what we see as crucial next steps.