Development of a Python Library for the Analysis of Embedded Non-Volatile Memories Test Data

This thesis covers the development of Clappy, a Python-based software library for analyzing test data generated from embedded non-volatile memories (NVMs). The work, which was carried out with Infineon Technologies Italy S.r.l., aims to address the increasing demand for reliable, adaptable, and scalable tools to handle the complex data generated during semiconductor testing, especially in the automotive industry where quality and dependability are crucial. The Clappy library makes it easier to develop extensive data analysis workflows, allowing engineers to obtain useful information from test data. The architecture is divided into two primary packages: core and io. The core package implements advanced data structures and domain-specific methods for handling memory scan results, validating test outputs, identifying anomalies, and computing statistical indicators. It offers enhanced modularity and reusability by supporting both functional and object-oriented programming paradigms. The io package handles file input/output operations and supports formats such as CSV, Apache Parquet, and the EDA Flat Table Format (EFF) that Infineon uses internally to store test data. Clappy works smoothly with Infineon’s current NVM testing and data pipelines, ensuring real-world application. It enables users to read, filter, transform, and visualize data collected from tester machines. Specific features include memory scan analysis, test failure detection, binary pattern decoding, and data format conversion for external tools like Tableau. Furthermore, in order to reduce dimensionality and identify important failure patterns, the thesis investigates the use of Sparse Principal Component Analysis (Sparse PCA) on memory test data. Sparse PCA results revealed insights into test redundancy and variability, which could guide possible test selection and data interpretation optimizations. Clappy increases traceability, reproducibility, and efficiency throughout the testing process by offering a uniform and expandable toolkit for analyzing NVM test data. The library helps Infineon’s Automotive division achieve its broader aims through providing efficient data analysis operations for semiconductor devices.