Analysis efficiency by reuse by creating and applying libraries
This topic is migrated from NXP's Excel based requirement list for medini, entries MEDNXP009, MEDNXP016 and MEDNXP060. Background: Semiconductor design usually applies a high degree of reuse, primarily IP, in order to achieve high development efficiency. The quantitative safety analysis towards the end of a development phase is considerable effort and can date and delay the release of a product. It is a high desire to be able to perform quantitative safety analysis very efficiently and fast. For that purpose, any possibility of reuse from previous products shall be leveraged. medini has a library concept. Part of models or analyses can be put into libraries. Typical examples applied by NXP today already are failure mode libraries, safety mechanism libraries. Request: Guidance is needed on how to create and leverage in a useful way additional kinds of libraries, e.g.: Library of (partial) safety analyses for reusable IP platforms, being the baseline for product families Library of Temperature profile (standardized ones and collected customer ones from different application domains) Library with data about different semiconductor device packages ...? Questions: Creation of a library is probably not so much different than creating any other medini project. The challenge is more how to organize, distribute content and integrate (partial or fully) these libraries into the SoC specific analyses, so that the expected effort reduction and speed-up manifests, particularly for reused IP and platforms. Questions are e.g. How can all or part of an existing product safety analysis be turned into a reusable library project? How shall models be architectured so that they can easily be integrated / imported / referenced from SoC specific safety analyses How can configuration management for these libraries best be done How can libraries also support multi-die system-in-package analyses When using a library, what needs to be done to make sure it matches the current project. What shall reviewers focus on when the have to verify / confirm analyses which make use of library content? These are just few questions about use of libraries.
