Automated Circuit Edit Description Generation using ML/AI models

הרקע לפרויקט:

The inefficiency in the post silicon validation segment is dragging down the innovation and profitability of the semiconductor industry. A discipline in which this inefficiency is especially pronounced is Circuit Edit (CE). CE is a technology which utilizes particle accelerators (FIB - Focus Ion Beam) to pathfind problems and to prototype solutions for problems found in chips. The issue here is that this technology is currently utilized entirely manually from the bug localization step to the CE execution step. This project aims to take an important step towards the automation of this discipline.

מטרת הפרויקט:

This project's goal is to automate one of the crucial steps in the CE flow - the CE description (or request) step, in order to enable greater speed and efficiency in this discipline.
The final result of the project will be a set of synthetic data (data created by the students), and software which is capable of creating a high level CE description in electrical schematic level, given a problem description and localization.

תכולת הפרויקט:

1. Learn how to read semiconductor electrical schematics and semiconductor basics (logic gates, chip architecture).
2. Learn how to work with basic semiconductor EDA design tools (specifically with Altair's design tools).
3. Create a comprehensive list of the problem types that can occur in chip design.
4. Create a data set of problems on a working chip design, based on the researched types of problems.
5. Form the relevant data set that will be the input of the ML model, which will consist of the bug description and localization (which devices or signals cause the problem).
6. Research the ML models that would be the most suitable for the task at hand.
7. Apply the ML models to the data set and analyze the results.
8. Create corrective measures (in the weights of the model or in the feedback loop) to maximize the ML model accuracy, based on ML results analysis.
9. Document all steps of the process.
10. Propose potential improvement to the model (the application of which would be outside the scope of the project).

קורסי קדם:

Semiconductor Devices, Logic Gates, Python Programming, Machine Learning (Using Python).

דרישות נוספות:

CAD Software, Advanced Statistics, SQL Querying, Data Mining, High level Technical English, Jira, Confluence.

מקורות:

NA