1 / 16

Modelling in an industrial setting.

Modelling in an industrial setting. Jan Friso Groote. Silicon wafers for integrated circuits. ASML waferscanners. I am working one day a week at ASML (even more this spring). What is a waferscanner. Load path. Unload path. Measure. Expose. Vacuum.

Download Presentation

Modelling in an industrial setting.

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Modelling in an industrial setting. Jan Friso Groote

  2. Silicon wafers for integrated circuits. / Computer science

  3. ASML waferscanners. I am working one day a week at ASML (even more this spring). / Computer science

  4. What is a waferscanner Load path Unload path Measure Expose Vacuum / Computer science

  5. Some approximate facts about the machines. Number of software components: 2500. Number of lines of code: 50M. Number of software engineers: 1000. Price per machine: up to 100Meuro. Failure of the software is highly undesirable. Customers want to have new products as quicklyas possible. / Computer science

  6. Model based software engineering. ASD: State based. Data in the form of state and data variables. Code generation. Verification: deadlock freedom, absence of life lock, validity of invariants, initialisation of variables, absence of illegal behaviour, failure trace inclusion of implementation and interfaces. / Computer science

  7. Verum: ASD Language based version: Dezyne. / computer science

  8. Results for ASML Use of formal techniques lead to: • Efficiency up by a factor 2-3. • Learn in time down by a factor 2-3. • Quality: the number of problems in the field did go down dramatically. / Computer Science

  9. Results at Philips Healthcare Necessary: design for verification. At Philips Healthcare software is developed by staff using formal methods. Is this better? Yes, up to 10 times less bugs, up to 3 times faster. Industry standard 5-50 bugs/Kloc. Formal techniques 0.7 bugs/Kloc. / Computer science

  10. Verification is done by FDR2/mCRL2 Translation ASD FDR2/mCRL2 Result/counterexample Verification of the systems with hundreds of components can be done in minutes. / Computer Science

  11. ASD design philosophy. Provided interface Interface P Design model D Required interfaces Interface R1 Interface R2 Interface R3 Proof obligation: D || R1 || R2 || R3 ⊆ P (weak failure trace inclusion) / Computer science

  12. Structure of ASD control systems. Upper interface Controlled hardware/software Largest component at ASML: LOPW responsible for wafer/reticle logistics in the machine. Approximately 250 interfaces, 200 design components. 200.000 rule cases. / Computer Science

  13. Issues: the ‘diamond’ problem. Common component. Verification is only possible if the common component is ‘stateless’. / Computer Science

  14. Issue: end to end properties. On MotorOn If On happens, the action MotorOn will unavoidably occur. [true*·On]μX.[MotorOn]X∧〈true〉true / Computer Science

  15. Issue: model complexity. Typical software complexity reoccur when formal models become big. • Models require documentation. What is the appropriate style? • Which style of modelling leads to maximal verifiability/maintainability? • Model metrics assess the quality of a model. Which metrics are appropriate? / Computer Science

  16. Conclusion. Model based design and verification of behaviour leads to a 10-fold increase in quality and a 3-fold increase in development speed. More complex properties on more complex ‘model architectures’ should be verified. The set of models is becoming so large that theyspark interest in meta model questions like: - comprehensive documentation of models.- methods to assess the quality of metrics.- ... / Computer science

More Related