1 / 21

A New Methodology for the Simulation of Soft Errors on Digital Processors : a Case Study

A New Methodology for the Simulation of Soft Errors on Digital Processors : a Case Study. S. Rodríguez INTA Torrejón de Ardoz 28850 (Spain). S. Rezgui , R. Velazco TIMA Laboratory Group QUALIFICATION http://tima.imag.fr. R. Ecoffet CNES Toulouse - France. Motivation.

ulani
Download Presentation

A New Methodology for the Simulation of Soft Errors on Digital Processors : a Case Study

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. A New Methodology for the Simulation of Soft Errors on Digital Processors : a Case Study S. Rodríguez INTA Torrejón de Ardoz 28850 (Spain) S. Rezgui, R. Velazco TIMA Laboratory Group QUALIFICATION http://tima.imag.fr R. Ecoffet CNES Toulouse - France

  2. Motivation Ground testing of digital processors is generally performed with “static strategies” (register test) or simple benchmarks ? Significance of derived results for final space applications Techniques of Fault Injection Qualification of the TS6833216A (military version) for CESAR project

  3. Outline 1. Fault Injection Techniques => CEU Injection Principle 2. The CESAR Project 3. Test Methodology 4. Targets of CEU Injection 5. Experimental Results 6. Error Rate Prediction 7. Conclusion and Future Work

  4. 1. Fault Injection Techniques • Simulator / Emulator [Asenek 98] • 80C51 ? Automation • Xception : Transient Fault Injection[Carreira 98] • Power PC 601 ? Generalization • CEU Injection (80C51 & TMS320C50)[Velazco 00] • Predicting Error Rate for microprocessor-Based Digital Architectures through C.E.U. (Code Emulating Upsets) Injection • Accepted for publication at IEEE Transaction of Nuclear Science, Dec. 2000.

  5. Currently executed program CEU Code Instruction sequence provoking the bit flip of the chosen target RETI 3 IT request (random occurrence) 1 4 PC . .. 2 stack area RAM Memory 2. CEU Injection Principle

  6. Error Rates predicted with an accuracy of 5%

  7. 2. The CESAR Project • Earth Observation Satellite Mission • Developed by: • National Institute for the Aerospace Techniques • (INTA) from Spain. • National Commission for Space Activities (CONAE) • from Argentina. • Primary Objectives of CESAR : • Cartography • Topography • Thermal Studies • Geophysics

  8. 2. The CESAR Project (cont’d) MEGA Spectrometer

  9. 2. The CESAR Project (cont’d) MEGA

  10. Our Objective • Goal: Prediction of the error rate of different applications running on the TS68332 • Project: Implementation of CEU injection technique for benchmark programs & MEGA THESIC

  11. DUT 68332 THESIC Mother Board RS232 Serial Link Daugther Board MMI User Interface DUT Board 3. Test Methodology THESIC

  12. 3. Test Methodology (cont’d) • A daughterboard designed and developed for the TS6833216A • Implementation of CEU injection technique for : • Matrix Multiplication, SORT and F.F.T. • Final application (MEGA) • Radiation Testing • Comparison CEU injection Vs.Radiation Tests

  13. TS6833216A 3. Test Methodology (cont’d)

  14. 4. Targets of CEU Injection • Codes CEU • Internal Memory • Registers (TPU,…) • Program Counter & Status Register • Stack Pointer • …

  15. 5.Experimental Results TABLE I: Sensitive area occupation for each tested program TABLE II: Results of CEU fault injection experiments

  16. 5. Experimental Results (cont’d) • Different error rates • Dependence with the software • Overestimation in case of register testing • Exhaustive experiment in Program Counter TABLE III: Results of CEU fault injection experiments in the PC

  17. 6. Error Rate Prediction • CEU-injection sessions performed with the studied application program • CEU error rate (CEU) • Radiation testing (static strategy) • Underlying SEU cross-section (SEU) • SEU error rate prediction • SEU= SEU X SEU

  18. 6. Error Rate …. Duty Cycles (Cont’d) • Radiation testing (static strategy) •  Individual registers SEU cross-section (i) • Duty cycle predicted fi (performed by exhaustive experiment in cycles and position) • Prediction of error rate of an application

  19. Underlying cross-section Error Rate Saturation 10E-2 SEU Measured 10E-4 SEUm SEUp 10E-6 Predicted Threshold LET 10E-8 10 20 30 40 50 LET (Mev/mg/Cm2 ) 6. Error Rate Prediction (cont’d)

  20. 7. Conclusion and Future Work • CEU injection performed for three benchmark programs • Advantages • Reduced intrusiveness • Low cost • Flexibility of automation • Limitations • Effects of SEUs during instruction execution can not be simulated • Not all sensitive areas can be reached

  21. 7.Future Work (cont’d) • Ground testing for the TS68332 under heavy ions • Cf. 252 (ONERA - DESP Toulouse - France) • Tandem Van de Graaff (IPN -Orsay - France) • Conclude about the error rate prediction efficiency

More Related