11-28-2018 11:10 AM
My desire is to model the Zynq-UltraScale+ device for an avionics application using the Xilinx SEU Estimator. The Estimator for avionics (2016) does not have this device, the 2017 version does but it has limitations: 1) The max elevation is 15000 ft and 2) the ability to change the Neutron Flux (Cross-Section) is grayed out. How can I extrapolate from a known elevation and latitude (for instance the results using New York City) to any other altitude and latitude?
11-28-2018 12:41 PM
You need the avionics version. It is not general access, so you must request it through the support portal. Avionics is weird as no one agrees on what the flux vs. altitude really is (even the experts disagree). Is 45 Kft 600x sea level, or 900x sea level? Will nearby fuel increase thermal neutron flux? Every avionics company has their own rules. Xilinx does the best it can guessing, but what is the flight profile you are trying to estimate and what are the rules at where you work? You can use the commercial estimator and just increase your mean time to upset by your preferred factor if you need for now while you await access to the avionics lounge to get the compromise spreadsheet estimator that everyone disagrees with anyway...
Once you get it, you still need to integrate failure rates over your flight profile (time at altitudes) anyway (the tool does not help you there).
11-28-2018 01:55 PM
Thanks, unfortunately the avionics version doesn't have the Chip I want to model, the other spreadsheet does but only for altitudes below 15000 ft.....Xilinx said they are looking at it.
11-29-2018 08:03 AM
All MPSoC devices use a common library of elements, so pick one that is supported. The final numbers will be quite close enough. The estimator is only good to +/- 20% at BEST...
More important are the choices you make: ECC on vs. off, use of caches, OCM, etc.
12-06-2018 12:15 PM
Thanks, the no-avionics version has the chip I need, and although it only shows sea level New York, I can use that flux density (12.9) and then extrapolate that to any other flux density.