Can ISO 26262 Random Fault Analysis be Scaled to Complex SoCs Containing 3rd Party IP? - Sponsored By Synopsys, Inc.

  • Kevin Rich, NVIDIA

Analysis of the effects of random faults on complex SoCs presents unprecedented challenges for functional safety tools. NVIDIA's state-of-the-art SoCs targeting autonomous driving consist of a dizzying array of units/IP, both proprietary and third party. The failure mode analysis of hundreds of units is necessarily performed by the hundreds of engineers with domain expertise. The SoC can be subdivided for the purpose of analysis, and in the case of third party IP, such as Arm® cores, this analysis is performed by the IP vendor. These analyses must eventually be combined into a coherent, SoC-level FMEDA (failure modes, effects, and diagnostic analysis). The more challenging aspects of this process and how they are addressed in VC Functional Safety Manager, an FMEDA tool that is the product of a Synopsys-NVIDIA collaboration, will be presented. VC FSM's integrated FMEA (failure mode effects analysis) and FMEDA development flow supporting tens of simultaneous users will be described. Particular focus will be given to the advantages of leveraging a purpose-built tool that utilizes a centralized database. Sophisticated features that smooth the path from FMEA to FMEDA, including integration with fault injection, will be discussed.

  • Date:Wednesday, October 17
  • Time:3:30 PM - 4:20 PM
  • Location:Executive Ballroom 210D
  • Session Type:Sponsored Session
  • Room:Executive Ballroom 210D
  • Pass Type:All-Access Pass, Expo Floor Pass
Kevin Rich