Alec Stanculescu, Fintronic USA - PowerPoint PPT Presentation

1 / 20
About This Presentation
Title:

Alec Stanculescu, Fintronic USA

Description:

Radiation Hardness using Simulation Farms Alec Stanculescu, Fintronic USA Alex Zamfirescu, ASC MAPLD 2004 September 8-10, 2004 alec_at_fintronic.com alexz_at_ascinc.com – PowerPoint PPT presentation

Number of Views:68
Avg rating:3.0/5.0
Slides: 21
Provided by: klabsOrgm
Learn more at: http://klabs.org
Category:

less

Transcript and Presenter's Notes

Title: Alec Stanculescu, Fintronic USA


1
Design Verification Method for Radiation Hardness
using Simulation Farms
  • Alec Stanculescu, Fintronic USA
  • Alex Zamfirescu, ASC
  • MAPLD 2004
  • September 8-10, 2004
  • alec_at_fintronic.com alexz_at_ascinc.com

2
Summary
Fintronic USA
  • Problem Addressed
  • Solution Proposed
  • Conclusion
  • Proposed Plan

3
Introduction
  • Circuits malfunction due to radiations
  • Transient errors may lead to Soft errors if an
    erroneous data is latched
  • Cumulative effects may lead to permanent ( hard)
    errors (total dose)

4
Problem Addressed
  • Reduce the cost of assessing the behavior of very
    large circuits under different radiation
    conditions by providing fast feedback to circuit
    designers regarding
  • statistics of faults
  • reliability
  • nature of degradation

5
Solution Proposed
  • Simulate radiation effect by injecting faults in
    the simulation based on the radiation failure
    susceptibility of the library cells used
  • Use efficiently a simulation farm
  • Automate the analysis of results

6
Radiation Hardness Analyzer
7
Radiation Failure Susceptibility
  • For each cell, for each wire, record
  • lttime windowgt ltfault_info_tripletgt
  • ltfault_info_tripletgt
  • ltp of a given fault-type occurring in the given
    windowgt,
  • lttype of faultgt,
  • ltp of at most one fault of the given type
    occurring in the given windowgt
  • where p stands for probability

8
Creation of Radiation Failure Susceptibility Data
  • Calibration of Simulation
  • High correlation between simulation results and
    manufactured circuit
  • Calibration Methods
  • Test all cells at various radiation environments

9
Calibration Methods
  • Fabricate all supported cells on tested chip
    selection of tests may affect calibration
  • Use existing circuits for testing
  • Use an un-hardened version of the actual circuit
  • Use transistor susceptibility based on specific
    measurements

10
FinFarmVerilog Simulation Farm
  • Enables one engineer to manage many simultaneous
    simulations
  • Network of computers
  • Farming methodology
  • Farming tools
  • Current farms use between 10 and more than 1000
    licenses

11
Farming Methodology
  • Un-partitioned circuits
  • small footprint 64-bit simulation helps
  • Split stimulus
  • Simultaneous verification of alternative
    components
  • Program feedback-based verification

12
Farming Tools
  • Perform automated
  • simulation launching
  • simulation queuing
  • collection reporting of simulation results
  • collection merging of code coverage results
  • Perform real-time monitoring
  • Perform programmed re-launching of simulations

13
FinFarm Usage
  • Regression Testing
  • Extensive Exploration of Design Space
  • Evaluation of Reusable IPs
  • Analysis of Design Trade-Offs Power, Size,
    Speed and search for optimal solution using
    feedback loop
  • Radiation Hardening

14
Rad. Hard. Simulation
  • Describing effects on circuit by a given
    radiation environment, using VCD-F format (VCD
    fault format)
  • VCD-F Reader injects corresponding faults in
    simulation
  • Launching simultaneous simulations using
    simulation farm
  • Collecting and processing results

15
VCD-F Format
  • VCD IEEE std 1364 results
  • VCD-F specifies random faults, i.e. values are
    replaced by
  • Fault types
  • Randomization info

16
VCD-F Reader
  • VCD-F Reader injects faults in simulation.
  • Nature of faults permanent, transient
  • Kind of faults stuck at, bridging, cross talk,
    gate rupture, value toggle, timing, etc.

17
Simulation Launching
  • Save simulations at various check points, just
    before inserting faults
  • Re-start simulation at latest checkpoint before
    the injected fault which caused errors
  • Launch simultaneous simulations on FinFarm

18
Collecting and Processing Results
  • Record results databases
  • Compare Results Databases
  • Isolate circuits containing at least one error,
    for detailed analysis
  • List of possible fault type
  • Produce reports

19
Summary of Solution
  • Cells are measured once and results are stored in
    IEEE standard
  • Use Monte Carlo technique to establish radiation
    susceptibility of complex ICs using simulation
  • Fast feedback to designers
  • Major cost reduction

20
Proposed Plan
  • Establish a radiation assessment lab
  • Create oversight task force
  • Government
  • Industry
  • Academia
  • Create standards for describing
  • Radiation failure susceptibility of cells
  • Distribution of faults in circuits in time
Write a Comment
User Comments (0)
About PowerShow.com