Presentation is loading. Please wait.

Presentation is loading. Please wait.

Paper by F.L. Kastensmidt, G. Neuberger, L. Carro, R. Reis Talk by Nick Boyd 1.

Similar presentations


Presentation on theme: "Paper by F.L. Kastensmidt, G. Neuberger, L. Carro, R. Reis Talk by Nick Boyd 1."— Presentation transcript:

1 Paper by F.L. Kastensmidt, G. Neuberger, L. Carro, R. Reis Talk by Nick Boyd 1

2  Exploring techniques for detecting and dealing with radiation-induced faults in FPGAs  Why?  Drive to use commercial off-the-shelf to minimize cost and development time (for space apps)  As technology gets smaller, radiation becomes an issue even at ground level 2

3  Incident radiation deposits energy  creation of electron-hole pairs and secondary ionizations produce transient current pulse  Can change a ‘0’ to a ‘1’ or vice versa, often called “bit flip”  In combinational logic: Single Event Transient (SET)  In sequential (or memory): Single Event Upset (SEU) 3

4  In FPGAs there are further considerations  SEU in the configuration SRAM (logic, routing)  SET in combinational FPGA fabric  SEU in BlockRAM 4

5  Effects of SEU in configuration fabric 5

6  TMR = Triple Modular Redundancy  Logic is triplicated and results are accepted by majority vote  Everything is tripled; including combinational, sequential, routing and i/o 6

7  Benefits  Able to detect and correct SEU\SET anywhere in the FPGA  No performance penalty  Drawbacks  Very large area/resource penalty (particularly problematic for i/o pads) 7

8  A new technique proposed by the authors of this paper  DMR-CED: Double Modular Redundancy with Concurrent Error Detection  Motivation: Want to find a way that is as reliable as TMR in detecting/correcting errors with less area overhead 8

9  CED = Concurrent Error Detection  Exploits some property of the logic block to find error  Time-redundant examples:  bit-wise inversion  re-computing with shifted operands (RESO)  re-computing with swapped operands (REWSO) 9

10  Result calculated from direct input and stored  Input then encoded, new result calculated and decoded  Two outputs compared – should be equal 10

11  How can we use CED?  Only duplicate combinational logic  Use CED to determine the faulty module only if there is disagreement 11

12 12

13  Three sample sequential circuits tested  8-bit multiplier  8-bit ALU  FIR filter  Sample circuits generated then each node was replaced with a multiplexor which either passes ‘correct’, ‘0’, or ‘1’  Able to simulate every possible SEU fault 13

14 14

15 15

16  Benefits  Reduces area required for combinational logic (by a significant amount in some cases)  Drawbacks  Significantly more complicated due to CED  CED circuit needs to be chosen to be optimized for each combinational circuit you protect  Speed reduced by as much as 50% 16

17  Reasonably well written and complete  Necessary to read the references to understand the minutiae of underlying principles  DMR-CED probably only useful under very specific conditions 17


Download ppt "Paper by F.L. Kastensmidt, G. Neuberger, L. Carro, R. Reis Talk by Nick Boyd 1."

Similar presentations


Ads by Google