zhengqigao / CACHEFE

SRAM Failure Rate Estimation

CACHEFE

This tool "CACHEFE" is designed for estimating SRAM failure rate.

1. Introduction

Currently, the tool supports modeling three kinds of SRAM Failure Rate: Read Failure, Write failure, Sense Amplifier Failure. For more details about the failure , please refer to:

[1] Mukhopadhyay S, Mahmoodi H, Roy K. Modeling of failure probability and statistical design of SRAM array for yield enhancement in nanoscaled CMOS[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2005, 24(12):1859-1880.

Basically, MOSFET failure rate is so small ( typically ) that Monte Carlo method would be costly ( simulations required ), So here we use algorithm described in :

[2] Yu H, Tao J, Liao C, et al. Efficient statistical analysis for correlated rare failure events via Asymptotic Probability Approximation[C]// Ieee/acm International Conference on Computer-Aided Design. IEEE, 2017:18.

2. Prerequest

Currently, the source codes request g++ ( version > 4.6 ) to compile on Linux | Mac OSX platform.

3. Installation

1. Change directory to the path where your unzipped file are located.
2. Compile the source code.
3. Run the programm.

$ cd "your path"
$ make
$ ./CACHEFE -C (--Cell) argument [options]

4. Usage

./CACHEFE -C (--Cell) argument [-A (--Algo) arguement] [-T (--Type) argument]

[-C (--Cell) argument]: represents the cell number, the argument must be a positive integer and must be given.

[-A (--Algo) argument]: represents the algorithm used to calculate SRAM Failure Rate, the argument must be one of "APA", "APE", default is "APA".

[-T (--Type) argument]: represents the type of the failure,the argument must be one of"all","read","write","sa", default is "all".

5. Example

5.1. Usage Example

./CACHEFE --Cell 64 : Using algorithm APA to simulate all types of failure of a SRAM with 64 Cells.

./CACHEFE -C 32 -T read : Using algorithm APA to simulate read failure of a SRAM with 32 Cells.

5.2. Result Example

When you run the code in demo.cpp (by uncommenting the corresponding part in main.cpp), you can get results in ./CACHEFE_res, stored in txt (I have already put the demo reults there). If you have Matlab, you can use enter run draw.m in the command window to see the following images :

(1) Sense Amplifire Failure:

(2) Read Failure:

(3) WriteFailure:

6. Acknowledgement

1. Thanks to a Github contributor with cmdline parser, you can find it here:https://github.com/tanakh/cmdline

7. Supports

If you have any questions related to this tool, please feel free to contact with me: [e-mail] or [e-mail]