Project 1-ECE469

1. Computer System Design – Project 1
Group 9
James Kimball (651812434)
jkimba3@uic.edu
Lakshmi Yasaswi Kamireddy(651771619)
lkamir2@uic.edu
Rajkumar Balachandran(670383564)
rbalac4@uic.edu
Rohith Prasanna(654760321)
rprasa3@uic.edu

2. Results for Part (a) :
A plot of number of faults VS value of K in K detects
Fig 1: C17 UD=0 Fig 2: C432 UD=4 Fig 3: C499 UD=8
Fig 4: C880 UD=0 Fig 5: C1355 UD=8 Fig 6: C1908 UD=9
UD-Undetected faults i.e 0 detect

3. Fig 7: C2670 UD=117 Fig 8: C3540 UD=137 Fig 9: C5315 UD=59
Fig 10: C6288 UD=34 Fig 11: C7552 UD=132 Fig 12: Trends for only 1
detect

4. Conclusions for Part (a):-
• Referring to the plot in Fig 12, it is observed that as the
number of faults in a circuit increases the percentage of one
and only one detect decreases.
• Manually the number of faults can be computed from the
plots by summation of all the peak values that appear in the
plots.
• The erratic variation in the plots for the benchmarks c499 and
c6288 can be attributed to a large number of false paths that
are created during the simulation of the circuitry.
• There is a decrease in number of faults as K increases .

5. Results for Part (b):-
A plot of number of faults VS value of K in K detects
Fig 13: c17 Fig 14: c432 Fig 15: c499
Fig 16: c880 Fig 17: c1355 Fig 18: c1908

6. Fig 19: c2670 Fig 20: c3540 Fig 21: c5315
Fig 22: c6288 Fig 23: c7552

7. Conclusions for part (b):-
• The K detect capability of a random test set is less than that of the
minimum test set generated by Atalanta for same test vector set
size.
• The random sets with sizes greater than the minimum set shown
better performance in reducing bad faults (i.e the faults which have
low k-detect ) and also shown increase of k in some cases.
• This might be because the random vectors generated are a super
set of the minimum set or atleast have most of the vectors from
minimum set.
• Random test vector does not cover all faults as atalanta does
because none of the techniques used in atalanta are considered for
the generation of test vectors.
• The random test set that was generated with 10xsize of the
minimum set could detect up to 80-90 % of faults ,but still could not
reach as good as the minimum set created by Atalanta.
Fig 24

8. Results for Part (c):-
A Plot of Number of faults VS the value of K in K detect
Fig 25 : c17 Fig 26: c499
Fig 27: c880 Fig 28 : c1355

9. Conclusions for Part (c):-
• Enlarging the size of the test vector set would help in improving the
capability of K detect.
• Comparing the plots for part (a) and part (c) it can be observed that, the k
value as well as the number of faults for a specific k-value increase for the
same set of bench marks. Due to increase in k in part (c) a single fault can
be detected by more vectors when compared to part (a).
• The percentage of one only one detect decreases in part (c) when
compared to part (a) as shown in the plot above.
Fig29

10. Percentage and contribution of each group member:-
Names Percentage of Contribution Contribution
James Kimball 25% Coding
Lakshmi Yasaswi Kamireddy 25% Data interpretation+ Data
analysis
Rajkumar Balachandran 25% Data interpretation+ Data
analysis
Rohith Prasanna 25% Data Presentation