1. A New ATPG Algorithm for 21st Century: The wojoRithm
John Sunwoo
Electrical & Computer Engineering
Auburn University, AL

2. Table of Contents Flow Diagram PI Assignment Mechanism
Logic Simulation (How & Why + Demo)
Backtrace()
Example
Results (Developing)
Significance & Suggestions on future work
Q & A

3. High Level Algorithm 1)Assign PIs fully (Unique) 2)Simulate
3)Fault site activated? If not, then change PIs and goto step 2
4)Fault propagates to PO? (D or D-bar detected?) If not, set an additional objective to make it propagate (change PIs) and goto step 2
5)Save the vector
6)Done
Activate the fault site
propagate the fault
3) save the vector

4. Why Start with Fully Assigned PI?
Start with X-state
D, PODEM
FAN, TOPS, SOCRATES
Before let the algorithm to think (ex. Backtrace), why don’t we give it a shot. (FEMT, Free Extra Missile Theory)
Can be lucky

5. High Level Flow Diagram
Start
Define an objective ( ex . Z - sa 1 ,
then Objective is Z = )
Objective s
achieved ? { _
Check
Backtrace
To PI
change a bit of PI } no
Assign all PIs to its non
controlling
values
Apply PIs to the circuit
Fault signal
propagates through output
Find the conflict
make
it as a secondary
objective
yes
Save Vector
Exit Done

6. Initial Primary Input Assignment

7. Initial Primary Input Assignment (2)
Waste of step!
Why non-controlling values, not Random?
Random = 50% to activate fault site accidentally
Non-controlling values = ? (Add-in Intelligence)
Higher chance to make an effective change on the circuit 1
Random 1 1 1 N - C 1 1 1

8. Logic Simulator Reads in circuit.lev
Logic evaluation from lowest level to highest level.
myVector : A linked list to be updated simultaneously.
Demo

9. Backtrace( ) Backtrace(label, value) { while (PI has not updated) {
1) check input of gate ‘l’
2) change one of the input ‘x’ which needs to be change to ‘v’. If none of the input value needs to be change then backtrack.
3) l = l
4) x = v }
return new myVector

10. Example (c17.bench)
myVector = [13, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1$, 0$]

11. Example Decision Steps (c17.bench)6 1 12 10 1 2 1 10 1 8 3 1 1 7 4 1 1 13 11 1 D* D 5
sa1 9 1
Initial objective: set [9] to “0”
Apply vector
Backtrace from [9] & *save the path (learning)
Apply updated vector
D-bar propagates?
Additional objective: set [8] to “1”
7. Backtrace from [8]
8. Apply updated vector
9. D detected at PO [13]
10. Save vector, Done

12. Data <Demo> wojoRithm 5 01111 10011 10001 11111 11100 01101
11011
11101
10100
HITEC 5
00011
00010
00100
10001
10111
11111
10000
01001
10011
01100
One vector short!
Both got 100% Fault coverage
Compact test vector size
is possible

13. Significance & Suggestion
Complete implementation on software
Experiment on bigger bench circuits
Artificial Intelligence (AI)
Random PI VS Static non-controlling PI

14. Conclusions Algorithm is not fully implemented More analysis is needed
The simplest but powerful
All node is in known state = faster % simpler backtracing.

15. Questions?