Focus group: Statistical synthesis

1
Focus groupStatistical synthesis
2
Top reasons to go for statistical
3
Objects
YES (Yuji, Michel)
NO (10 is too much)

• How to keep clock as a golden reference?
• regular structures (meshes)
• common path pessimism removal

4
Objects
- Is analysis needed?
YES (Yuji, Michel)
- Are we desperate?
NO (10 is too much)

• How to keep clock as a golden reference?
• regular structures (meshes)
• common path pessimism removal

5
Break in Design Chain
Foundries
EDA companies
6
Rule of games I. Corner cases
More corner cases (up to 16)
Problems - Worst and best are far
7
Rule of games I. Corner cases
More corner cases (up to 16)
Problems - Worst and best are far
- Monotonicity???
Assumption Life in a silicon world is boring
8
Rule of games I. Corner cases
More corner cases (up to 16)
Problems - Worst and best are far
- Monotonicity???
Life is beautiful!!! (Benini)
Corner case analysis cannot solve the problem
9
Rule of games II. Avoiding risk
Foundry
EDA Company
Sigmas are added
10
Optimization room
Confidence margin must be big (chips work) But it
is fully unknown
Statistical analysis and synthesis might help to
quantify risk (reduce confidence margin and be
structure specific)
Statistical analysis might help to trade off
confidence margin and yield!
11
Optimization room
Confidence margin
Confidence margin must be big (chips work) But it
is fully unknown
worst
Statistical analysis and synthesis might help to
quantify risk (reduce confidence margin and be
structure specific)
Statistical analysis might help to trade off
confidence margin and yield!

• But testing is a big obstacle!!!
• delay fault testing is very difficult
• - at-speed testing is very costly (BIST could be
  an option)

12
Optimization room
Confidence margin
worst
Focus of STA
13
Can I start coding now?
Critical path
14
Can I start coding now?
?
Reconvergence needs some care

• Numerical computation of a distribution
• Approximate convolution (5 accuracy)
• Use upper and lower bounds (10 diff. Blaauw03)

Conclusion we are almost there (check normality
assumption, justify approximations)
15
Todays vitamins or medicine?

• Use of statistical models from foundries
• Measuring systematic component of variability
  (IBM inverter rings)
• Exploiting locality in placement of critical
  parts
• Averaging variability by using deeper logic (or
  balancing logic)
• Latch-based designs to relax synchronization
  constraints

16
Tomorrows medicine?

• Regular non-standard cell design structures
  (PLA???)

Gates dominate wires -gt Standard cells are
good Submicron paradigm shift cheap computation
vs costly communication

• Introducing redundancy

• Conscious redundant encoding (a-la asynchronous)
  for smart
• Duplication of critical parts (different
  implementation) for dummies