Introduction to Formal Verification and Logic Synthesis
ο»ΏIntroduction to Formal Verification and Logic Synthesis
Yukio Miyasaka
Contents
Formal Verification ? Equivalence Checking
? Binary Decision Diagram ? Boolean Satisfiability Problem
Logic Synthesis ? Two-Level Logic Synthesis ? Two-Level Logic to Multi-Level Logic ? Multi-Level Logic Optimization
Formal Verification
? Prove that a given logic circuit meets some given properties ? Often compared to random simulation, here are some tradeoffs:
Coverage Time
Formal Verification 100%
~O(exp(circuit size))
Random Simulation #test patterns / #all possible patterns
O(#test patterns * circuit size)
Note: coverage has a different meaning for bounded model checking
Equivalence Checking
? Given two implementations, prove they are functionally equivalent
? For simplicity, assume the implementations are combinational logic circuits ? Optimized implementation v.s. Golden model (Spec)
? Exhaustive simulation is enough to prove this property ? However, it takes an exponential amount of time
? N inputs -> 2^N patterns to simulate ? O(exp(#inputs))
? Can we do it better?
Binary Decision Diagram (BDD)
? Binary tree ? Redundant nodes (equivalent cofactors) are removed ? Each node has a unique function
Randal E. Bryant, "Symbolic Boolean manipulation with ordered binary-decision diagrams," in ACM Computing Surveys, vol. 24, no. 3, pp. 293?318, 1992.
BDD Operations
? We can construct a BDD for the result of operation by a recursive procedure ? e.g. BDD z = AND(BDD x, BDD y);
? Runtime is bounded by O(#BDD nodes) while #BDD nodes is usually smaller than exp(#inputs)
x
y
z
Symbolic Simulation
n13
x0
1c4
x2
1c3
x0
0
1
x2
x1
n15
x1
1cb
x2
1ca
x3
1c5
0
1
x2 x3
n12
x2
1c4
x3
1c3
0
1
x4
n14
x2
1c8
x3
1c7
x4
1c6
0
1
Output BDDs to be compared with the golden model
n16
x0
1d3
x1
1d2
1cc
x2
1d1
1c5
y0 x3
1d0
0
1
y1 x1 x2 x3 x4
n17
1da
1d7
1d9
1c6
1d8
1cd
1
0
Boolean Satisfiability Problem
? Problem definition . = 1.
? One of the most famous NP-complete problems
? is usually given as CNF (Conjunctive Normal Form) a.k.a. POS
? Variable ? Literal ? Clause ? CNF
1, 2, ... 1, 1 , 2, 2 , ... 1 + 2 + 3, ... ...
? Many heuristics have been proposed
? Can be solved much faster than the other NP-complete problems
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related download
- exploring logic optimizations with reinforcement learning and
- a simulation guided paradigm for logic synthesis and verification
- introduction to formal verification and logic synthesis
- why logic synthesis how does it work why not logic synthesis
- l02 logic synthesis 2019 mit
- logic synthesis meets machine learning trading exactness for
Related searches
- introduction to logic pdf
- introduction to logic book
- introduction to logic gensler pdf
- introduction to java programming and data structures
- concise introduction to logic pdf
- introduction to logic in mathematics
- introduction to mathematical logic pdf
- mathematical introduction to logic pdf
- hurley introduction to logic pdf
- an introduction to formal logic
- introduction to logic pdf gensler
- introduction to leadership concepts and practice