Using Unreachable State Information in Sequential Logic Synthesis

Using Unreachable State Information in Sequential Logic Synthesis Robert Brayton UC Berkeley

Windowing a Node in the Network Ø Definition l Ø A window for a node in the network is the context in which the don’t-cares are computed A window includes n levels of the TFI l m levels of the TFO l all re-convergent paths captured in this scope Ø Window with its PIs and POs can be considered as a separate network Boolean network Window POs m=3 l n=3 Window PIs

Don’t-Care Computation Framework “Miter” constructed for the window POs … f f x x s s Window Same window with inverter

Implementation of Don’t-Care Computation Compute the care set l • Simulate the miter using random patterns • Collect x minterms, for which the output of miter is 1 • This is a subset of a care set l 1 Simulation n Satisfiability • Derive set of network clauses • Add the negation of the current care set, • Assert the output of miter to be 1, • Enumerate through the SAT assignments • Add these assignments to the care set s x n s x

Resubstitution considers a node in a Boolean network and expresses it using a different set of fanins X X Computation can be enhanced by use of don’t cares

Resubstitution with Don’t-Cares Overview Consider all or some nodes in Boolean network Ø Create window Ø Select possible fanin nodes (divisors) Ø For each candidate subset of divisors l l l Rule out some subsets using simulation Check resubstitution feasibility using SAT Compute resubstitution function using interpolation • A low-cost by-product of completed SAT proofs Ø Update the network if there is an improvement

Resubstitution with Don’t Cares Ø Given: l l l Ø C(x) Find: l Ø node function F(x) to be replaced care set C(x) for the node candidate set of divisors {gi(x)} for reexpressing F(x) A resubstitution function h(y) such that F(x) = h(g(x)) on the care set SPFD Theorem: Function h exists if and only if every pair of care minterms, x 1 and x 2, distinguished by F(x), is also distinguished by gi(x) for some i g 1 g 2 g 3 = F(x) C(x) F(x) h(g) g 1 g 2 g 3

Example of Resubstitution Ø Any minterm pair distinguished by F(x) should also be distinguished by at least one of the candidates gi(x) Given: F(x) = (x 1 x 2)(x 2 x 3) Two candidate sets: {g 1= x 1’x 2, g 2 = x 1 x 2’x 3}, {g 3= x 1 x 2, g 4 = x 2 x 3} Set {g 3, g 4} cannot be used for resubstitution while set {g 1, g 2} can. x F(x) g 1(x) g 2(x) g 3(x) g 4(x) 000 0 0 001 0 0 010 1 1 0 011 1 1 0 1 1 100 0 1 0 101 1 0 110 0 1 0 111 0 0 0 1 1

Checking Resubstitution using SAT Miter for resubstitution check F F Note use of care set. Resubstitution function exists if and only if SAT problem is unsatisfiable.

Computing Dependency Function h by Interpolation (Theory) Consider two sets of clauses, A(x, y) and B(y, z), such that A(x, y) B(y, z) = 0 l y are the only variables common to A and B. Ø An interpolant of the pair (A(x, y), B(y, z)) is a function h(y) depending only on the common variables y such that A(x, y) h(y) B(y, z) Ø Boolean space (x, y, z) h(y) A(x, y) B(y, z)

Computing Dependency Function h by Interpolation (Implementation) Problem: Ø Find function h(y), such that C(x) [h(g(x)) F(x)], i. e. F(x) is expressed in terms of {gi}. Solution: Ø Prove the corresponding SAT problem “unsatisfiable” Ø Derive unsatisfiability resolution proof [Goldberg/Novikov, DATE’ 03] Ø Divide clauses into A clauses and B clauses Ø Derive interpolant from the unsatisfiability proof [Mc. Millan, CAV’ 03] Ø Use interpolant as the dependency function, h(g) Ø Replace F(x) by h(g) if cost function improved A h B Notes on this solution Ø uses don’t cares Ø does not use Espresso Ø is more scalable A B y

Don’t Care Based Simplification Using Care Clauses Representing Reached States We compute an inductive invariant consisting of a set of care clauses on k-cuts of an AIG representing the sequential circuit. (see indcut paper submitted to CAV’ 08) Ø In order to use these as well as SDCs and ODCs in a scalable way, windowing is used Ø l l l An outer window is used to compute cares for inner window We look for those clauses whose variables all fall within the outer window We add these to the outer window in computing the care set in terms of variables x

Care Clauses in Outer Window z care clause 1 1 care clause f 1 y inner window x Outer window s Entire circuit care clause 1

Care Computation Using Care Clauses Adding care clauses restricts what values s can have … z 2 z 1 Outer window W 1 care clause Outer window W 2 care clause f f x x s s 1 Window 1 Same window with inverter