Timing Analysis. Andreas Kuehlmann. A k = max{a 1 +D k1, A 2 +D k2,a 3 +D k3 } S k. S j. Required times: S ki. given required times on primary outputs

Transcription

1 EECS 9B Spring 3 Timing Analysis - Delay Models Simple model : D k Ak A A A3 Timing Analysis A k = arrival time = max(a,a,a 3 ) + D k D k is the delay at node k, parameterized according to function f k and fanout node k Andreas Kuehlmann Simple model : Ak D k D D k k3 A3 A A A A A3 Can also have different times for rise time and fall time Ak A k = max{a +D k, A +D k,a 3 +D k3 } Static delay analysis Required Times // level of PI nodes initialized to, // the others are set to -. // Invoke LEVEL from PO Algorithm LEVEL(k) { // levelize nodes if( k.level!= -) return(k.level) else k.level = +max{level(k i ) k i fanin(k)} return(k.level) } // Compute arrival times: // Given arrival times on PI s Algorithm ARRIVAL() { for L = to MAXLEVEL for {k k.level = L} A k = MAX{A ki } + D k } Required times: given required times on primary outputs Traverse in reverse topological order (i.e. from primary outputs to primary inputs) if (k i, k ) is an edge between k i and k, R ki,k = R k - D k (this is the edge required time) Hence, the required time of output of node k is R k = min ( R k,kj k j fanout(k) ) k i k S k k i S ki S j j 3

2 Propagating Slacks Sequential networks Slacks: slack at the output node kiss k = R k -A k Since R ki, k =R k -D k S ki, k =R ki, k - A ki S ki, k + A ki = R k -D k = S k + A k - D k Since A k = max {A kj } + D k S ki, k = S k + max {A kj } - A ki k j, k i fanin (k ) S ki = min{s ki, j } j fanout (k i ) k i k S k S ki k i S j j l l C C3 l5 C C l l3 Notes: Each edge is the graph has a slack and a required time Negative slack is bad. Arrival times known at l and l Required times known at l 3, l, and l 5 Delay analysis gives arrival and required times (hence slacks) for C, C, C 3, C 5 6 Static critical paths Min-Max problem:minimize max{-s i, } A static critical path of a Boolean network is a path P = {i,i,,i p } where S ik, ik+ < Note: if a node k is on a static critical path, then at least one of the fanin edges of kis critical. Hence, all critical paths reach from an input to an output. Note: There may be several critical paths Example: Static critical paths A8= R=5 - R= A9= A=6 R=5 A=5 R=5 S =- R3=3 S = R7= S 3, =- R9=- S, = - S, = S 5, = S 6,3 = S 7,3 = - S 7, = - S 7,5 = S 8,6 = S 9,7 = - critical path edges S ki, k = S k + max{a kj } - A ki, k j,k i fanin(k) S k = min{s k, kj }, k j fanout(k) 7 8

3 Timing analysis problems Functional Timing Analysis We want to determine the true critical paths of a circuit in order to: determine the minimum cycle time that the circuit will function identify critical paths from performance optimization - don t want to try to optimize the wrong (non-critical) paths Implications: Don t want false paths (produced by static delay analysis) Delay model is worstcase model. Need to ensure correctness for case where i th gate delay D i M What is Timing Analysis? Estimate when the output of a given circuit gets stable clock Combinational block T 9 Why Timing Analysis? Timing verification Verifies whether a design meets a given timing constraint Example: cycle-time constraint Timing optimization Needs to identify critical portion of a design for further optimization Critical path identification In both applications, the more accurate, the better Timing Analysis - Basics Naïve approach - Simulate all input vectors with SPICE Accurate, but too expensive Gate-level timing analysis Focus of this lecture Less accurate than SPICE due to the level of abstraction, but much more efficient Scenario: Gate/wire delays are pre-characterized (accuracy loss) Perform timing analysis of a gate-level circuit assuming the gate/wire delays 3

4 Gate-level Timing Analysis Example: -bit Carry-skip Adder False path aware arr(z)? z A naive approach is topological analysis Easy longest-path problem Linear in the size of a network Not all paths can propagate signal events False paths If all longest paths are false, topological analysis gives delay overestimate Functional timing analysis= false-pathaware timing analysis Compute false-path-aware arrival time c_in a b a b Length 5 Length s s c_out x x arr(x )= arr(x )= mux ripple carry adder False Path Analysis - Basics Controlling/Non-Controlling Values Is a path responsible for delay? If the answer is no, can ignore the path for delay computation Check the falsity of long paths until we find the longest true path How can we determine whether a path is false? Controlled value of AND Delay underestimation is unacceptable Can lead to overlooking a timing violation Delay overestimation is not desirable, but acceptable Topological analysis can give overestimate, but never give underestimate Controlling value of AND Non-Controlling value of AND Controlled value of OR Controlling value of OR Non-Controlling value of OR 5 6

5 Static Sensitization A path is statically-sensitizable if there exists an input vector such that all the side inputs to the path are set to non-controlling values This is independent of gate delays Controlling value! Static Sensitization The (dashed)path is responsible for delay! Delay underestimation by static sensitization (delay = when true delay = 3) incorrect condition t= t= t= These paths are not statically-sensitizable 3 The longest true path is of length? 7 8 What is Wrong with Static Sensitization? Timing Simulation The idea of forcing non-controlling values to side inputs is okay, buttiming was ignored The same signal can have a controlling value at one time and a non-controlling value at another time. How about timing simulation as a correct method? 3 9 Implies that delay = for these inputs BUT! 5

6 Timing Simulation What is Wrong with Timing Simulation? -> 3 3 If gate delays are reduced, delay estimates can increase Not acceptable since Gate delays are just upper-bounds, actual delay is in [,d] Delay uncertainty due to manufacturing We are implicitly analyzing a familyof circuits where gate delays are within the upper-bounds Implies that delay = with the same set of inputs. Monotone Speedup Property Timing Simulation Revisited Definition: For any circuit C, if C is obtained from C by reducing some gate delays, and delay_estimate(c ) delay_estimate(c), then delay_estimate has Monotone Speedup property Timing simulation does nothave this property 3 3 means that the rising signal occurs anywhere between t = -infinity and t =. X-valued simulation 6

7 Timing Simulation Revisited Timed 3-valued (,,X) simulation called X-valued simulation Monotone speedup property is satisfied. Underlying model of floating mode condition [Chen, Du] Applies to simple gate networks only viability [McGeer, Brayton] Applies to general Boolean networks False Path Analysis Algorithms Checking the falsity of every path explicitly is too expensive - exponential # of paths State-of -the-art approach:. Start: set L = L top -! = topological longest path delay -! L old =. Binary search: If (Delay(L)) ( * )! L = L-L old /, L old = L, L = L +! L Else,! L = L-L old /, L old = L, L = L -! L If (L > L top or! L < threshold), L = L old, done (* ) Delay(L) = if there an input vector under which an output gets stable only at time t where L t? Can be reduced to a SAT problem [McGeer, Saldanha, Brayton, ASV] or a timed-atpg [Devadas, Keutzer, Malik] SAT-based False Path Analysis Example Decision problem: Is there an input vector under which the output gets stable only after t = T? Idea:. characterize the set of all input vectors S(T) that make the output stable no later than t = T. check if S(T) contains S = all possible input vectors This check is solved as a SAT problem: Is S \ S(T) empty? - set difference + emptiness check Let F and F(T) be the characteristic functions of S and S(T) Is F!F(T) satisfiable? a b c e d Assume all the PIs arrive at t =, all gate delays = Is the output stable time t >? f g 7 8 7

8 Example g(,t=) : the set of input vectors under which g gets stable to value = no later than t = a b c e d g(,t=) = d(,t=) f(,t=) = (a(,t=) b(,t=)) (c(,t=) e(,t=)) =!a!b(c ) =!a!bc = S (t=) g(,t= ) = onset =!a!bc = g(,t=) = S 9 f g Onset: stabilized by t=? Example a b c g(,t=) : the set of input vectors under which g gets stable to value = no later than t= e d g(,t=) = d(,t=) f(,t=) = (a(,t=) b(,t=)) (c(,t=) e(,t=)) = (a+b) + (!c ) = a+b = S (t=) g(,t= ) = offset = a+b+!c = S 3 f g Example a b c g(,t=) : the set of input vectors under which g gets stable to no later than t= g(,t=) = a+b e d g(,t= ) = offset = a+b+!c g(,t= ) \ g(,t=) = (a+b+!c)!(a+b) =!a!b!c = satisfiable f g Offset: NOTstabilized by t= under abc= 3 Summary False-path-aware arrival time analysis is wellunderstood Practical algorithms exist Can handle industrial circuits easily Remaining problems Incremental analysis (make it so that a small change in the circuit does not make the analysis start all over) Integration with logic optimization DSM issues such as cross-talk-aware false path analysis 3 8

9 Timed ATPG Yet another way to solve the same decision problem A generalization of regular ATPG: regular ATPG find an input vector that differentiates a fault-freecircuit and a faulty circuit in terms of functionality Timed ATPG find an input vector that exhibits a given timed behavior Timed extension of PODEM 33 9