Event manipulation for discrete simulations requiring large numbers of events

Abstract

The event-manipulation system presented here consists of two major parts. The first part addresses the familiar problem of event scheduling efflciency when the number of scheduled events grows large. The second part deals with the less apparent problem of providing efficiency and flexibility as scheduled events are accessed to be executed. Additional features and problems dealt with include the proper handling of simultaneous events; that certain events must be created, scheduled, and executed at the same points in simulated time; that inflnite loops caused by the concatenation of such "zero-time" events are possible and must be diagnosed; that maintaining various event counts is practical and economical; and that a capability for handling "tune-displaceable" events is desirable and possible.The scheduling algorithm described here is an improvement over a previous version [9] and combines the standard linear list [1. 3, 4] scheduling method with the more recent time-mapping [7, 8. 6] method. The basic strategy used is break up the (potentially) very long standard-event list into many short lists [7. 6, 10. 11. 2] for which event scheduling can be done more efficiently. The scheduling algonthm uses a "converging lists" data structure, with lists originating in a time-mapping table and converging into a single list. This structure facilitates time-mapping scheduling without the necessity to separate "imminent" and "remote" events, and also permits adjustments of the simulation tune scale whenever the density of imminent events is consistently high or low.The scheduling algorithm deals with the problem of simultaneous events (5]. and actually gains in overall efficiency as the percentage of simultaneous events increases. Simultaneous events are scheduled into "vertical" sublists suspended from normal "horizontal" event lists.A crucial and integral part ofthe event-manipulation system is an "interpretive-branching" mechanism used for event accessing. This mechanism is based on event codes routinely injected into "object" events and on permanent codes located within permanent simulationcontrol events. This mechanism is flexible and fast because it permits recognition and accessing of control events and object events (distinguished from each other by type and/or attributes) without sequential testing. Adding to this basic flexibihty, the event interpretation itself may be modifted dynamically by modifying entries within a traverse table used for the interpretation process.Several related algorithms recently published [10, 11. 2] are characterized by a theoretical preoccupation with the pure problem of event scheduling, and do not consider the problems of event accessing, event counting, simultaneousness, and zero-time event loops.Furthermore, even the performance results reported for event scheduling (see Figures 4 and 5 in [2]) are not optimal, apparently because the scheduhng techniques described are burdened with a high per-event overhead. Specifically, the reported overheads of approximately 100 microseconds (usin...

Performance analysis and design of a logic simulation machine

“…Although some evaluations may take more time than others, we assume that all event/function evaluations take the average event/function evaluation time and that this time is invariant with the number of processors. This assumption assumes the existence of a nearconstant-time event-list management capabilities [14].…”

Conjugate symmetry

“…Of the many different simulation algorithms [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33] that have appeared over the years, we have chosen the Inversion Algorithm for comparison with our enhanced version of HyperSim. The Inversion Algorithm is extremely fast, and competes well with Levelized Compiled Code simulation [18].…”

Using conjugate symmetries to enhance gate-level simulations

“…Over the past many years, many different techniques have been explored, and there is an active and continuing interest in simulation problems today. See [1][2][3][4][5][6][7][8][9][10][11][12][13][14] for a small sampling of this work.…”