This paper presents a database containing 'ground truth' segmentations produced by humans for images of a wide variety of natural scenes. We define an error measure which quantifies the consistency between segmentations of differing granularities and find that different human segmentations of the same image are highly consistent. Use of this dataset is demonstrated in two applications: (1) evaluating the performance of segmentation algorithms and (2) measuring probability distributions associated with Gestalt grouping factors as well as statistics of image region properties.

/pdf/a-database-of-human-segmented-natural-images-and-its-2ve1vk356s.pdf

A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics

https://deim.urv.cat/~alexandre.arenas/publicacions/pdf/review.pdf

Synchronization in complex networks

This survey covers rollback-recovery techniques that do not require special language constructs. In the first part of the survey we classify rollback-recovery protocols into checkpoint-based and log-based.Checkpoint-based protocols rely solely on checkpointing for system state restoration. Checkpointing can be coordinated, uncoordinated, or communication-induced. Log-based protocols combine checkpointing with logging of nondeterministic events, encoded in tuples called determinants. Depending on how determinants are logged, log-based protocols can be pessimistic, optimistic, or causal. Throughout the survey, we highlight the research issues that are at the core of rollback-recovery and present the solutions that currently address them. We also compare the performance of different rollback-recovery protocols with respect to a series of desirable properties and discuss the issues that arise in the practical implementations of these protocols.

/pdf/a-survey-of-rollback-recovery-protocols-in-message-passing-bnrpate93a.pdf

A survey of rollback-recovery protocols in message-passing systems

Parallel discrete event simulation (PDES), sometimes called distributed simulation, refers to the execution of a single discrete event simulation program on a parallel computer. PDES has attracted a considerable amount of interest in recent years. From a pragmatic standpoint, this interest arises from the fact that large simulations in engineering, computer science, economics, and military applications, to mention a few, consume enormous amounts of time on sequential machines. From an academic point of view, parallel simulation is interesting because it represents a problem domain that often contains substantial amounts of parallelism (e.g., see [59]), yet paradoxically, is surprisingly difficult to parallelize in practice. A sufficiently general solution to the PDES problem may lead to new insights in parallel computation as a whole. Historically, the irregular, data-dependent nature of PDES programs has identified it as an application where vectorization techniques using supercomputer hardware provide little benefit [14].A discrete event simulation model assumes the system being simulated only changes state at discrete points in simulated time. The simulation model jumps from one state to another upon the occurrence of an event. For example, a simulator of a store-and-forward communication network might include state variables to indicate the length of message queues, the status of communication links (busy or idle), etc. Typical events might include arrival of a message at some node in the network, forwarding a message to another network node, component failures, etc.We are especially concerned with the simulation of asynchronous systems where events are not synchronized by a global clock, but rather, occur at irregular time intervals. For these systems, few simulator events occur at any single point in simulated time; therefore parallelization techniques based on lock-step execution using a global simulation clock perform poorly or require assumptions in the timing model that may compromise the fidelity of the simulation. Concurrent execution of events at different points in simulated time is required, but as we shall soon see, this introduces interesting synchronization problems that are at the heart of the PDES problem.This article deals with the execution of a simulation program on a parallel computer by decomposing the simulation application into a set of concurrently executing processes. For completeness, we conclude this section by mentioning other approaches to exploiting parallelism in simulation problems.Comfort and Shepard et al. have proposed using dedicated functional units to implement specific sequential simulation functions, (e.g., event list manipulation and random number generation [20, 23, 47]). This method can provide only a limited amount of speedup, however. Zhang, Zeigler, and Concepcion use the hierarchical decomposition of the simulation model to allow an event consisting of several subevents to be processed concurrently [21, 98]. A third alternative is to execute independent, sequential simulation programs on different processors [11, 39]. This replicated trials approach is useful if the simulation is largely stochastic and one is performing long simulation runs to reduce variance, or if one is attempting to simulate a specific simulation problem across a large number of different parameter settings. However, one drawback with this approach is that each processor must contain sufficient memory to hold the entire simulation. Furthermore, this approach is less suitable in a design environment where results of one experiment are used to determine the experiment that should be performed next because one must wait for a sequential execution to be completed before results are obtained.

/pdf/parallel-discrete-event-simulation-m759whzicd.pdf

Parallel discrete event simulation

A number of library based parallel and sequential network simulators have been designed. The paper describes a library, called GloMoSim (Global Mobile system Simulator), for parallel simulation of wireless networks. GloMoSim has been designed to be extensible and composable: the communication protocol stack for wireless networks is divided into a set of layers, each with its own API. Models of protocols at one layer interact with those at a lower (or higher) layer only via these APIs. The modular implementation enables consistent comparison of multiple protocols at a given layer. The parallel implementation of GloMoSim can be executed using a variety of conservative synchronization protocols, which include the null message and conditional event algorithms. The paper describes the GloMoSim library, addresses a number of issues relevant to its parallelization, and presents a set of experimental results on the IBM 9076 SP, a distributed memory multicomputer. These experiments use models constructed from the library modules.

GloMoSim: a library for parallel simulation of large-scale wireless networks

State-saving and reverse computation are two different approaches by which rollback is realized in Time Warp-based parallel simulation systems. Of the two approaches, state-saving is, in general, more memory-intensive than reverse computation. When executed on a state-of-the-art commercial CC-NUMA (Cache Coherent Non-Uniform Memory Architecture) multiprocessor, our Time Warp system runs almost 6 times slower if state-saving is used than if reverse computation is used. The focus of this paper is to understand why state-saving yields such poor performance when compared to reverse computation on a CC-NUMA multiprocessor. To address this question, we examined the low level machine performance statistics, especially those that relate to memory system performance, such as caching, and translation look-aside buffer (TLB) misses. The outcome of the performance study suggests that TLB misses are the primary culprit for state-saving's performance degradation.

/pdf/the-effect-of-state-saving-in-optimistic-simulation-on-a-3cqlxhd21m.pdf

The effect of state-saving in optimistic simulation on a cache-coherent non-uniform memory access architecture

Real-time distributed simulations, such as on-line gaming or military training simulations are normally considered to be non-deterministic. Analysis of these simulations is therefore difficult depending solely on logging and runtime observations. This paper explores an approach for removing one major source of non-determinism in these simulations, thereby allowing repeatable executions. Specifically, we use a synchronization protocol to ensure repeatable delivery of messages. Through limited instrumentation of the simulation code, we maintain a virtual time clock, by which message delivery is governed. The additional overhead imposed by the scheme is shown to be reasonable, although additional reductions in this overhead are anticipated. The results are demonstrated in the context of a simple combat model, whose only source of non-determinism is communications latency. The simulation is shown to be made repeatable, and the perturbation on the execution compared to the non-repeatable execution small. The paper is one step in bridging the gap between the traditional PDES perspective and real-time simulation world.

Repeatability in real-time distributed simulation executions

In this paper we discuss new synchronization algorithms for Parallel and Distributed Discrete Event Simulations (PDES) which exploit the capabilities and behavior of the underlying communications network. Previous work in this areas has assumed the network to be a Black Box which provides a one-to-one, reliable and in-order message passing paradigm. In our work, we utilize the Broadcast capability of the ubiquitous Ethernet for synchronization computations, and both unreliable and reliable protocols for message passing, to achieve more efficient communications between the participating systems.We describe two new algorithms for computation of a distributed snapshot of global reduction operations on monotonically increasing values. The algorithms require O(N) messages (where N is the number of systems participating in the snapshot) in the normal case. We specifically target the use of this algorithm for distributed discrete event simulations to determine a global lower bound on time-stamp (LBTS), but expect the algorithm has applicability outside the simulation community.

Network aware time management and event distribution

The behavior of n interacting processes synchronized by the “Time Warp” rollback mechanism is analyzed under the constraint that the total amount of memory to execute the program is limited. In Time Warp, a protocol called “cancelback” has been proposed to reclaim storage when the system runs out of memory. A discrete state, continuous time Markov chain model for Time Warp augmented with the cancelback protocol is developed for a shared memory system with n homogeneous processors and homogeneous workload. The model allows one to predict speedup as the amount of available memory is varied. To our knowledge, this is the first model to achieve this result. The performance predicted by the model is validated through direct performance measurements on an operational Time Warp system executing on a shared-memory multiprocessor using a workload similar to that in the model. It is observed that Time Warp with only a few additional message buffers per processor over that required in the corresponding sequential execution can achieve approximately the same or even greater performance than Time Warp with unlimited memory, if GVT computation and fossil collection can be efficiently implemented.

Performance analysis of “Time Warp” with limited memory

The suitability of the Time Warp mechanism to perform simulations with real-time constraints is examined. A model for Time Warp is developed that accounts for overheads such as state saving, state restoration, and sending and transmitting positive and negative messages. A criterion called R-schedulability is defined to indicate whether or not computations can meet real-time deadlines. It is shown that if false events (events that will be rolled back or cancelled later) are generated, and there are no committed events with timestamps equal to those of the false events, Time Warp cannot meet the R-schedulability criterion. Further, if aggressive cancellation is used, scheduling guarantees still cannot be made even in the absence of such false events. However, Time Warp using lazy cancellation is shown to be R-schedulable provided such false events do not exist. Finally, based on these results, bounds on the execution time of a Time Warp simulation are derived.

Richard M. Fujimoto

Papers

The effect of state-saving in optimistic simulation on a cache-coherent non-uniform memory access architecture

Repeatability in real-time distributed simulation executions

Network aware time management and event distribution

Performance analysis of “Time Warp” with limited memory

Time Warp simulation in time constrained systems