arXiv subjects

Ganesh Gopalakrishnan

Publications and source records attributed to Ganesh Gopalakrishnan.

31 records · Page 2Linked to original sources

arXiv2023

Efficient Linearizability Checking for Actor-based Systems

Recent demand for distributed software had led to a surge in popularity in actor-based frameworks. However, even with the stylized message passing model of actors, writing correct distributed software is still difficult. We present our work on linearizability checking in DS2, an integrated framework for specifying, synthesizing, and testing distributed actor systems. The key insight of our approach is that often subcomponents of distributed actor systems represent common algorithms or data structures (e.g.\ a distributed hash table or tree) that can be validated against a simple sequential model of the system. This makes it easy for developers to validate their concurrent actor systems without complex specifications. DS2 automatically explores the concurrent schedules that system could arrive at, and it compares observed output of the system to ensure it is equivalent to what the sequential implementation could have produced. We describe DS2's linearizability checking and test it on several concurrent replication algorithms from the literature. We explore in detail how different algorithms for enumerating the model schedule space fare in finding bugs in actor systems, and we present our own refinements on algorithms for exploring actor system schedules that we show are effective in finding bugs.

cs.DC↗

arXiv2023

Understanding the Effect of the Long Tail on Neural Network Compression

Network compression is now a mature sub-field of neural network research: over the last decade, significant progress has been made towards reducing the size of models and speeding up inference, while maintaining the classification accuracy. However, many works have observed that focusing on just the overall accuracy can be misguided. E.g., it has been shown that mismatches between the full and compressed models can be biased towards under-represented classes. This raises the important research question, can we achieve network compression while maintaining "semantic equivalence" with the original network? In this work, we study this question in the context of the "long tail" phenomenon in computer vision datasets observed by Feldman, et al. They argue that memorization of certain inputs (appropriately defined) is essential to achieving good generalization. As compression limits the capacity of a network (and hence also its ability to memorize), we study the question: are mismatches between the full and compressed models correlated with the memorized training data? We present positive evidence in this direction for image classification tasks, by considering different base architectures and compression schemes.

cs.LG↗

arXiv2021

Going Beyond Classification Accuracy Metrics in Model Compression

With the rise in edge-computing devices, there has been an increasing demand to deploy energy and resource-efficient models. A large body of research has been devoted to developing methods that can reduce the size of the model considerably without affecting the standard metrics such as top-1 accuracy. However, these pruning approaches tend to result in a significant mismatch in other metrics such as fairness across classes and explainability. To combat such misalignment, we propose a novel multi-part loss function inspired by the knowledge-distillation literature. Through extensive experiments, we demonstrate the effectiveness of our approach across different compression algorithms, architectures, tasks as well as datasets. In particular, we obtain up to $4.1\times$ reduction in the number of prediction mismatches between the compressed and reference models, and up to $5.7\times$ in cases where the reference model makes the correct prediction; all while making no changes to the compression algorithm, and minor modifications to the loss function. Furthermore, we demonstrate how inducing simple alignment between the predictions of the models naturally improves the alignment on other metrics including fairness and attributions. Our framework can thus serve as a simple plug-and-play component for compression algorithms in the future.

cs.CV↗

arXiv2021

Automated Dynamic Concurrency Analysis for Go

The concurrency features of the Go language have proven versatile in the development of a number of concurrency systems. However, correctness methods to address challenges in Go concurrency debugging have not received much attention. In this work, we present an automatic dynamic tracing mechanism that efficiently captures and helps analyze the whole-program concurrency model. Using an enhancement to the built-in tracer package of Go and a framework that collects dynamic traces from application execution, we enable thorough post-mortem analysis for concurrency debugging. Preliminary results about the effectiveness and scalability (up to more than 2K goroutines) of our proposed dynamic tracing for concurrent debugging are presented. We discuss the future direction for exploiting dynamic tracing towards accelerating concurrent bug exposure.

cs.DC↗

arXiv2020

A Programmable Approach to Neural Network Compression

Deep neural networks (DNNs) frequently contain far more weights, represented at a higher precision, than are required for the specific task which they are trained to perform. Consequently, they can often be compressed using techniques such as weight pruning and quantization that reduce both the model size and inference time without appreciable loss in accuracy. However, finding the best compression strategy and corresponding target sparsity for a given DNN, hardware platform, and optimization objective currently requires expensive, frequently manual, trial-and-error experimentation. In this paper, we introduce a programmable system for model compression called Condensa. Users programmatically compose simple operators, in Python, to build more complex and practically interesting compression strategies. Given a strategy and user-provided objective (such as minimization of running time), Condensa uses a novel Bayesian optimization-based algorithm to automatically infer desirable sparsities. Our experiments on four real-world DNNs demonstrate memory footprint and hardware runtime throughput improvements of 188x and 2.59x, respectively, using at most ten samples per search. We have released a reference implementation of Condensa at https://github.com/NVlabs/condensa.

cs.LG↗

arXiv2020

FPDetect: Efficient Reasoning About Stencil Programs Using Selective Direct Evaluation

We present FPDetect, a low overhead approach for detecting logical errors and soft errors affecting stencil computations without generating false positives. We develop an offline analysis that tightly estimates the number of floating-point bits preserved across stencil applications. This estimate rigorously bounds the values expected in the data space of the computation. Violations of this bound can be attributed with certainty to errors. FPDetect helps synthesize error detectors customized for user-specified levels of accuracy and coverage. FPDetect also enables overhead reduction techniques based on deploying these detectors coarsely in space and time. Experimental evaluations demonstrate the practicality of our approach.

cs.DC↗

arXiv2020

An Abstraction-guided Approach to Scalable and Rigorous Floating-Point Error Analysis

Automated techniques for rigorous floating-point round-off error analysis are important in areas including formal verification of correctness and precision tuning. Existing tools and techniques, while providing tight bounds, fail to analyze expressions with more than a few hundred operators, thus unable to cover important practical problems. In this work, we present Satire, a new tool that sheds light on how scalability and bound-tightness can be attained through a combination of incremental analysis, abstraction, and judicious use of concrete and symbolic evaluation. Satire has handled problems exceeding 200K operators. We present Satire's underlying error analysis approach, information-theoretic abstraction heuristics, and a wide range of case studies, with evaluation covering FFT, Lorenz system of equations, and various PDE stencil types. Our results demonstrate the tightness of Satire's bounds, its acceptable runtime, and valuable insights provided.

cs.PL↗

arXiv2019

Message Scheduling for Performant, Many-Core Belief Propagation

Belief Propagation (BP) is a message-passing algorithm for approximate inference over Probabilistic Graphical Models (PGMs), finding many applications such as computer vision, error-correcting codes, and protein-folding. While general, the convergence and speed of the algorithm has limited its practical use on difficult inference problems. As an algorithm that is highly amenable to parallelization, many-core Graphical Processing Units (GPUs) could significantly improve BP performance. Improving BP through many-core systems is non-trivial: the scheduling of messages in the algorithm strongly affects performance. We present a study of message scheduling for BP on GPUs. We demonstrate that BP exhibits a tradeoff between speed and convergence based on parallelism and show that existing message schedulings are not able to utilize this tradeoff. To this end, we present a novel randomized message scheduling approach, Randomized BP (RnBP), which outperforms existing methods on the GPU.

cs.DC↗

arXiv2019

Multi-level analysis of compiler induced variability and performance tradeoffs

Successful HPC software applications are long-lived. When ported across machines and their compilers, these applications often produce different numerical results, many of which are unacceptable. Such variability is also a concern while optimizing the code more aggressively to gain performance. Efficient tools that help locate the program units (files and functions) within which most of the variability occurs are badly needed, both to plan for code ports and to root-cause errors due to variability when they happen in the field. In this work, we offer an enhanced version of the open-source testing framework FLiT to serve these roles. Key new features of FLiT include a suite of bisection algorithms that help locate the root causes of variability. Another added feature allows an analysis of the tradeoffs between performance and the degree of variability. Our new contributions also include a collection of case studies. Results on the MFEM finite-element library include variability/performance tradeoffs, and the identification of a (hitherto unknown) abnormal level of result-variability even under mild compiler optimizations. Results from studying the Laghos proxy application include identifying a significantly divergent floating-point result-variability and successful root-causing down to the problematic function over as little as 14 program executions. Finally, in an evaluation of 4,376 controlled injections of floating-point perturbations on the LULESH proxy application, we showed that the FLiT framework has 100 precision and recall in discovering the file and function locations of the injections all within an average of only 15 program executions.

cs.SE↗

arXiv2018

Sparse Matrix Code Dependence Analysis Simplification at Compile Time

Analyzing array-based computations to determine data dependences is useful for many applications including automatic parallelization, race detection, computation and communication overlap, verification, and shape analysis. For sparse matrix codes, array data dependence analysis is made more difficult by the use of index arrays that make it possible to store only the nonzero entries of the matrix (e.g., in A[B[i]], B is an index array). Here, dependence analysis is often stymied by such indirect array accesses due to the values of the index array not being available at compile time. Consequently, many dependences cannot be proven unsatisfiable or determined until runtime. Nonetheless, index arrays in sparse matrix codes often have properties such as monotonicity of index array elements that can be exploited to reduce the amount of runtime analysis needed. In this paper, we contribute a formulation of array data dependence analysis that includes encoding index array properties as universally quantified constraints. This makes it possible to leverage existing SMT solvers to determine whether such dependences are unsatisfiable and significantly reduces the number of dependences that require runtime analysis in a set of eight sparse matrix kernels. Another contribution is an algorithm for simplifying the remaining satisfiable data dependences by discovering equalities and/or subset relationships. These simplifications are essential to make a runtime-inspection-based approach feasible.

cs.PL↗

arXiv2017

An Operational Semantic Basis for OpenMP Race Analysis

OpenMP is the de facto standard to exploit the on-node parallelism in new generation supercomputers.Despite its overall ease of use, even expert users are known to create OpenMP programs that harbor concurrency errors, of which one of the most insidious of errors are {\em data races}.OpenMP is also a rapidly evolving standard, which means that future data races may be introduced within unfamiliar contexts.A simple and rigorous operational semantics for OpenMP can help build reliable race checkers and ward off future errors through programmer education and better tooling.This paper's key contribution is a simple operational semantics for OpenMP, with primitive events matching those generated by today's popular OpenMP runtimes and tracing methods such as OMPT.This makes our operational semantics more than a theoretical document for intellectual edification; it can serve as a blueprint for OpenMP event capture and tool building.We back this statement by summarizing the workings of a new data race checker for OpenMP being built based on this semantics.The larger purpose served by our semantics is to serve the needs of the OpenMP community with regard to their contemplated extensions to OpenMP, as well as future tooling efforts.

cs.DC↗

arXiv2017

Report of the HPC Correctness Summit, Jan 25--26, 2017, Washington, DC

Maintaining leadership in HPC requires the ability to support simulations at large scales and fidelity. In this study, we detail one of the most significant productivity challenges in achieving this goal, namely the increasing proclivity to bugs, especially in the face of growing hardware and software heterogeneity and sheer system scale. We identify key areas where timely new research must be proactively begun to address these challenges, and create new correctness tools that must ideally play a significant role even while ramping up toward exacale. We close with the proposal for a two-day workshop in which the problems identified in this report can be more broadly discussed, and specific plans to launch these new research thrusts identified.

cs.DC↗

arXiv2016

PRESAGE: Protecting Structured Address Generation against Soft Errors

Modern computer scaling trends in pursuit of larger component counts and power efficiency have, unfortunately, lead to less reliable hardware and consequently soft errors escaping into application data ("silent data corruptions"). Techniques to enhance system resilience hinge on the availability of efficient error detectors that have high detection rates, low false positive rates, and lower computational overhead. Unfortunately, efficient detectors to detect faults during address generation (to index large arrays) have not been widely researched. We present a novel lightweight compiler-driven technique called PRESAGE for detecting bit-flips affecting structured address computations. A key insight underlying PRESAGE is that any address computation scheme that flows an already incurred error is better than a scheme that corrupts one particular array access but otherwise (falsely) appears to compute perfectly. Enabling the flow of errors allows one to situate detectors at loop exit points, and helps turn silent corruptions into easily detectable error situations. Our experiments using PolyBench benchmark suite indicate that PRESAGE-based error detectors have a high error-detection rate while incurring low overheads.

cs.SE↗