arXiv subjects

Keshav Pingali

Publications and source records attributed to Keshav Pingali.

27 records · Page 2Linked to original sources

arXiv2020

Pangolin: An Efficient and Flexible Graph Pattern Mining System on CPU and GPU

There is growing interest in graph pattern mining (GPM) problems such as motif counting. GPM systems have been developed to provide unified interfaces for programming algorithms for these problems and for running them on parallel systems. However, existing systems may take hours to mine even simple patterns in moderate-sized graphs, which significantly limits their real-world usability. We present Pangolin, a high-performance and flexible in-memory GPM framework targeting shared-memory CPUs and GPUs. Pangolin is the first GPM system that provides high-level abstractions for GPU processing. It provides a simple programming interface based on the extend-reduce-filter model, which enables users to specify application-specific knowledge for search space pruning and isomorphism test elimination. We describe novel optimizations that exploit locality, reduce memory consumption, and mitigate the overheads of dynamic memory allocation and synchronization. Evaluation on a 28-core CPU demonstrates that Pangolin outperforms existing GPM frameworks Arabesque, RStream, and Fractal by 49x, 88x, and 80x on average, respectively. Acceleration on a V100 GPU further improves performance of Pangolin by 15x on average. Compared to state-of-the-art hand-optimized GPM applications, Pangolin provides competitive performance with less programming effort.

cs.DC↗

arXiv2019

An Elementary Introduction to Kalman Filtering

Kalman filtering is a classic state estimation technique used in application areas such as signal processing and autonomous control of vehicles. It is now being used to solve problems in computer systems such as controlling the voltage and frequency of processors. Although there are many presentations of Kalman filtering in the literature, they usually deal with particular systems like autonomous robots or linear systems with Gaussian noise, which makes it difficult to understand the general principles behind Kalman filtering. In this paper, we first present the abstract ideas behind Kalman filtering at a level accessible to anyone with a basic knowledge of probability theory and calculus, and then show how these concepts can be applied to the particular problem of state estimation in linear systems. This separation of concepts from applications should make it easier to understand Kalman filtering and to apply it to other problems in computer systems.

eess.SY↗

arXiv2019

SLAMBooster: An Application-aware Controller for Approximation in SLAM

Simultaneous Localization and Mapping (SLAM) is the problem of constructing a map of a mobile agent's environment while localizing the agent within the map. Dense SLAM algorithms perform reconstruction and localization at pixel granularity. These algorithms require a lot of computational power, which has hindered their use on low-power resource-constrained devices. Approximate computing can be used to speed up SLAM implementations as long as the approximations do not prevent the agent from navigating correctly through the environment. Previous studies of approximation in SLAM have assumed that the entire trajectory of the agent is known before the agent starts, and they have focused on offline controllers that set approximation knobs at the start of the trajectory. In practice, the trajectory is not known ahead of time, and allowing knob settings to change dynamically opens up more opportunities for reducing computation time and energy. We describe SLAMBooster, an application-aware, online control system for dense SLAM that adaptively controls approximation knobs during the motion of the agent. SLAMBooster is based on a control technique called proportional-integral-derivative (PID) controller but our experiments showed this application-agnostic controller led to an unacceptable reduction in localization accuracy. To address this problem, SLAMBooster also exploits domain knowledge for controlling approximation by performing smooth surface detection and pose correction. We implemented SLAMBooster in the open-source SLAMBench framework and evaluated it on several trajectories. Our experiments show that on the average, SLAMBooster reduces the computation time by 72% and energy consumption by 35% on an embedded platform, while maintaining the accuracy of localization within reasonable bounds. These improvements make it feasible to deploy SLAM on a wider range of devices.

cs.RO↗

arXiv2017

Dynamic Load Balancing Strategies for Graph Applications on GPUs

Acceleration of graph applications on GPUs has found large interest due to the ubiquitous use of graph processing in various domains. The inherent \textit{irregularity} in graph applications leads to several challenges for parallelization. A key challenge, which we address in this paper, is that of load-imbalance. If the work-assignment to threads uses node-based graph partitioning, it can result in skewed task-distribution, leading to poor load-balance. In contrast, if the work-assignment uses edge-based graph partitioning, the load-balancing is better, but the memory requirement is relatively higher. This makes it unsuitable for large graphs. In this work, we propose three techniques for improved load-balancing of graph applications on GPUs. Each technique brings in unique advantages, and a user may have to employ a specific technique based on the requirement. Using Breadth First Search and Single Source Shortest Paths as our processing kernels, we illustrate the effectiveness of each of the proposed techniques in comparison to the existing node-based and edge-based mechanisms.

cs.DC↗

arXiv2017

Capri: A Control System for Approximate Programs

Approximate computing trades off accuracy of results for resources such as energy or computing time. There is a large and rapidly growing literature on approximate computing that has focused mostly on showing the benefits of approximation. However, we know relatively little about how to control approximation in a disciplined way. This document briefly describes our published work of controlling approximation for non-streaming programs that have a set of "knobs" that can be dialed up or down to control the level of approximation of different components in the program. The proposed system, Capri, solves this control problem as a constrained optimization problem. Capri uses machine learning to learn cost and error models for the program, and uses these models to determine, for a desired level of approximation, knob settings that optimize metrics such as running time or energy usage. Experimental results with complex benchmarks from different problem domains demonstrate the effectiveness of this approach. This report outlines improvements and extensions to the existing Capri system to address its limitations, including a complete rewrite of the software, and discusses directions for follow up work. The document also includes instructions and guidelines for using the new Capri infrastructure.

cs.PL↗

arXiv2016

Adaptive Work-Efficient Connected Components on the GPU

This report presents an adaptive work-efficient approach for implementing the Connected Components algorithm on GPUs. The results show a considerable increase in performance (up to 6.8$\times$) over current state-of-the-art solutions.

cs.DC↗

arXiv2016

Lowering IrGL to CUDA

The IrGL intermediate representation is an explicitly parallel representation for irregular programs that targets GPUs. In this report, we describe IrGL constructs, examples of their use and how IrGL is compiled to CUDA by the Galois GPU compiler.

cs.PL↗

arXiv2012

Processor Allocation for Optimistic Parallelization of Irregular Programs

Optimistic parallelization is a promising approach for the parallelization of irregular algorithms: potentially interfering tasks are launched dynamically, and the runtime system detects conflicts between concurrent activities, aborting and rolling back conflicting tasks. However, parallelism in irregular algorithms is very complex. In a regular algorithm like dense matrix multiplication, the amount of parallelism can usually be expressed as a function of the problem size, so it is reasonably straightforward to determine how many processors should be allocated to execute a regular algorithm of a certain size (this is called the processor allocation problem). In contrast, parallelism in irregular algorithms can be a function of input parameters, and the amount of parallelism can vary dramatically during the execution of the irregular algorithm. Therefore, the processor allocation problem for irregular algorithms is very difficult. In this paper, we describe the first systematic strategy for addressing this problem. Our approach is based on a construct called the conflict graph, which (i) provides insight into the amount of parallelism that can be extracted from an irregular algorithm, and (ii) can be used to address the processor allocation problem for irregular algorithms. We show that this problem is related to a generalization of the unfriendly seating problem and, by extending Turán's theorem, we obtain a worst-case class of problems for optimistic parallelization, which we use to derive a lower bound on the exploitable parallelism. Finally, using some theoretically derived properties and some experimental facts, we design a quick and stable control strategy for solving the processor allocation problem heuristically.

cs.PL↗

arXiv2000

Fractal Symbolic Analysis

Restructuring compilers use dependence analysis to prove that the meaning of a program is not changed by a transformation. A well-known limitation of dependence analysis is that it examines only the memory locations read and written by a statement, and does not assume any particular interpretation for the operations in that statement. Exploiting the semantics of these operations enables a wider set of transformations to be used, and is critical for optimizing important codes such as LU factorization with pivoting. Symbolic execution of programs enables the exploitation of such semantic properties, but it is intractable for all but the simplest programs. In this paper, we propose a new form of symbolic analysis for use in restructuring compilers. Fractal symbolic analysis compares a program and its transformed version by repeatedly simplifying these programs until symbolic analysis becomes tractable, ensuring that equality of simplified programs is sufficient to guarantee equality of the original programs. We present a prototype implementation of fractal symbolic analysis, and show how it can be used to optimize the cache performance of LU factorization with pivoting.

cs.PL↗