arXiv subjects

Muktikanta Sa

Publications and source records attributed to Muktikanta Sa.

7 recordsLinked to original sources

arXiv2020

Dynamic Graph Operations: A Consistent Non-blocking Approach

Graph algorithms enormously contribute to the domains such as blockchains, social networks, biological networks, telecommunication networks, and several others. The ever-increasing demand of data-volume, as well as speed of such applications, have essentially transported these applications from their comfort zone: static setting, to a challenging territory of dynamic updates. At the same time, mainstreaming of multi-core processors have entailed that the dynamic applications should be able to exploit concurrency as soon as parallelization gets inhibited. Thus, the design and implementation of efficient concurrent dynamic graph algorithms have become significant. This paper reports a novel library of concurrent shared-memory algorithms for breadth-first search (BFS), single-source shortest-path (SSSP), and betweenness centrality (BC) in a dynamic graph. The presented algorithms are provably non-blocking and linearizable. We extensively evaluate C++ implementations of the algorithms through several micro-benchmarks. The experimental results demonstrate the scalability with the number of threads. Our experiments also highlight the limitations of static graph analytics methods in a dynamic setting.

cs.DC↗

arXiv2019

A Pragmatic Non-Blocking Concurrent Directed Acyclic Graph

In this paper, we have developed two algorithms for maintaining acyclicity in a concurrent directed graph. The first algorithm is based on a wait-free reachability query and the second one is based on partial snapshot-based obstruction-free reachability query. Interestingly, we are able to achieve the acyclic property in the dynamic setting without the need of helping using descriptors by other threads or clean double collect mechanism. We present a proof to show that the graph remains acyclic at all times in the concurrent setting. We also prove that the acyclic graph data-structure operations are linearizable. We implement both the algorithms in C++ and test through a number of micro-benchmarks. Our experimental results show an average of 7x improvement over the sequential and global lock implementation.

cs.DC↗

arXiv2019

A Concurrent Unbounded Wait-Free Graph

In this paper, we propose an efficient concurrent wait-free algorithm to construct an unbounded directed graph for shared memory architecture. To the best of our knowledge that this is the first wait-free algorithm for an unbounded directed graph where insertion and deletion of vertices and/or edges can happen concurrently. To achieve wait-freedom in a dynamic setting, threads help each other to perform the desired tasks using operator descriptors by other threads. To enhance performance, we also developed an optimized wait-free graph based on the principle of fast-path-slow-path. We also prove that all graph operations are wait-free and linearizable. We implemented our algorithms in C++ and tested its performance through several micro-benchmarks. Our experimental results show an average of 9x improvement over the global lock-based implementation.

cs.DC↗

arXiv2019

A Simple and Practical Concurrent Non-blocking Unbounded Graph with Reachability Queries

Graph algorithms applied in many applications, including social networks, communication networks, VLSI design, graphics, and several others, require dynamic modifications -- addition and removal of vertices and/or edges -- in the graph. This paper presents a novel concurrent non-blocking algorithm to implement a dynamic unbounded directed graph in a shared-memory machine. The addition and removal operations of vertices and edges are lock-free. For a finite sized graph, the lookup operations are wait-free. Most significant component of the presented algorithm is the reachability query in a concurrent graph. The reachability queries in our algorithm are obstruction-free and thus impose minimal additional synchronization cost over other operations. We prove that each of the data structure operations are linearizable. We extensively evaluate a sample C/C++ implementation of the algorithm through a number of micro-benchmarks. The experimental results show that the proposed algorithm scales well with the number of threads and on an average provides 5 to 7x performance improvement over a concurrent graph implementation using coarse-grained locking.

cs.DC↗

arXiv2018

Proving Correctness of Concurrent Objects by Validating Linearization Points

Concurrent data structures or CDS such as concurrent stacks, queues, sets etc. have become very popular in the past few years partly due to the rise of multi-core systems. But one of the greatest challenges with CDSs has been developing correct structures and then proving the correctness of these structures. We believe that techniques that help prove the correctness of these CDSs can also guide in developing new CDSs. An intuitive technique to prove the correctness of CDSs is using Linearization Points or LPs. An LP is an atomic event in the execution interval of each method such that the execution of the entire method seems to have taken place in the instant of that event. One of the main challenges with the LP based approach is to identify the correct LPs of a CDS. Identifying the correct LPs can be deceptively wrong in many cases. In fact, in many cases, the LP identified or even worse the CDS itself could be wrong. To address these issues, several automatic tools for verifying linearizability have been developed. But we believe that these tools don't provide insight to a programmer to develop the correct concurrent programs or identify the LPs. Considering the complexity of developing a CDS and verifying its correctness, we address the most basic problem of this domain in this paper: given the set of LPs of a CDS, how to show its correctness? We assume that we are given a CDS and its LPs. We have developed a hand-crafted technique of proving the correctness of the CDS by validating its LPs. As observed earlier, identifying the correct LPs is very tricky and erroneous. But since our technique is hand-crafted, we believe that the process of proving correctness might provide insight to identify the correct LPs, if the currently chosen LP is incorrect. We also believe that this technique might also offer the programmer some insight to develop more efficient variants of the CDS.

cs.DC↗

arXiv2018

Maintenance of Strongly Connected Component in Shared-memory Graph

In this paper, we present an on-line fully dynamic algorithm for maintaining strongly connected component of a directed graph in a shared memory architecture. The edges and vertices are added or deleted concurrently by fixed number of threads. To the best of our knowledge, this is the first work to propose using linearizable concurrent directed graph and is build using both ordered and unordered list-based set. We provide an empirical comparison against sequential and coarse-grained. The results show our algorithm's throughput is increased between 3 to 6x depending on different workload distributions and applications. We believe that there are huge applications in the on-line graph. Finally, we show how the algorithm can be extended to community detection in on-line graph.

cs.DC↗

arXiv2017

Building Efficient Concurrent Graph Object through Composition of List-based Set

In this paper, we propose a generic concurrent directed graph (for shared memory architecture) that is concurrently being updated by threads adding/deleting vertices and edges. The graph is constructed by the composition of the well known concurrent list-based set data-structure from the literature. Our construction is generic, in the sense that it can be used to obtain various progress guarantees, depending on the granularity of the underlying concurrent set implementation - either blocking or non-blocking. We prove that the proposed construction is linearizable by identifying its linearization points. Finally, we compare the performance of all the variants of the concurrent graph data-structure along with its sequential implementation. We observe that our concurrent graph data-structure mimics the performance of the concurrent list based set.

cs.DC↗