ANJU M A
Department of Computer Science, Texas State University
Hello!
I am a post-doctoral researcher in the Department of Computer Science at Texas State University, where I am part of the Efficient Computing Laboratory (ECL), led by Dr. Martin Burtscher. My research interests are in high performance computing, graph algorithms, and massively parallel computations. I received my PhD (and M.Tech) in Computer Science from IIT Madras. My PhD dissertation work (supervised by Dr. Rupesh Nasre, IIT Madras) was on locking techniques for improving concurrency of hierarchical applications. Prior to that, I worked as a software developer at Amdocs, Pune, India.
Education & Work Experience
Academic/career timeline:
September 2023 - Present: PostDoc (Department of Computer Science, Texas State University, USA)
Area of research: Parallel graph algorithms, GPU computing
December 2016 - August 2023: PhD (Department of Computer Science and Engineering , IIT Madras, India)
Area of research: Parallel computing
July 2016 - November 2016: Adhoc faculty (Department of Computer Science and Engineering , NIT Calicut, India)
Taught Operating Systems (theory and lab) to 3rd year B Tech students
July 2012 - May 2014: M Tech (Department of Computer Science and Engineering , IIT Madras, India)
Area of work: Computer Architecture, Memory
Sept 2008 - May 2012: Java developer (Amdocs Development Center India Private Ltd, Pune, India)
Worked in different phases of SDLC (mainly in Java back end development)
July 2004 - May 2008: BE (Hons) (Department of Computer Science, BITS Pilani, India)
Major courses studied:
Program Analysis
GPU Programming
Parallel Computer Architecture
Advanced Data Structures and Algorithms
Mathematical Concepts of Computer Science
Computer Architecture
Advanced Computer Architecture
Concurrent Programming
Advanced Programming Lab (C++)
Research
My research work is in the area of parallelization of programs, with special focus on optimizing the degree of concurrency and thread synchronization. I am interested in finding efficient parallelization techniques suitable for various applications (on both multi-core and many-core systems) that handle static as well as dynamic data. The real-life impact of a parallel application comes mainly from its faster execution and completion compared to the sequential version of the program. I am excited about making applications run faster!
We collaborated with Shell on a project which simulates oil repository formation on a parallel computer. We achieved a reduction in runtime from more than 24 hours to less than 1 hour for an application that simulates the percolation in porous rock structures. (The natural phenomenon takes years to form a repository!)
We worked on dynamic clustering, where the data points are dynamic (data points may be added / removed or their location may be changed). We came up with an efficient parallel implementation of single linkage clustering, with focus on scenarios such as splitting and merging of clusters. As part of another work, we came up with a technique to optimize the degree of concurrency of dominator-based locking in parallel applications that handle hierarchical data. We also proposed a technique that allows the co-existence of fine-grained and hierarchical locks in multi-threaded applications that support dynamic updates.
Publications:
Anju, M. A., and Rupesh Nasre. Multi-Interval DomLock: Toward Improving Concurrency in Hierarchies. ACM Transactions on Parallel Computing 9.3, TOPC (2022).
Mongandampulath Akathoott, Anju, and Rupesh Nasre. Single‐Linkage Clustering of Dynamic Data. Concurrency and Computation: Practice and Experience: e7447, CCPE, (2023).
Anju, M. A., and Rupesh Nasre. FlexiGran: Flexible Granularity Locking in Hierarchies. 30th International European Conference on Parallel and Distributed Computing, EuroPar (2024).
Contact
Address:
Efficient Computing Laboratory, M13,
Department of Computer Science,
Texas State University, San Marcos, Texas, USA
78666
Email:
msd [dot] anju [at] gmail [dot] com / anju [dot] m [dot] a [at] txstate.edu