Publications

Journal Articles

A Survey of Distributed Asynchronous Many-Task Models and Their Applications

Published in ACM Computing Surveys, 2026

A comprehensive survey of distributed asynchronous many-task (AMT) runtime systems — including Charm++, HPX, Legion, PaRSEC, Uintah, Chapel, and StarPU — covering their design principles, execution models, and the application domains where they provide tangible benefits over MPI+X.

Recommended citation: Joseph Schuchart, Patrick Diehl, Michael Bauer, Aurelien Bouteiller, Gregor Daiss, Engin Kayraklioglu, Shreyas Khandekar, Thomas Herault, John Holmen, Ritvik Rao, Alexander Strack, Elliott Slaughter, Jennifer Spinti, Jeremy Thornock, Alex Aiken, Olivier Aumage, Martin Berzins, George Bosilca, Bradford Chamberlain, Hartmut Kaiser, and Laxmikant Kale. (2026). "A Survey of Distributed Asynchronous Many-Task Models and Their Applications." ACM Computing Surveys. https://doi.org/10.1145/3840389
Download Paper

Conference Papers

Reconverse: A New Scheduling Library for Multithreaded Task-Based Runtime Systems

Published in 9th Annual Parallel Applications Workshop, Alternatives To MPI+X (PAW-ATM), held in conjunction with SC26 — accepted, to appear, 2026

We present Reconverse, a new scheduling library for task-based runtime systems that replaces Charm++’s aging Converse layer, showing improved performance and speedups of up to 2x on large-scale applications.

Recommended citation: Ritvik Rao, Jiakun Yan, Aditya Bhosale, and Laxmikant Kalé. (2026). "Reconverse: A New Scheduling Library for Multithreaded Task-Based Runtime Systems." 9th Annual Parallel Applications Workshop, Alternatives To MPI+X (PAW-ATM), held in conjunction with SC26. To appear.
Download Paper

Efficient and Portable Support for Overdecomposition on Distributed GPU Platforms

Published in 9th Annual Parallel Applications Workshop, Alternatives To MPI+X (PAW-ATM), held in conjunction with SC26 — accepted, to appear, 2026

We show that overdecomposition can be sustained efficiently and portably on GPUs across vendors, by mitigating host-side serialization costs transparently at the runtime layer in a single-source Charm++/Kokkos environment running unchanged on NVIDIA and AMD systems.

Recommended citation: Aditya Bhosale, Anant Jain, Shourya Goel, Ritvik Rao, Peddoju Sateesh Kumar, and Laxmikant Kalé. (2026). "Efficient and Portable Support for Overdecomposition on Distributed GPU Platforms." 9th Annual Parallel Applications Workshop, Alternatives To MPI+X (PAW-ATM), held in conjunction with SC26. To appear.
Download Paper

Efficient Cosmological Friends-of-Friends Computation with Fine-Grained Optimizations

Published in 16th IEEE/ACM Workshop on Irregular Applications: Architectures and Algorithms (IA³), held in conjunction with SC26 — accepted, to appear, 2026

We present a series of fine-grained optimizations to a parallel friends-of-friends (FoF) cluster-finding implementation built on the ParaTreeT and Union-Find libraries, achieving significant performance gains on highly imbalanced cosmological datasets without compromising cluster accuracy.

Recommended citation: Ritvik Rao and Laxmikant Kalé. (2026). "Efficient Cosmological Friends-of-Friends Computation with Fine-Grained Optimizations." 16th IEEE/ACM Workshop on Irregular Applications: Architectures and Algorithms (IA3), held in conjunction with SC26. To appear.
Download Paper

An Adaptive Asynchronous Approach for the Single-Source Shortest Paths Problem

Published in 14th IEEE/ACM Workshop on Irregular Applications: Architectures and Algorithms (IA³), held in conjunction with SC24, 2024

We introduce ACIC, a highly asynchronous approach to the single-source shortest paths problem that uses continuous concurrent introspection and adaptation to reduce speculative loss without constraining parallelism.

Recommended citation: Ritvik Rao, Kavitha Chandrasekar, and Laxmikant V. Kalé. (2024). "An Adaptive Asynchronous Approach for the Single-Source Shortest Paths Problem." SC24-W: Workshops of the International Conference for High Performance Computing, Networking, Storage and Analysis. pp. 697–702.
Download Paper