Skip to main content
Press Release | Mathematics and Computer Science

Argonne to highlight research at the ACM Federated Computing Research conferences

FCRC 2023 assembles a spectrum of affiliated research conferences and workshops into a week-long, co-located meeting.

The data reduction and management team in the Mathematics and Computer Science (MCS) division at Argonne National Laboratory recently had six papers accepted for two of the FCRC affiliated conferences: the 37th ACM International Conference on Supercomputing (ICS) and the 32nd International Symposium on High-Performance Parallel and Distributed Computing (HPDC). Both ICS and HPDC are considered premier international forums for the presentation of research in high-performance computing and computing systems. Moreover, three other papers by MCS division staff also were accepted at the FCRC conferences.

The acceptance of so many papers at ICS and HPDC is for a single team is exceptional,” said Franck Cappello, a senior scientist in Argonne’s MCS division and coauthor of five of the papers. Having a single paper accepted for either conference has been considered an achievement for a research team.”

The papers represent challenging areas of research raised by high-performance computing systems. The massive amounts of data being produced by high-performance applications require novel compression methods – both lossless (in which the file size is reduced but the original data can be reconstructed from the compressed data) and lossy (in which some of the data considered less important is discarded). Four of the papers accepted for the conferences deal with compression issues (boldface means an MCS division staff member):

  • Jinyang Liu, Sheng Di, Kai Zhao, Xin Liang, Zizhong Chen, Franck Cappello, FZ: A Flexible Auto-tuned Modular Error-Bounded Compression Framework for Scientific Data,” ICS23.  This is one of three papers nominated for Best Paper Award of ICS’23 (to be determined at the conference)
  • Boyuan Zhang, Jiannan Tian, Sheng Di, Xiaodong Yu, Martin Swany, Dingwen Tao, Franck Cappello, GPULZ: Optimizing LZSS Lossless Compression for Multi-byte Data on Modern GPUs,” ICS23
  • Milan Shah, Xiaodong Yu, Sheng Di, Michela Becchi, Franck Cappello, Lightweight Huffman Coding for Efficient GPU Compression,” ICS23
  • Boyuan Zhang, Jiannan Tian, Sheng Di, Xiaodong Yu, Yunhe Feng, Xin Liang, Dingwen Tao, Franck Cappello, Z-GPU: A Fast and High-Ratio Lossy Compressor for Scientific Computing Applications on GPUs,” HPDC23

Another issue facing high-performance computing is the need to manage large-scale intermediate data generated by complex applications. Checkpointing is increasingly being used, often relying on accelerators such as graphic processing units, or GPUs. One of the papers addresses this issue, exploiting GPU memory to enable informed caching and prefetching.

  • Avinash Maurya, Bogdan Nicolae, Aleena Rafique, Thierry Tonellot, Franck Cappello, J. Hussain. GPU-Enabled Asynchronous Multi-level Checkpoint Caching and Prefetching,” HPDC23

The ability to share and reuse deep learning (DL) models also is becoming increasing important, especially in the adoption of artificial intelligence in industrial and scientific applications. One of the team’s papers accepted for the ICS23 conference describes a lightweight, scalable DL repository to make access to deep learning models easier.

  • Meghana Madhyastha, Robert Underwood, R. Burns, Bogdan Nicolae, DStore: A Lightweight Scalable Learning Model Repository with Fine-Grain Tensor-Level Access.” ICS23

In addition to these six papers by the data reduction and management team, three other papers by MCS division researchers have been accepted for FCRC2023. Two of these will be presented at ICS23: a paper by Srinivas Eswar et al. titled Distributed Memory Parallel JointNMF” and a paper by Xiaodong Yu (with collaborators) titled HEAT: A Highly Efficient and Affordable Training System for Collaborative Filtering Based Recommendations on Multi-core CPUs.” A third paper, by Ji Liu (with collaborators), titled PULSE-Level Variational Quantum Algorithms for Molecular Energy,” will be presented at the Quantum Classical Cooperative Computing workshop held in conjunction with HPDC23.

This year’s FCRC conference will take place in Orlando, Florida, in June. For more information, see the ACM Federated Computing Research Conference website at https://​fcrc​.acm​.org/.