Recent Publication at ICPP 2026

A big shout out to Drs. Hussain, Ballard and Devarakonda! The paper, entitled “High-Performance Star-M SVD for Big Data Compression,” is accepted to the International Conference on Parallel Processing (ICPP 2026). “High-Performance star-M SVD for Big Data Compression,” presents the first high-performance implementation of the star-M tensor SVD, a tensor-based method for compressing large multidimensional scientific datasets. The implementation scales to 64 shared-memory cores and is demonstrated on climate, computational fluid dynamics, and X-ray crystallography data. On climate data, the method preserves physically relevant extreme events better than standard matrix-based compression, while achieving speedups over existing MATLAB implementations of star-M SVD.

This work is in collaboration with Argonne National Laboratory and the Indian Institute of Science.

Congratulations!