Adaptive Spatially Aware I/O for Multiresolution Particle Data Layouts

Will Usher, Xuan Huang, Steve Petruzza, Sidharth Kumar, Stuart R. Slattery, Sam T. Reeve, Feng Wang, Chris R. Johnson, and Valerio Pascucci

In IEEE International Parallel & Distributed Processing Symposium (IPDPS), 2021.

**Fig. 1:** An overview of our adaptive two-phase I/O pipeline. (a) Given the number of particles on each rank, rank 0 constructs the Aggregation Tree to create leaves with similar numbers of particles. Each leaf is assigned to a rank responsible for aggregating the data and writing it to disk. (b) Each rank sends its data to its aggregator. (c) Each aggregator constructs our multiresolution data layout and writes it to disk. (d) The aggregators send the local value ranges and root node bitmaps for each attribute to rank 0, which populates the Aggregation Tree with the bitmaps and writes it out.

Abstract

Large-scale simulations on nonuniform particle distributions that evolve over time are widely used in cosmology, molecular dynamics, and engineering. Such data are often saved in an unstructured format that neither preserves spatial locality nor provides metadata for accelerating spatial or attribute subset queries, leading to poor performance of visualization tasks. Furthermore, the parallel I/O strategy used typically writes a file per process or a single shared file, neither of which is portable or scalable across different HPC systems. We present a portable technique for scalable, spatially aware adaptive aggregation that preserves spatial locality in the output. We evaluate our approach on two supercomputers, Stampede2 and Summit, and demonstrate that it outperforms prior approaches at scale, achieving up to 2.5× faster writes and reads for nonuniform distributions. Furthermore, the layout written by our method is directly suitable for visual analytics, supporting low-latency reads and attribute-based filtering with little overhead.

Content

PDF

Publisher's Version

Presentation

Adaptive Spatially Aware I/O Library

Benchmark scripts used to run the evaluation

ExaMPM modified to use our I/O library

BibTeX

      @inproceedings{usher_aggtree_2021,
  booktitle = {35th IEEE International Parallel & Distributed Processing Symposium (IPDPS)},
  title = {{Adaptive} {Spatially} {Aware} {I}/{O} for {Multiresolution} {Particle} {Data} {Layouts}},
  author = {Usher, Will and Huang, Xuan and Petruzza, Steve and Kumar, Sidharth
    and Slattery, Stuart R. and Reeve, Sam T. and Wang, Feng and Johnson, Chris. R.
    and Pascucci, Valerio},
  year = {2021},
  doi={10.1109/IPDPS49936.2021.00063}
}