A5078397602- Advanced Data Storage Technologies
- Parallel Computing and Optimization Techniques
- Distributed and Parallel Computing Systems
- Caching and Content Delivery
- Numerical Methods and Algorithms
- Scientific Computing and Data Management
- Computer Graphics and Visualization Techniques
- Model Reduction and Neural Networks
- Algorithms and Data Compression
New Jersey Institute of Technology
2018-2024
Scientific simulations generate large amounts of floating-point data, which are often not very compressible using the traditional reduction schemes, such as deduplication or lossless compression. The emergence lossy compression holds promise to satisfy data demand from HPC applications; however, has been widely adopted in science production. We believe a fundamental reason is that there lack understanding benefits, pitfalls, and performance on scientific data. In this paper, we conduct...
With the high volume and velocity of scientific data produced on high-performance computing systems, it has become increasingly critical to improve compression performance. Leveraging general tolerance reduced accuracy in applications, lossy compressors can achieve much higher ratios with a user-prescribed error bound. However, they are still far from satisfying reduction requirements applications. In this paper, we propose evaluate idea that need be preconditioned prior compression, such...
High-performance computing (HPC) applications generate large amounts of floating-point data that need to be stored and analyzed efficiently extract the insights advance knowledge discovery. With growing disparities between compute I/O, optimizing storage stack alone may not suffice cure I/O problem. There has been a strong push in HPC communities perform reduction before is transmitted order lower cost. However, as now, neither lossless nor lossy compressors can achieve adequate ratio...
Scientific simulations on high performance computing (HPC) platforms generate large quantities of data. To bridge the widening gap between compute and I/O, enable data to be more efficiently stored analyzed, simulation outputs need refactored, reduced, appropriately mapped storage tiers. However, a systematic solution support these steps has been lacking in current HPC software ecosystem. that end, this paper develops SIRIUS, progressive JPEG-like management scheme for storing analyzing big...
As high-performance computing (HPC) is being scaled up to exascale accommodate new modeling and simulation needs, I/O has continued be a major bottleneck in the end-to-end scientific processes. Nevertheless, prior work this area mostly aimed maximize average performance, there been lack of study solutions that can manage performance variation on HPC systems. This aims take advantage storage characteristics explore application level are interference-aware. In particular, we monitor data...