Developing and Optimizing the error-bounded lossy compression on GPU

PI Name Sheng Di, MCS
PI Institution Argonne National Laboratory
Project Description

This research is focused on the development and performance optimization for the state-of-the-art lossy compressor (SZ) on GPU. Specifically, we have developed a cuda version for SZ. We will keep developing one or more versions (e.g., based on Kokkos or oneAPI) for other GPU architectures such as Intel GPU, and also optimize the performance.

Testbed

Intel GPUs (oneAPI)