Reliability, Availability, and Serviceability for Petascale High-End Computing and Beyond, (Final Report).

摘要

Our project is a multi-institutional research effort that adopts interplay of RELIABILITY, AVAILABILITY, and SERVICEABILITY (RAS) aspects for solving resilience issues in highend scientific computing in the next generation of supercomputers. results lie in the following tracks: Failure prediction in a large scale HPC; Investigate reliability issues and mitigation techniques including in GPGPU-based HPC system; HPC resilience runtime & tools.