2021 · 42 citations · 37 references
ComputingComputer ScienceHigh-performance ComputingHeterogeneous ComputingParallel AlgorithmsParallel ComputingCommunication OverheadData Movement OptimizationData ParallelismMemory Hierarchy OptimizationParallel Computing SystemsParallel Performance AnalysisBandwidthKernel FusionKernel OptimizationKernel Performance OptimizationMemory BandwidthMemory Bandwidth OptimizationParallel Performance OptimizationVectorization
We study and evaluate performance optimization techniques for the HPCG benchmark on the newest generation Sunway supercomputer. Specifically, a two-level blocking scheme is proposed to expose adequate parallelism in the symmetric Gauss-Seidel kernel while keeping a fast convergence rate, a fine-grained kernel fusion technique is developed to alleviate the bandwidth load on local storage with small capacity, and a low overhead thread collaboration method is presented to efficiently move data between threads and hide its cost with data transfer operations. Test results show that the optimized HPCG code is able to exploit 73.0% of the theoretical memory bandwidth, and scale to over 42 million heterogeneous cores with 95.5% weak-scaling efficiency and 5.91 Pflop/s performance. We also study how the performance can be improved if the specific rules of HPCG are not fully obeyed, and design dependency preserving parallelization and vectorization methods, further boosting performance to 27.6 Pflop/s.
37
George Karypis, Vipin Kumar · University of Minnesota Digital Conservancy (University of Minnesota) · 1997 · 1.2K citations · Full text
The Sunway TaihuLight supercomputer: system and applications
Haohuan Fu, Junfeng Liao, Jinzhe Yang et al. · Science China Information Sciences · 2016 · 463 citations · Full text