Improving the ratio of memory operations to floating-point operations in loops
ACM Transactions on Programming Languages and Systems (TOPLAS)
Proceedings of the 2002 ACM/IEEE conference on Supercomputing
A 15.2 TFlops Simulation of Geodynamo on the Earth Simulator
Proceedings of the 2004 ACM/IEEE conference on Supercomputing
Hardware-aware analysis and optimization of stable fluids
Proceedings of the 2008 symposium on Interactive 3D graphics and games
Proceedings of the 2007 ACM/IEEE conference on Supercomputing
Roofline: an insightful visual performance model for multicore architectures
Communications of the ACM - A Direct Path to Dependable Software
Performance evaluation of NEC SX-9 using real science and engineering applications
Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis
A scalable method for ab initio computation of free energies in nanoscale systems
Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis
Proceedings of the 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis
Hi-index | 0.00 |
Mechanisms of interactions among different scale phenomena play important roles for forecasting of weather and climate. Multi-scale Simulator for the Geoenvironment (MSSG), which deals with multi-scale multi-physics phenomena, is a coupled non-hydrostatic atmosphere-ocean model designed to be run efficiently on the Earth Simulator. We present its simulation results with the world-highest 1.9km horizontal resolution for the entire globe. To gain high performance by exploiting the system capabilities, we propose novel performance evaluation metrics that incorporate the effects of the data caching mechanism between CPU and memory. A potentially attainable computational performance is also introduced by evaluating both computational and memory intensities. With the useful code optimization guideline based on such metrics, we demonstrate that MSSG can achieve an excellent peak performance ratio of 32.2% on the Earth Simulator with the single-core performance found to be a key to reduced time-to-solution.