Abstract
The scientific demand for more accurate modeling of the climate system calls for more computing power to support higher resolutions, inclusion of more component models, more complicated physics schemes, and larger ensembles. As the recent improvements in computing power mostly come from the increasing number of nodes in a system and the integration of heterogeneous accelerators, how to scale the computing problems onto more nodes and various kinds of accelerators has become a challenge for the model development. This paper describes our efforts on developing a highly scalable framework for performing global atmospheric modeling on heterogeneous supercomputers equipped with various accelerators, such as GPU (Graphic Processing Unit), MIC (Many Integrated Core), and FPGA (Field Programmable Gate Arrays) cards. We propose a generalized partition scheme of the problem domain, so as to keep a balanced utilization of both CPU resources and accelerator resources. With optimizations on both computing and memory access patterns, we manage to achieve around 8 to 20 times speedup when comparing one hybrid GPU or MIC node with one CPU node with 12 cores. Using a customized FPGA-based data-flow engines, we see the potential to gain another 5 to 8 times improvement on performance. On heterogeneous supercomputers, such as Tianhe-1A and Tianhe-2, our framework is capable of achieving ideally linear scaling efficiency, and sustained double-precision performances of 581 Tflops on Tianhe-1A (using 3750 nodes) and 3.74 Pflops on Tianhe-2 (using 8644 nodes). Our study also provides an evaluation on the programming paradigm of various accelerator architectures (GPU, MIC, FPGA) for performing global atmospheric simulation, to form a picture about both the potential performance benefits and the programming efforts involved.
Highlights
As one of the traditional high-performance computing (HPC) applications, atmospheric models have been one of the major consumers of supercomputer computing cycles [1] and major drivers of new HPC technologies [2,3,4].Especially in recent years, the more and more urgent economic and social challenges brought by global warming are asking for better scientific understanding of the climate change mechanism and more accurate climate models to make predictions into the future climate risks
While previous climate models are mostly based on CPU-only clusters, the large-scale supercomputers that are built these years are mostly employing heterogeneous systems that rely on accelerators, such as GPUs [8], Many Integrated Core (MIC) [9], or even reconfigurable FPGAs [10]
Similar results have been reported for existing atmospheric models that employ cubed-sphere meshes, such as Community Atmospheric Model (CAM)-SE [32], and atmospheric component (AM3) of the Geophysical Fluid Dynamics Laboratory (GFDL) coupled model (CM3) [33]
Summary
In recent years, the more and more urgent economic and social challenges brought by global warming are asking for better scientific understanding of the climate change mechanism and more accurate climate models to make predictions into the future climate risks These demands, along with the inherent development of climate science, are calling for significantly more computing power, to support higher modeling resolutions [5], to include descriptions of more complex physics processes [6], and to enable more accurate modeling through larger ensembles [7]. While previous climate models are mostly based on CPU-only clusters, the large-scale supercomputers that are built these years are mostly employing heterogeneous systems that rely on accelerators, such as GPUs [8], MICs [9], or even reconfigurable FPGAs [10]. While these hybrid systems are providing more computing power, the inclusion of different architectures within one system is leading to more complicated programming patterns, and huge challenges on porting existing climate model softwares
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
