Achieving a high sustained simulation performance is the most important concern in the HPC community. To this end, many kinds of HPC system architectures have been proposed, and the diversity of the HPC systems grows rapidly. Under this circumstance, a vector-parallel supercomputer SX-ACE has been designed to achieve a high sustained performance of memory-intensive applications by providing a high memory bandwidth commensurate with its high computational capability. This paper examines the potential of the modern vector-parallel supercomputer through the performance evaluation of SX-ACE using practical engineering and scientific applications. To improve the sustained simulation performances of practical applications, SX-ACE adopts an advanced memory subsystem with several new architectural features. This paper discusses how these features, such as MSHR, a large on-chip memory, and novel vector processing mechanisms, are beneficial to achieve a high sustained performance for large-scale engineering and scientific simulations. Evaluation results clearly indicate that the high sustained memory performance per core enables the modern vector supercomputer to achieve outstanding performances that are unreachable by simply increasing the number of fine-grain scalar processor cores. This paper also discusses the performance of the HPCG benchmark to evaluate the potentials of supercomputers with balanced memory and computational performance against heterogeneous and cutting-edge scalar parallel systems.
In this paper, we propose a runtime performance prediction model for automatic selection of accelerators to execute kernels in OpenCL. The proposed method is a history-based approach that uses profile data for performance prediction. The profile data are classified into some groups, from each of which its own performance model is derived. As the execution time of a kernel depends on some runtime parameters such as kernel arguments, the proposed method first identifies parameters affecting the execution time by calculating the correlation between each parameter and the execution time. A parameter with weak correlation is used for the classification of the profile data and the selection of the performance prediction model. A parameter with strong correlation is used for building a linear model for the prediction of the kernel execution time by using only the classified profile data. Experimental results clearly indicate that the proposed method can achieve more accurate performance prediction than conventional history-based approaches because of the profile data classification.
27 November 2019 in Lobachevskii Journal of Mathematics
The main goal of this paper is to demonstrate that the newest generation of NEC SX-Aurora TSUBASA architecture can perform large-scale graph processing extremely efficiently. This paper proposes...
Combinatorial clustering based on the Ising model is drawing attention as a high-quality clustering method. However, conventional Ising-based clustering methods using the Euclidean distance cannot handle irregular data. To overcome this problem, this paper proposes an Ising-based kernel clustering method. The kernel clustering method is designed based on two critical ideas. One is to perform clustering of irregular data by mapping the data onto a high-dimensional feature space by using a kernel trick. The other is the utilization of matrix–matrix calculations in the numerical libraries to accelerate preprocess for annealing. While the conventional Ising-based clustering is not designed to accept the transformed data by the kernel trick, this paper extends the availability of Ising-based clustering to process a distance matrix defined in high-dimensional data space. The proposed method can handle the Gram matrix determined by the kernel method as a high-dimensional distance matrix to handle irregular data. By comparing the proposed Ising-based kernel clustering method with the conventional Euclidean distance-based combinatorial clustering, it is clarified that the quality of the clustering results of the proposed method for irregular data is significantly better than that of the conventional method. Furthermore, the preprocess for annealing by the proposed method using numerical libraries is by a factor of up to million × from the conventional naive python’s implementation. Comparisons between Ising-based kernel clustering and kernel K-means reveal that the proposed method has the potential to obtain higher-quality clustering results than the kernel K-means as a representative of the state-of-the-art kernel clustering methods.