Yuhua Tang

Design of Counters Based on Memristors

Memristor is a nonlinear resistor with memory. It has special circuitry characteristics, which give it the ability of both storing information and operating logics. In 2010, researchers from HP Lab found that, with given stimulating voltages, memristors could finish implication logic operations. Memristor has attracted the researchers from both universities and the industry. Counter is a circuit implementing counting in digital systems and there has not been any research on counters designed with memristors. So this paper aims at designing a counter based on memristors. We first analyze and model the characteristics of memristor with HSPICE, and then design counters based on logic operations with memristors. The experimental results and analysis show that our design is right in function and has good scalability.

Reducing Static Energy in Supercomputer Interconnection Networks Using Topology-Aware Partitioning

The key to reducing static energy in supercomputers is switching off their unused components. Routers are the major components of a supercomputer. Whether routers can be effectively switched off or not has become the key to static energy management for supercomputers. For many typical applications, the routers in a supercomputer exhibit low utilization. However, there is no effective method to switch the routers off when they are idle. By analyzing the router occupancy in time and space, for the first time, we present a routing-policy guided topology partitioning methodology to solve this problem. We propose topology partitioning methods for three kinds of commonly used topologies (mesh, torus and fat-tree) equipped with the three most popular routing policies (deterministic routing, directionally adaptive routing and fully adaptive routing). Based on the above methods, we propose the key techniques required in this topology partitioning based static energy management in supercomputer interconnection networks to switch off unused routers in both time and space dimensions. Three topology-aware resource allocation algorithms have been developed to handle effectively different job-mixes running on a supercomputer. We validate the effectiveness of our methodology by using Tianhe-2 and a simulator for the aforementioned topologies and routing policies. The energy savings achieved on a subsystem of Tianhe-2 range from 3.8 to 79.7 percent. This translates into a yearly energy cost reduction of up to half a million US dollars for Tianhe-2.

The way to develop software towards exascale computing

This paper preliminarily discusses how to develop software towards exascale computing. Two typical development models are studied, one is “all-round contract” which means all the functional modules are implemented in a single framework, and the other is “co-design” which using the existing packages to realize functional modules. As a widely used CFD software whose numerical solving module is implemented in a typical “all-round contract” manner, OpenFOAM is concerned in this work. After redesigning and reimplementing the solving module of OpenFOAM in a “co-design” way by inserting PETSc, the source lines of the code decrease dramatically, which improves the development efficiency. Tests of two CFD benchmark cases and a practical large-scale case on a 128-node cluster show that the newly implemented numerical module also has a higher solving efficiency, compared with the original numerical module in OpenFOAM. Therefore, it is recommended to develop software in a “co-design” manner towards exascale computing, and softwares already implemented in “all-round contract” way should also be reconsidered.

A multi-user performance analysis framework for CFD simulations

This paper proposes a multi-user performance analysis framework for computational fluid dynamics (CFD) simulations. Independent performance datasets are given according to different requirements of computer programmers, algorithm developers and application developers. The performance analysis framework is designed based on the general procedure of CFD processing: pre-processing, problem solving and post-processing. Optimisation of the data acquisition is carried out based on data dependence to realise lightweight. A profiler is implemented in OpenFOAM based on the framework, and experiments are conducted to verify the performance information extraction and analysis. The results on a subsystem of Tianhe-1A indicate that the multiuser performance tool can probe the performance information successfully during the parallel execution. And from the performance results, we found that the direction of mesh partition turns out to be a critical factor of simulation performance for CFD problems.

A Hybrid Parallel Algorithm for Solving Eeuler Equation Using Explicit RKDG Method Based on OpenFOAM

OpenFOAM is a framework of the open source C CFD toolbox for flexible engineering simulation, which uses finite volume method (FVM) in the discretization of partial differential equations (PDEs). The problem solving procedure in OpenFOAM consists in equations dicretization stage, equations solving stage and field limiting stage. The best parallelism is limited by the equation solving stage, which contains communications. Compared to FVM, discontinuous Galerkin (DG) method is a high-order discretization method, which can accelerate the convergence of the residuals over same mesh scale and has higher resolution of the flow. Based on OpenFOAM with DG method, the ratio of overhead in equations discretization stage increases, especially when solving Euler equations using an explicit method. The equations discretization stage has a better potential parallelism than the other two stages due to no existence of communication. In this paper, we will analysis the difference of time cost in these three stages between original OpenFOAM and OpenFOAM with DG method. By decoupling these three stages, a hybrid parallel algorithm for solving PDEs is proposed and implemented based on OpenFOAM with DG method. The experimental results show that the simulation time is reduced by 16%, and the relative speedup of the hybrid parallel algorithm is up to 2.88 compared to the original parallel algorithm with the same degree of parallelism.

The Curve Boundary Design and Performance Analysis for DGM Based on OpenFOAM

OpenFOAM is a widely used numerical simulation software, and Discontinuous Galerkin method (DGM), a high-order numerical method, has been developed on OpenFOAM. In order to obtain meaningful numerical simulations, curve boundary is needed, but it has not been implemented on OpenFOAM. In this paper, based on codeStream function of original OpenFOAM, we design and implement curve boundary interface with reference to the interface of original OpenFOAM, so that users can use C++ code to describe curve boundary. Furthermore, in order to move the high-order points on the linear boundary to the curve boundary, we propose an algorithm to move each high-order point to a specific position on the curve, where the normal of this position passes through the origin point. Experimental results based on the flow around a cylinder show that curve boundary is needed by DGM numerical simulation, and DGM high-order simulation is much more efficient than DGM low-order. Typically, when the error of drag coefficient is about 0.03, the DGM high-order can save \(89.6\%\) time cost and \(83.0\%\) memory cost.

A High Order Discontinuous Galerkin Method Based RANS Turbulence Framework for OpenFOAM

Discontinuous Galerkin is one of the most promising high order method in CFD. However, there is a surprising lack of user-friendly software based on DGM. HopeFOAM is a high order extension of OpenFOAM. This paper designed a high order discontinuous Galerkin method based RANS turbulence framework on HopeFOAM, by inheriting the principle and user interfaces. The one equation turbulence model Spalart-Allmaras is integrated into the framework. User-friendly interfaces are provided to configure turbulence parameters and choosing calculation methods. The turbulence model and wall distance calculation are discretized by high order DGM. Results show that this framework can achieve similar simulation results comparing to literature.

Practical Performance Models for High-Order CFD Simulation

This paper discusses the performance model for practical computational fluid dynamics (CFD) simulation. Based on the analysis of the conventional qualitative model and empirical knowledge in the simulation, a set of practical models for the linear solution in high-order finite element method (FEM) simulation are established: a linear function model for the storage consumption, and a power function model for the solution time. The unknown parameters in the models are estimated by linear fitting and nonlinear fitting, respectively. Tests on a typical high-order FEM platform show that the proposed models are highly accurate and practical. In addition, it has been tested that nonlinear fitting is better than linear fitting in our case.

An Adaptive Visualization Tool for High Order Discontinuous Galerkin Method with Quadratic Elements

High order discretization method is one of the most popular research topics on Computational Fluid Dynamics (CFD). However, the development of visualization tools suited for high order methods obtains not as much attention. Most of the software used to visualize numerical solutions do not support high order data format. To visualize the high order solution, the most common method is to tessellate the solution with linear subdivision and then generate the values by resampling. This paper describes an efficient approach for Discontinuous Galerkin Method (DGM) based on OpenFOAM and VTK. We design a method to estimate the visualization error and introduce gaussian quadrature method to calculate it. Under the limit of visualization error set by user, the high order DGM solution is tessellated into a set of quadratic elements and converted into VTK data. By implementing the interpolation interface, this tools can support other discretization methods. The tool is tested with a series of cases. Accurately visualization of geometric information and field attributes are obtained. Comparing to linear methods, less computation and space cost are needed to reach the same visualization error limit.

Detailed and clock-driven simulation for HPC interconnection network

Performance and energy consumption of high performance computing (HPC) interconnection networks have a great significance in the whole supercomputer, and building up HPC interconnection network simulation platform is very important for the research on HPC software and hardware technologies. To effectively evaluate the performance and energy consumption of HPC interconnection networks, this article designs and implements a detailed and clock-driven HPC interconnection network simulation platform, called HPC-NetSim. HPC-NetSim uses applicationdriven workloads and inherits the characteristics of the detailed and flexible cycle-accurate network simulator. Besides, it offers a large set of configurable network parameters in terms of topology and routing, and supports router’s on/off states.We compare the simulated execution time with the real execution time of Tianhe-2 subsystem and the mean error is only 2.7%. In addition, we simulate the network behaviors with different network structures and low-power modes. The results are also consistent with the theoretical analyses.