Showing 1–3 of 3 results for author: Ahnert, K

Programming CUDA and OpenCL: A Case Study Using Modern C++ Libraries

Authors: Denis Demidov, Karsten Ahnert, Karl Rupp, Peter Gottschling

Abstract: We present a comparison of several modern C++ libraries providing high-level interfaces for programming multi- and many-core architectures on top of CUDA or OpenCL. The comparison focuses on the solution of ordinary differential equations and is based on odeint, a framework for the solution of systems of ordinary differential equations. Odeint is designed in a very flexible way and may be easily a… ▽ More We present a comparison of several modern C++ libraries providing high-level interfaces for programming multi- and many-core architectures on top of CUDA or OpenCL. The comparison focuses on the solution of ordinary differential equations and is based on odeint, a framework for the solution of systems of ordinary differential equations. Odeint is designed in a very flexible way and may be easily adapted for effective use of libraries such as Thrust, MTL4, VexCL, or ViennaCL, using CUDA or OpenCL technologies. We found that CUDA and OpenCL work equally well for problems of large sizes, while OpenCL has higher overhead for smaller problems. Furthermore, we show that modern high-level libraries allow to effectively use the computational resources of many-core GPUs or multi-core CPUs without much knowledge of the underlying technologies. △ Less

Submitted 26 April, 2013; v1 submitted 27 December, 2012; originally announced December 2012.

Comments: 21 pages, 4 figures, submitted to SIAM Journal of Scientific Computing and accepted

Odeint - Solving ordinary differential equations in C++

Authors: Karsten Ahnert, Mario Mulansky

Abstract: Many physical, biological or chemical systems are modeled by ordinary differential equations (ODEs) and finding their solution is an every-day-task for many scientists. Here, we introduce a new C++ library dedicated to find numerical solutions of initial value problems of ODEs: odeint (www.odeint.com). odeint is implemented in a highly generic way and provides extensive interoperability at top per… ▽ More Many physical, biological or chemical systems are modeled by ordinary differential equations (ODEs) and finding their solution is an every-day-task for many scientists. Here, we introduce a new C++ library dedicated to find numerical solutions of initial value problems of ODEs: odeint (www.odeint.com). odeint is implemented in a highly generic way and provides extensive interoperability at top performance. For example, due to it's modular design it can be easily parallized with OpenMP and even runs on CUDA GPUs. Despite that, it provides a convenient interface that allows for a simple and easy usage. △ Less

Submitted 15 October, 2011; originally announced October 2011.

Comments: 4 pages, 1 figure

Journal ref: IP Conf. Proc. - September 14, 2011 - Volume 1389, pp. 1586-1589

Metaprogramming Applied to Numerical Problems

Authors: Mario Mulansky, Karsten Ahnert

Abstract: From the discovery that the template system of C++ forms a Turing complete language in 1994, a programming technique called Template Metaprogramming has emerged that allows for the creation of faster, more generic and better code. Here, we apply Template Metaprogramming to implement a generic Runge-Kutta scheme that can be used to numerically solve ordinary differential equations. We show that usi… ▽ More From the discovery that the template system of C++ forms a Turing complete language in 1994, a programming technique called Template Metaprogramming has emerged that allows for the creation of faster, more generic and better code. Here, we apply Template Metaprogramming to implement a generic Runge-Kutta scheme that can be used to numerically solve ordinary differential equations. We show that using Template Metaprogramming results in a significantly improved performance compared to a classical implementation. △ Less

Submitted 14 October, 2011; originally announced October 2011.

Comments: 4 pages,2 figures,3 listings

Journal ref: AIP Conf. Proc. 1389, pp. 1582-1585 (2011)