V. A. Roganov, A. A. Kuznetsov, G. Matveev, V. I. Osipov, “Implementation of T-system with an open architecture for CUDA devices supporting dynamic parallelism and for hybrid computing clusters”, Program Systems: Theory and Applications, 2015, Volume 6, Issue 1,Pages <nobr>175

This article is cited in 2 papers

Hardware and Software for Supercomputers

Implementation of T-system with an open architecture for CUDA devices supporting dynamic parallelism and for hybrid computing clusters

V. A. Roganov, A. A. Kuznetsov, G. Matveev, V. I. Osipov

Program Systems Institute of RAS

Abstract: The article describes methods used to adapt the OpenTS parallel programming system and runtime for the optimal performance on hybrid clusters with computational nodes having accelerator hardware based on NVIDIA CUDA technology. (In Russian).

Key words and phrases: dynamic parallelization, T-system with an open architecture, OpenTS, T++ programming language, GPU accelerator, hybrid cluster systems, hash functions.

UDC: 519.682.3

Received: 22.12.2014
Accepted: 26.02.2015