Efficiency of Programming Model Based on Explicit Creating Light Threads

The results of research of efficiency of execution of parallel programs created on the basis of the different programming models are presented. Programs run on an SMP formed a motherboard with few multi-core crystals. One programming model is based on the implicit to the user allocation of generated parallel threads on the provided SMP hardware resources and traditional approaches to the synchronization of threads. This model is represented by Open MP. The second model uses the explicit allocation of lightweights threads on resources SMP with the use of developed at Sandia National Laboratory library Qthreads.

Keywords

scalable parallel programs, the effectiveness of parallel computing, lightweight threads.

References

1. Kogge P. ExaScale Computing Study: Technology Challenges in Achieving Exascale Systems. University of Notre Dame. September 28, 2008. Available at: http://www.sc.doe.gov/ascr/Research/CS/DARPA%20exascale%20%20hardware%20(2008).pdf (accessed 4 April 2012)
2. Wheeler K. Qthreads: An API for Programming with Millions of Lightweight Threads. Workshop on Multithreaded Architectures and Applications at IEEE IPDPS, April, 2008, pp. 1 - 8.
3. Korneev V.V. Next Supercomputers Generation [Sleduyushchee pokolenie superkomputerov]. Otkrytye sistemy [Open systems], 2008, no. 8, pp. 14 - 19.
4. Korneev V.V. An Approach to the Programming of Supercomputers on the Basis of Multicore Multithreaded Chips [Podkhod k programmirovaniu superkomputerov na base mnogoyadernykh multitredovykh kristallov]. Vychislitel'nye metody i programmirovanie [Numerical Methods and Programming], 2009, vol. 10, pp. 123 - 128.
5. Wen X., Vishkin U. FPGA-Based Prototype of a PRAM-On-Chip Processor. CF'08, May 5-7, 2008, Ischia, Italy, pp. 55 - 66.
6. Qthreads (2012). Available at http://www.cs.sandia.gov/qthreads (accessed 2 March 2012).