Zhiyi Huang, John Hine, Laurent Lefevre, Tony McGregor, Yi Pan, Hong Shen, Zhiwei Xu
2010 Journal of Supercomputing  
This special issue titled "Emerging Research in Parallel and Distributed Computing" reflects the recent research efforts of authors of the Ninth International Conference on Parallel and Distributed Computing, Applications and Technologies (PDCAT 2008). This special issue is divided into two sections: the first addresses the issues in multicore computing, while the second contains the new developments in grid computing. Parallel computing is facing new challenges with the advent of multicore and
more » ... many-core architectures. In the past decades, CPU speeds doubled every 18 months by increasing the clock speed. Software has had a "free ride" achieving better performance through increasing hardware speed. However, this free ride is now over due to physical limitations such as heat dissipation and energy consumption in the manufacture of CPU chips. Because of the stalling CPU clock speeds, parallel computing becomes the only way for software to increase performance. Researchers are now working hard to provide suitable tools and environments for parallel computing, from languages to runtime systems, from hardware accelerators to scheduling algorithms for operating systems. In the invited paper "The Cilk++ Concurrency Platform", Charles Leiserson gives an overview of a state-of-the-art parallel programming environment-Cilk++, which is derived from MIT Cilk. Cilk++ consists of a compiler, a runtime system, and a race-detection tool. It is simple to use and guarantees to load-balance computations within theoretical bounds. One recent prominent feature of Cilk++ is the "hyperobject" library to mitigate the data race problem. In "Data Race: Tame the Beast", Kai-
doi:10.1007/s11227-010-0399-x fatcat:pkotq634inctzph4azqoixotsy