Search

Different communication-aware mapping techniques were proposed in recent years for improving the performance of distributed systems based on both, off-chip and on-chip networks. Some of these proposals were based on heuristic search for finding pseudo-optimal assignments of tasks and processing elements. However, the technology integration improvements have allowed a significant increase in the number of network nodes, requiring the acceleration of the heuristic search. In this paper, we propose a comparative study of the local search method used in a communication-aware mapping technique, when implemented on different parallel architectures. We compare the performance provided by a version of the local search method when executed on a single Graphics Processing Unit (GPU) with the one provided by the MPI version executed on a supercomputer with the same theoretical performance of the GPU platform, in order to study a fair scenario. We have considered a GPU based on the Fermi architecture, evaluating the improvements achieved by some new architectural features of this platform. The results show that a mixed parallel implementation on a single GPU outperforms the MPI implementation of the local search method. These results validate the GPU implementation as a very cost-effective accelerator for the local search method. [ABSTRACT FROM AUTHOR]

The computing capabilities of current multi-core and many-core architectures have been used in crowd simulations for both enhancing crowd rendering and simulating continuum crowds. However, improving the scalability of crowd simulation systems by exploiting the inherent parallelism of these architectures is still an open issue. In this paper, we propose different parallelization strategies for the collision check procedure that takes place in agent-based simulations. These strategies are designed for exploiting the parallelism in both multi-core and many-core architectures like graphic processing units (GPUs). As for the many-core implementations, we analyse the bottlenecks of a previous GPU version of the collision check algorithm, proposing a new GPU version that removes the bottlenecks detected. In order to fairly compare the GPU with the multi-core implementations, we propose a parallel CPU version that uses read--copy update (RCU), a new synchronization method which significantly improves performance. We perform a comparison study of these different implementations. On the one hand, the comparison study shows the first performance evaluation of RCU in a real user-space application with complex data structures. On the other hand, the comparison shows that the GPU greatly accelerates the collision test with respect to any other implementation optimized for multi-core CPUs. In addition, we analyse the efficiency of the different implementations taking into account the theoretical performance and power consumption of each platform. The evaluation results show that the GPU-based implementation consumes less energy and provides a minimum speedup of 45× with respect to any of the CPU-based implementations. Since interactivity is a hard constraint in crowd simulations, this acceleration of the collision check process represents a significant improvement in the overall system throughput and response time. Therefore, the simulations are significantly accelerated, and the system throughput and scalability are improved. [ABSTRACT FROM AUTHOR]

SUMMARY In this paper, we look at the acceleration of weakly coupled electromechanics using the graphics processing unit (GPU). Specifically, we port to the GPU a number of components of [ABSTRACT FROM AUTHOR]