During the last years a large number of research works has focused on problems related to multi-core processors. Due to the possibilities of many cores, the number of opportunities in High Performance Computing (HPC) ...
During the last years a large number of research works has focused on problems related to multi-core processors. Due to the possibilities of many cores, the number of opportunities in High Performance Computing (HPC) has grown a lot. In fact, new fields related to HPC and processor architecture increase the future possibilities of a Grid-on-Chip (GoC). The goal of this paper is to show a high-throughput MCNoC (Multi-Cluster Network-on-Chip) as an alternative architecture to support clusters of cores and Grid features. In this new scenario data throughput, flexibility, and scalability are very important. The results verify that MCNoC has a similar area occupation and a better data throughput than a traditional Network-on-Chip.
For the next processor generation, many cores and parallelprogramming will provide high-throughput and high-performance processing. As a consequence, research works have studied on-chip interconnection architectures ...
详细信息
For the next processor generation, many cores and parallelprogramming will provide high-throughput and high-performance processing. As a consequence, research works have studied on-chip interconnection architectures to identify alternatives capable of decreasing the communication latencies. The objective of this paper is to present the evaluation of three well-known architectures (bus, crossbar switch and a conventional network-on-chip) in order to propose a multi-cluster network-on-chip architecture for parallelprocessing. The results show that a NoC composed of programmable routers and crossbar switches to interconnect clusters of cores has a better performance than conventional NoCs.
In this paper, we present a parallelization of a filtering algorithm related to non-linear anisotropic diffusion, used to enhance the performance of an application in a paralleldistributed system. The anisotropic dif...
详细信息
ISBN:
(纸本)9780769532578
In this paper, we present a parallelization of a filtering algorithm related to non-linear anisotropic diffusion, used to enhance the performance of an application in a paralleldistributed system. The anisotropic diffusion is a well-established technique for image enhancement by means of diffusivity functions, which act as border attenuators. However, it requires a high computational cost when a large amount of data is used. The proposed implementation was parallelized considering both point-to-point and collective communications, adopting the MPI paradigm. Results from both approaches indicate that the proposed algorithm has reached interesting levels of performance (81% and 93% of efficiency, respectively) when compared to the execution of one process in a single computer node. In addition, our results indicate an enhancement of around 21% utilizing the collective communication strategy when compared to point-to-point communication.
暂无评论