Design and Implementation of Relay Nodes in SBC Multi-display System
Design and Implementation of Relay Nodes in SBC Multi-display System
复制标题
SBC多屏显示系统中继节点的设计与实现
DOI：
10.1007/978-3-030-98012-2_37
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Shinji Shimojo
中科院分区：
文献类型：
--
作者：
Yuga Takahata;Yoshiyuki Kido;Shinji Shimojo
多显示器系统使用虚拟化技术创建单个大屏幕，允许多人同时观看高分辨率视频和图像。因此，它们经常与高性能计算机系统结合使用，以可视化模拟结果。然而，多显示器系统的建造成本也往往很高，因为它们通常需要高性能的计算机。为了克服这一局限性，我们以前的研究提出了一种由一个头节点和多个显示节点组成的多显示系统，该系统使用单板计算机(SBCS)构建。不幸的是，在多显示器系统中增加显示器数量也会增加头节点上的处理负载，这最终会产生一个瓶颈，从而降低整个系统的速度。因此，本研究提出在头部和显示节点之间引入中继节点作为图像帧处理任务的分布和并行化的方式。在这里，我们的方法可以减少头节点的中央处理单元(CPU)的使用率，同时也提供了足够的余量在中继节点的CPU利用率。这些评估结果表明，在多显示系统中实现中继节点和执行分布式并行处理可以消除系统瓶颈。
Multi-display systems use virtualization technology to create single large screens that allow multiple people to view high-resolution videos and images simultaneously. As such, they are often used in conjunction with high-performance computer systems to visualize simulation results. However, multi-display systems also tend to be expensive to construct because they generally require high-performance computers. To overcome this limitation, our previous study proposed a multi-display system consisting of one head node and multiple display nodes constructed using single-board computers (SBCs). Unfortunately, increasing the number of displays in a multi-display system also increases the processing load on the head node, which eventually produces a bottleneck that slows the whole system. Therefore, this study proposes the introduction of relay nodes between the head and display nodes as a way to distribute and parallelize the image frame processing task. Herein, we show that our method can reduce the head node’s central processing unit (CPU) usage while also providing sufficient margin in the relay node CPU utilization rates. These evaluation results demonstrate that system bottlenecks can be eliminated by implementing relay nodes and performing distributed parallel processing in multi-display systems.