摘要
NVIDIA推出的CUDA(计算统一设备架构)是基于GPU进行通用计算的开发平台,非常适合大规模的并行数据计算。在GPU流处理器架构下用CUDA技术实现编码并行化,并针对流处理器架构特点进行内存读写等方面的优化。在此采用CUDA技术,实现了计算机桌面环境的多屏幕融合显示的纯软件拼接系统。该系统不但较以往单一的视频拼接系统功能更加强大,也较采用分屏器等硬件辅助的融合系统成本更低,适应性更强。目前实验表明,CUDA技术在并行处理方面的优越性使得该系统画面实时处理快,互动展示性好,具有很大的商业使用前景。
CUDA（compute unified device architecture）launched by NVIDIA is a development platform based on the GPU for general-purpose computation and suitable for large-scale parallel data computation. The parallel processing compression algo-rithm is implemented by CUDA under the stream processor architecture of GPU. The optimization of reading and writing memory was carried out according to the features of the stream processor architecture. The software mosaic system with multi-screen fu-sion display in the computer’s desktop environment was achieved with CUDA technology. A software-system based on the CUDA technology is introduced. This system is more powerful than any previous video fusion systems,and cheaper than the hardware-assisted fusion system such as VGA. The experiment shows that the CUDA technology has the advantages of parallel processing to make this system has fast real-time image processing and interactivity. It will be widely used in business filed.
出处
《现代电子技术》
2013年第20期94-96,共3页
Modern Electronics Technique
基金
西安石油大学青年创新基金资助项目(YS28032111)
