文件名称:高性能计算自选题目实验报告
介绍说明--下载内容均来自于网络,请自行研究使用
对于一副图像,比如8000*5333分辨率,我们在处理时,通常思路是从第1个像素开始,一直计算到最后一个像素。其实,目前不论手机还是个人电脑,处理器都是多核。那么完全可以将整副图像分成若干块,比如cpu为4核处理器,那么可以分成4块,每块图像大小为2000*5333,这样程序可以创建4个线程,每个处理器执行一个线程,每个线程处理一个图像块。虽然这样操作后,运算速度不会显著提升4倍,因为线程创建、释放、上下文切换都要耗些时间。但运算速度还是将明显提升,一般4核 vs 1核,运算时间将降低一半。(For an image, such as 8000 * 5333 resolution, we usually start with the first pixel and calculate it to the last pixel. In fact, at present, whether mobile phones or personal computers, processors are multi-core. Then the whole image can be divided into several blocks, such as CPU 4 core processor, then can be divided into four blocks, each image size of 2000*5333, so that the program can create four threads, each processor executes a thread, each thread processing an image block. Although this operation will not significantly increase the computing speed by four times, because thread creation, release, context switching will take some time. But the operation speed will be improved obviously. Generally, 4 cores vs 1 cores will reduce the operation time by half)
(系统自动生成,下载前可以参看下载内容)
下载文件列表
文件名 | 大小 | 更新时间 |
---|---|---|
高性能计算自选题目实验报告.docx | 5444624 | 2019-10-10 |