Parallel Programming---课程练习Q1

最新推荐文章于 2024-09-15 11:40:06 发布

qiqiqiqi77777

最新推荐文章于 2024-09-15 11:40:06 发布

阅读量76

点赞数

分类专栏： parallel 文章标签： c++ 算法数据结构性能优化

本文链接：https://blog.csdn.net/qiqiqiqi77777/article/details/130377519

版权

parallel 专栏收录该内容

1 篇文章 0 订阅

订阅专栏

Parallel Programming-Quiz1

#define SIZE 100
float waveData[SIZE];

void sequential() {
    for(int i = 0; i < SIZE; i++) {
        // Average the current, left, and right values into the current cell
        // i - 1 + SIZE % SIZE ensures that the returned index is always positive
        waveData[i] = (waveData[i - 1 + SIZE % SIZE] + waveData[i] + waveData[i + 1 % SIZE]) / 3;
    }
}

void parallel(int myIndex, int worldSize) {
    int blockSize = SIZE / worldSize;
    for (int i = myIndex * blockSize; i < (myIndex + 1) * blockSize; i++) {
        waveData[i] = (waveData[i - 1 + SIZE % SIZE] + waveData[i] + waveData[i+1 % SIZE]) / 3;
    }
}

The parallel implementation is called on two different threads with the parameters parallel(0, 2) and parallel(1,2). Assume that they are sharing memory in the waveData array.
两次的参数分别是(0,2), (1,2)
Q1:Will running the sequential program twice produce the same output?
Q2:Will running the parallel program twice produce the same output?
Q3:Will the sequential program output the same information as the parallel program?

Q1:Yes
Q2:Maybe
Q3:Maybe

Q1不用多说
Q2因为function的计算会涉及到 “交叉区域”，在该区域会因不同的thread执行的先后顺序差异产生不同的结果。
Q3因为有可能para就是很完美的先算完第一部分，再算第二部分，那么结果就会一样。