如何使用＆quot #pragma op paralallel for''正确地获取利润以填充阵列QVECTOR
我正在学习C ++，并尝试通过图表制作自己的计算器。在这一部分中，当我尝试用坐标填充数组时。我想使用多线程使其更快。但是当我这样做时，代码开始工作得更糟，速度较慢。我该如何解决问题，我在做什么错？ 它根本不起作用
omp_set_num_threads(2);
#pragma omp parallel for
for (current = start; current < finish; current ++) {
Calc b(a.get_string());
double y_value = b.parsing((double)current/1000);
#pragma omp critical
{
if (y_value > yL || y_value < yR) {
x->push_back((double)current/1000);
y->push_back(y_value);
}
}
顺便说一句，如果没有“ #pragma op critical” x和y，这是QVECTOR， 。 请尝试尽可能简单地表达自己，因为我只是在学习。非常感谢！
I'm learning c++ and trying to make my own calculator with charting. In this part, when I try to fill an array with coordinates. I want to make it faster using multithreading. But when I did so, the code began to work much worse, slower. How can I solve the problem what am I doing wrong? By the way, it doesn't work at all without "#pragma omp critical"
omp_set_num_threads(2);
#pragma omp parallel for
for (current = start; current < finish; current ++) {
Calc b(a.get_string());
double y_value = b.parsing((double)current/1000);
#pragma omp critical
{
if (y_value > yL || y_value < yR) {
x->push_back((double)current/1000);
y->push_back(y_value);
}
}
x and y this is QVector.
Please try to express yourself as simply as possible because I'm just learning. Thank you all very much!
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
当然不需要关键部分，因为条件
y_value＆gt; yl || y_value＆lt; yr可以并行完成，假设yl是yr剩下的未修改，并且因为xandy可以用大小完成启动初始化，然后您可以在向量上执行直接访问，例如x [current-start] =（double）Current/1000;。另外，请注意，使用“ private> private（current）”在并行指令中使用当前私有。The critical section is certainly not needed because the condition
y_value > yL || y_value < yRcan be done in parallel assumingyLareyRleft unmodified and becausexandycan be initialized with the sizefinish-startand you can then perform direct accesses on the vector likex[current-start] = (double)current/1000;. Also, please note that it is probably better to putcurrentprivate using the clauseprivate(current)in the parallel directive.