多维嵌套循环的OpenMP [英] Multi-dimensional nested OpenMP loop

问题描述

什么是并行OpenMP中多维尴尬并行循环的正确方法？维数在编译时是已知的，但它的尺寸将是大是没有的。它们中的任何可以是一个，两个，或一百万。当然，我不希望ñ OMP并行的一个N维环 ...

What is the proper way to parallelize a multi-dimensional embarrassingly parallel loop in OpenMP? The number of dimensions is known at compile-time, but which dimensions will be large is not. Any of them may be one, two, or a million. Surely I don't want N omp parallel's for an N-dimensional loop...

思考：

• 问题的概念很简单。只有最外层的大循环需要进行并行处理，但环尺寸在编译时未知的，可能会改变。

• The problem is conceptually simple. Only the outermost 'large' loop needs to be parallelized, but the loop dimensions are unknown at compile-time and may change.

将动态设置 OMP_SET_NUM_THREADS（1）和的#pragma OMP的时间表（静态，huge_number）使某些环路并行化无操作？这会不会有意外的副作用/开销？感觉就像一个杂牌组装电脑。

Will dynamically setting omp_set_num_threads(1) and #pragma omp for schedule(static, huge_number) make certain loop parallelizations a no-op? Will this have undesired side-effects/overhead? Feels like a kludge.

借助 OpenMP规范（2.10，A.38，A.39）告诉符合和不符合嵌套并行，但不建议这个问题的最好方法之间的区别。

The OpenMP Specification (2.10, A.38, A.39) tells the difference between conforming and non-conforming nested parallelism, but doesn't suggest the best approach to this problem.

重新排序的循环是可能的但可能会导致大量的缓存未命中。展开是可能的，但不平凡的。有另一种方式？

Re-ordering the loops is possible but may result in a lot of cache-misses. Unrolling is possible but non-trivial. Is there another way?

下面是想什么我并行：

for(i0=0; i0<n[0]; i0++) {
  for(i1=0; i1<n[1]; i1++) {
    ...
       for(iN=0; iN<n[N]; iN++) {
         <embarrasingly parallel operations>
       }
    ...
  }
}

谢谢！

推荐答案

的崩溃指令可能是你在找什么，所描述的此处。这将基本上形成单个环路，其然后parallized，并且被设计为恰好这些类型的情形。所以，你会怎么做：

The collapse directive is probably what you're looking for, as described here. This will essentially form a single loop, which is then parallized, and is designed for exactly these sorts of situations. So you'd do:

#pragma omp parallel for collapse(N)
for(int i0=0; i0<n[0]; i0++) {
  for(int i1=0; i1<n[1]; i1++) {
    ...
       for(int iN=0; iN<n[N]; iN++) {
         <embarrasingly parallel operations>
       }
    ...
  }
}

和准备就绪。

这篇关于多维嵌套循环的OpenMP的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！