Intel Threading Building Blocks

Cook Until Done: parallel_while

For some loops, the end of the iteration space is not known in advance, or the loop body may add more iterations to do before the loop exits. You can deal with both situations using the template class tbb::parallel_while.

A linked list is an example of an iteration space that is not known in advance. In parallel programming, it is usually better to use dynamic arrays instead of linked lists because accessing items in a linked list is inherently serial. But if you are limited to linked lists, if the items can be safely processed in parallel, and if processing each item takes at least a few thousand instructions, you can use parallel_while in a situation where the serial form is as shown in Example 4-1.

Example 4-1. Original list processing code

void SerialApplyFooToList( Item*root ) {
    for( Item* ptr=root; ptr!=NULL; ptr=ptr->next )
        Foo(pointer->data);
}

If Foo takes at least a few thousand instructions to run, you can get parallel speedup by converting the loop to use parallel_while. Unlike the templates described earlier, parallel_while is a class, not a function, and it requires two user-defined objects. The first object defines the stream of items. The object must have a method, pop_if_ present, such that when bool b =pop_if_present(v) is invoked, it sets v to the next iteration value if there is one and returns true. If there are no more iterations, it returns false. Example 4-2 shows a typical implementation of pop_if_present.

Example 4-2. pop_if_present ...