Calculating the sum of a large vector in parallel

Question

Problem background

I have a program that currently takes way too long to sum up large std::vectors of ~100 million elements using std::accumulat

Answer 1

You can use Boost Asio as a thread pool. But there's not a lot of sense in it unless you have... asynchronous IO operations to coordinate.

In this answer to "c++ work queues with blocking" I show two thread_pool implementations:

• Solution #1: one based on boost::asio::io_service
• Solution #2: the other based on boost::thread primitives

Both accept any void() signature compatible task. This means, you could wrap your function-that-returns-the-important-results in a packaged_task<...> and get the future from it.