Summary

Chapter summary

• It is an error to execute operations in parallel unless those operations are independent from each other.

• parallel() accesses the elements of a range in parallel.

• Tasks can explicitly be created, started, and waited for by task(), executeInNewThread(), and yieldForce(), respectively.

• The exceptions that are escaped from tasks can be caught later by most of the parallelism functions like yieldForce().

• asyncBuf() iterates the elements of an InputRange semi-eagerly in parallel.

• map() calls functions with the elements of an InputRange semi-eagerly in parallel.

• amap() calls functions with the elements of a RandomAccessRange fully- eagerly in parallel.

• reduce() makes calculations over the elements of a RandomAccessRange in parallel.

• map(), amap(), and reduce() can take multiple functions and return the results as tuples.

• When needed, TaskPool objects other than taskPool can be used.

---