Reduce Consider the reduce function: template T reduce (T* array , size_t n) { T result = array [0]; for (int i=1; i

2 Prefix sum Prefixsum is an algorithm that has many uses in parallel computing. The algorithm computes pr[i] = j

3 Merge Sort There is limited parallelism in merge sort because the merge operation is naturally sequential. Question: Rewrite merge to make it suitable to execute with P processors. (Hint: it is one of these cases where you may have to increase complexity slightly.) Question: What are the work and critical path of the merge algorithm you created? Question: If you used that parallel merge in merge sort, what would the work and critical path of merge sort become?