thrust

◆ reduce() [3/6]

template<typename DerivedPolicy , typename InputIterator , typename T >
__host__ __device__ T thrust::reduce ( const thrust::detail::execution_policy_base< DerivedPolicy > &  exec,
InputIterator  first,
InputIterator  last,
init 
)

reduce is a generalization of summation: it computes the sum (or some other binary operation) of all the elements in the range [first, last). This version of reduce uses init as the initial value of the reduction. reduce is similar to the C++ Standard Template Library's std::accumulate. The primary difference between the two functions is that std::accumulate guarantees the order of summation, while reduce requires associativity of the binary operation to parallelize the reduction.

Note that reduce also assumes that the binary reduction operator (in this case operator+) is commutative. If the reduction operator is not commutative then thrust::reduce should not be used. Instead, one could use inclusive_scan (which does not require commutativity) and select the last element of the output array.

The algorithm's execution is parallelized as determined by exec.

Parameters
execThe execution policy to use for parallelization.
firstThe beginning of the input sequence.
lastThe end of the input sequence.
initThe initial value.
Returns
The result of the reduction.
Template Parameters
DerivedPolicyThe name of the derived execution policy.
InputIteratoris a model of Input Iterator and if x and y are objects of InputIterator's value_type, then x + y is defined and is convertible to T.
Tis convertible to InputIterator's value_type.

The following code snippet demonstrates how to use reduce to compute the sum of a sequence of integers including an intialization value using the thrust::host execution policy for parallelization:

#include <thrust/reduce.h>
...
int data[6] = {1, 0, 2, 2, 1, 3};
int result = thrust::reduce(thrust::host, data, data + 6, 1);
// result == 10
See also
http://www.sgi.com/tech/stl/accumulate.html