thrust

◆ exclusive_scan() [5/6]

template<typename DerivedPolicy , typename InputIterator , typename OutputIterator , typename T , typename AssociativeOperator >
__host__ __device__ OutputIterator thrust::exclusive_scan ( const thrust::detail::execution_policy_base< DerivedPolicy > &  exec,
InputIterator  first,
InputIterator  last,
OutputIterator  result,
init,
AssociativeOperator  binary_op 
)

exclusive_scan computes an exclusive prefix sum operation. The term 'exclusive' means that each result does not include the corresponding input operand in the partial sum. More precisely, init is assigned to *result and the value binary_op(init, *first) is assigned to *(result + 1), and so on. This version of the function requires both an associative operator and an initial value init. When the input and output sequences are the same, the scan is performed in-place.

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.
resultThe beginning of the output sequence.
initThe initial value.
binary_opThe associatve operator used to 'sum' values.
Returns
The end of the output sequence.
Template Parameters
DerivedPolicyThe name of the derived execution policy.
InputIteratoris a model of Input Iterator and InputIterator's value_type is convertible to OutputIterator's value_type.
OutputIteratoris a model of Output Iterator and OutputIterator's value_type is convertible to both AssociativeOperator's first_argument_type and second_argument_type.
Tis convertible to OutputIterator's value_type.
AssociativeOperatoris a model of Binary Function and AssociativeOperator's result_type is convertible to OutputIterator's value_type.
Precondition
first may equal result but the range [first, last) and the range [result, result + (last - first)) shall not overlap otherwise.

The following code snippet demonstrates how to use exclusive_scan to compute an in-place prefix sum using the thrust::host execution policy for parallelization:

#include <thrust/scan.h>
...
int data[10] = {-5, 0, 2, -3, 2, 4, 0, -1, 2, 8};
thrust::exclusive_scan(thrust::host, data, data + 10, data, 1, binary_op); // in-place scan
// data is now {1, 1, 1, 2, 2, 2, 4, 4, 4, 4 }
See also
http://www.sgi.com/tech/stl/partial_sum.html