thrust::execution_policy

Defined in thrust/execution_policy.h

template<typename DerivedPolicy> struct execution_policy : public thrust::detail::execution_policy_base<DerivedPolicy>

execution_policy is the base class for all Thrust parallel execution policies like thrust::host, thrust::device, and each backend system’s tag type.

Custom user-defined backends should derive a policy from this type in order to interoperate with Thrust algorithm dispatch.

The following code snippet demonstrates how to derive a standalone custom execution policy from thrust::execution_policy to implement a backend which only implements for_each:

#include <thrust/execution_policy.h>
#include <iostream>

// define a type derived from thrust::execution_policy to distinguish our custom execution policy:
struct my_policy : thrust::execution_policy<my_policy> {};

// overload for_each on my_policy
template<typename Iterator, typename Function>
Iterator for_each(my_policy, Iterator first, Iterator last, Function f)
{
  std::cout << "Hello, world from for_each(my_policy)!" << std::endl;

  for(; first < last; ++first)
  {
    f(*first);
  }

  return first;
}

struct ignore_argument
{
  void operator()(int) {}
};

int main()
{
  int data[4];

  // dispatch thrust::for_each using our custom policy:
  my_policy exec;
  thrust::for_each(exec, data, data + 4, ignore_argument());

  // can't dispatch thrust::transform because no overload exists for my_policy:
  //thrust::transform(exec, data, data, + 4, data, ::cuda::std::identity{}); // error!

  return 0;
}