Thrust Integration¶
- group thrust_integrations
Typedefs
-
template<typename T>
using device_vector = thrust::device_vector<T, rmm::mr::thrust_allocator<T>>¶ Alias for a thrust::device_vector that uses RMM for memory allocation.
-
using thrust_exec_policy_t = thrust::detail::execute_with_allocator<mr::thrust_allocator<char>, thrust::cuda_cub::execute_on_stream_base>¶
Synchronous execution policy for allocations using thrust.
-
using thrust_exec_policy_nosync_t = thrust_exec_policy_t¶
When used with Thrust < 1.16.0, thrust_exec_policy_nosync_t is an alias for thrust_exec_policy_t
-
using exec_policy_nosync = exec_policy¶
When used with Thrust < 1.16.0, exec_policy_nosync is an alias for exec_policy.
-
class exec_policy : public thrust_exec_policy_t¶
- #include <exec_policy.hpp>
Helper class usable as a Thrust CUDA execution policy that uses RMM for temporary memory allocation on the specified stream.
Public Functions
-
inline explicit exec_policy(cuda_stream_view stream = cuda_stream_default, device_async_resource_ref mr = mr::get_current_device_resource())¶
Construct a new execution policy object.
- Parameters:
stream – The stream on which to allocate temporary memory
mr – The resource to use for allocating temporary memory
-
inline explicit exec_policy(cuda_stream_view stream = cuda_stream_default, device_async_resource_ref mr = mr::get_current_device_resource())¶
-
template<typename T>