cuda::experimental::stf::executable_graph_cache#

class executable_graph_cache#

Public Types

using per_device_map_t = ::std::unordered_multimap<::std::pair<size_t, size_t>, entry, hash_pair>#

Public Functions

inline executable_graph_cache()#
inline cuda::std::pair<::std::shared_ptr<cudaGraphExec_t>, bool> query(
size_t nnodes,
size_t nedges,
::std::shared_ptr<cudaGraph_t> g,
)#
struct entry#

Public Functions

inline entry(
executable_graph_cache *cache,
::std::shared_ptr<cudaGraphExec_t> exec_g_,
size_t footprint,
)#
inline void lru_refresh()#

Public Members

executable_graph_cache *cache#
::std::shared_ptr<cudaGraphExec_t> exec_g#
size_t last_use#
size_t footprint#
struct hash_pair#

Public Functions

inline size_t operator()(const std::pair<size_t, size_t> &p) const#