MueLu
Version of the Day
|
#include <MueLu_PerfModels_fwd.hpp>
Public Member Functions | |
PerfModels () | |
void | stream_vector_make_table (int KERNEL_REPEATS, int LOG_MAX_SIZE=20) |
bool | has_stream_vector_table () const |
double | stream_vector_copy_lookup (int SIZE_IN_BYTES) |
double | stream_vector_add_lookup (int SIZE_IN_BYTES) |
double | latency_corrected_stream_vector_copy_lookup (int SIZE_IN_BYTES) |
double | latency_corrected_stream_vector_add_lookup (int SIZE_IN_BYTES) |
double | stream_vector_lookup (int SIZE_IN_BYTES) |
double | latency_corrected_stream_vector_lookup (int SIZE_IN_BYTES) |
void | print_stream_vector_table (std::ostream &out, const std::string &prefix="") |
void | print_latency_corrected_stream_vector_table (std::ostream &out, const std::string &prefix="") |
void | pingpong_make_table (int KERNEL_REPEATS, int LOG_MAX_SIZE, const RCP< const Teuchos::Comm< int > > &comm) |
bool | has_pingpong_table () const |
double | pingpong_host_lookup (int SIZE_IN_BYTES) |
double | pingpong_device_lookup (int SIZE_IN_BYTES) |
void | print_pingpong_table (std::ostream &out, const std::string &prefix="") |
void | halopong_make_table (int KERNEL_REPEATS, int LOG_MAX_SIZE, const RCP< const Xpetra::Import< LocalOrdinal, GlobalOrdinal, Node > > &import) |
bool | has_halopong_table () const |
double | halopong_host_lookup (int SIZE_IN_BYTES_PER_MESSAGE) |
double | halopong_device_lookup (int SIZE_IN_BYTES_PER_MESSAGE) |
void | print_halopong_table (std::ostream &out, const std::string &prefix="") |
void | launch_latency_make_table (int KERNEL_REPEATS) |
bool | has_launch_latency_table () const |
double | launch_latency_lookup () |
void | print_launch_latency_table (std::ostream &out, const std::string &prefix="") |
Private Member Functions | |
void | print_stream_vector_table_impl (std::ostream &out, bool use_latency_correction, const std::string &prefix) |
Private Attributes | |
std::vector< int > | stream_sizes_ |
std::vector< double > | stream_copy_times_ |
std::vector< double > | stream_add_times_ |
std::vector< double > | latency_corrected_stream_copy_times_ |
std::vector< double > | latency_corrected_stream_add_times_ |
std::vector< int > | pingpong_sizes_ |
std::vector< double > | pingpong_host_times_ |
std::vector< double > | pingpong_device_times_ |
std::vector< int > | halopong_sizes_ |
std::vector< double > | halopong_host_times_ |
std::vector< double > | halopong_device_times_ |
double | launch_and_wait_latency_ |
Definition at line 51 of file MueLu_PerfModels_fwd.hpp.
MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::PerfModels | ( | ) |
Definition at line 331 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::stream_vector_make_table | ( | int | KERNEL_REPEATS, |
int | LOG_MAX_SIZE = 20 |
||
) |
Definition at line 339 of file MueLu_PerfModels_def.hpp.
|
inline |
Definition at line 81 of file MueLu_PerfModels_decl.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::stream_vector_copy_lookup | ( | int | SIZE_IN_BYTES | ) |
Definition at line 373 of file MueLu_PerfModels_def.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::stream_vector_add_lookup | ( | int | SIZE_IN_BYTES | ) |
Definition at line 379 of file MueLu_PerfModels_def.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::latency_corrected_stream_vector_copy_lookup | ( | int | SIZE_IN_BYTES | ) |
Definition at line 391 of file MueLu_PerfModels_def.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::latency_corrected_stream_vector_add_lookup | ( | int | SIZE_IN_BYTES | ) |
Definition at line 397 of file MueLu_PerfModels_def.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::stream_vector_lookup | ( | int | SIZE_IN_BYTES | ) |
Definition at line 385 of file MueLu_PerfModels_def.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::latency_corrected_stream_vector_lookup | ( | int | SIZE_IN_BYTES | ) |
Definition at line 403 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::print_stream_vector_table | ( | std::ostream & | out, |
const std::string & | prefix = "" |
||
) |
Definition at line 408 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::print_latency_corrected_stream_vector_table | ( | std::ostream & | out, |
const std::string & | prefix = "" |
||
) |
Definition at line 413 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::pingpong_make_table | ( | int | KERNEL_REPEATS, |
int | LOG_MAX_SIZE, | ||
const RCP< const Teuchos::Comm< int > > & | comm | ||
) |
Definition at line 461 of file MueLu_PerfModels_def.hpp.
|
inline |
Definition at line 106 of file MueLu_PerfModels_decl.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::pingpong_host_lookup | ( | int | SIZE_IN_BYTES | ) |
Definition at line 469 of file MueLu_PerfModels_def.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::pingpong_device_lookup | ( | int | SIZE_IN_BYTES | ) |
Definition at line 475 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::print_pingpong_table | ( | std::ostream & | out, |
const std::string & | prefix = "" |
||
) |
Definition at line 480 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::halopong_make_table | ( | int | KERNEL_REPEATS, |
int | LOG_MAX_SIZE, | ||
const RCP< const Xpetra::Import< LocalOrdinal, GlobalOrdinal, Node > > & | import | ||
) |
Definition at line 515 of file MueLu_PerfModels_def.hpp.
|
inline |
Definition at line 123 of file MueLu_PerfModels_decl.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::halopong_host_lookup | ( | int | SIZE_IN_BYTES_PER_MESSAGE | ) |
Definition at line 523 of file MueLu_PerfModels_def.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::halopong_device_lookup | ( | int | SIZE_IN_BYTES_PER_MESSAGE | ) |
Definition at line 529 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::print_halopong_table | ( | std::ostream & | out, |
const std::string & | prefix = "" |
||
) |
Definition at line 534 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::launch_latency_make_table | ( | int | KERNEL_REPEATS | ) |
Definition at line 570 of file MueLu_PerfModels_def.hpp.
|
inline |
Definition at line 137 of file MueLu_PerfModels_decl.hpp.
double MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::launch_latency_lookup | ( | ) |
Definition at line 594 of file MueLu_PerfModels_def.hpp.
void MueLu::PerfModels< Scalar, LocalOrdinal, GlobalOrdinal, Node >::print_launch_latency_table | ( | std::ostream & | out, |
const std::string & | prefix = "" |
||
) |
Definition at line 599 of file MueLu_PerfModels_def.hpp.
|
private |
Definition at line 418 of file MueLu_PerfModels_def.hpp.
|
private |
Definition at line 148 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 149 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 150 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 151 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 152 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 154 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 155 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 156 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 158 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 159 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 160 of file MueLu_PerfModels_decl.hpp.
|
private |
Definition at line 162 of file MueLu_PerfModels_decl.hpp.