get_kernel Subroutine

private subroutine get_kernel(dims, kernel_type, effort, base_storage, props, tile_size, block_rows, kernel, force_effort, neighbor_data)

Uses

  • proc~~get_kernel~~UsesGraph proc~get_kernel get_kernel iso_c_binding iso_c_binding proc~get_kernel->iso_c_binding iso_fortran_env iso_fortran_env proc~get_kernel->iso_fortran_env

Compiles kernel and caches it. Returns compiled kernel.

Arguments

Type IntentOptional Attributes Name
integer(kind=int32), intent(in) :: dims(:)

Local dimensions to process
type(kernel_type_t), intent(in) :: kernel_type

Type of kernel to build
type(dtfft_effort_t), intent(in) :: effort

How thoroughly dtFFT searches for the optimal transpose kernel
integer(kind=int64), intent(in) :: base_storage

Number of bytes needed to store single element
type(device_props), intent(in) :: props

GPU architecture properties
integer(kind=int32), intent(out) :: tile_size

Size of the tile in shared memory
integer(kind=int32), intent(out) :: block_rows

Number of rows in each block processed by each thread
type(CUfunction), intent(out) :: kernel

Compiled kernel to return
logical, intent(in), optional :: force_effort

Should effort be forced or not
integer(kind=int32), intent(in), optional :: neighbor_data(:)

Neighbor data for pipelined kernels

Calls

proc~~get_kernel~~CallsGraph proc~get_kernel get_kernel fname fname proc~get_kernel->fname interface~cudaeventcreate cudaEventCreate proc~get_kernel->interface~cudaeventcreate interface~cudaeventdestroy cudaEventDestroy proc~get_kernel->interface~cudaeventdestroy interface~cudaeventelapsedtime cudaEventElapsedTime proc~get_kernel->interface~cudaeventelapsedtime interface~cudaeventrecord cudaEventRecord proc~get_kernel->interface~cudaeventrecord interface~cudaeventsynchronize cudaEventSynchronize proc~get_kernel->interface~cudaeventsynchronize interface~cudafree cudaFree proc~get_kernel->interface~cudafree interface~cudamalloc cudaMalloc proc~get_kernel->interface~cudamalloc interface~cudastreamsynchronize cudaStreamSynchronize proc~get_kernel->interface~cudastreamsynchronize interface~to_str to_str proc~get_kernel->interface~to_str mpi_abort mpi_abort proc~get_kernel->mpi_abort proc~create_nvrtc_module create_nvrtc_module proc~get_kernel->proc~create_nvrtc_module proc~cudageterrorstring cudaGetErrorString proc~get_kernel->proc~cudageterrorstring proc~culaunchkernel cuLaunchKernel proc~get_kernel->proc~culaunchkernel proc~evaluate_analytical_performance evaluate_analytical_performance proc~get_kernel->proc~evaluate_analytical_performance proc~generate_candidates generate_candidates proc~get_kernel->proc~generate_candidates proc~get_conf_configs_to_test get_conf_configs_to_test proc~get_kernel->proc~get_conf_configs_to_test proc~get_conf_forced_kernel_optimization get_conf_forced_kernel_optimization proc~get_kernel->proc~get_conf_forced_kernel_optimization proc~get_conf_kernel_optimization_enabled get_conf_kernel_optimization_enabled proc~get_kernel->proc~get_conf_kernel_optimization_enabled proc~get_conf_log_enabled get_conf_log_enabled proc~get_kernel->proc~get_conf_log_enabled proc~get_conf_measure_iters get_conf_measure_iters proc~get_kernel->proc~get_conf_measure_iters proc~get_conf_measure_warmup_iters get_conf_measure_warmup_iters proc~get_kernel->proc~get_conf_measure_warmup_iters proc~get_conf_stream get_conf_stream proc~get_kernel->proc~get_conf_stream proc~get_kernel_args get_kernel_args proc~get_kernel->proc~get_kernel_args proc~get_kernel_instance get_kernel_instance proc~get_kernel->proc~get_kernel_instance proc~get_kernel_launch_params get_kernel_launch_params proc~get_kernel->proc~get_kernel_launch_params proc~get_kernel_string get_kernel_string proc~get_kernel->proc~get_kernel_string proc~is_unpack_kernel is_unpack_kernel proc~get_kernel->proc~is_unpack_kernel proc~pop_nvtx_domain_range pop_nvtx_domain_range proc~get_kernel->proc~pop_nvtx_domain_range proc~push_nvtx_domain_range push_nvtx_domain_range proc~get_kernel->proc~push_nvtx_domain_range proc~sort_candidates_by_score sort_candidates_by_score proc~get_kernel->proc~sort_candidates_by_score proc~write_message write_message proc~get_kernel->proc~write_message proc~double_to_string double_to_string interface~to_str->proc~double_to_string proc~float_to_string float_to_string interface~to_str->proc~float_to_string proc~int32_to_string int32_to_string interface~to_str->proc~int32_to_string proc~int64_to_string int64_to_string interface~to_str->proc~int64_to_string proc~int8_to_string int8_to_string interface~to_str->proc~int8_to_string proc~create_nvrtc_module->interface~to_str proc~create_nvrtc_module->proc~get_conf_log_enabled proc~create_nvrtc_module->proc~get_kernel_string proc~create_nvrtc_module->proc~write_message none~check nvrtc_module%check proc~create_nvrtc_module->none~check proc~add nvrtc_module_cache%add proc~create_nvrtc_module->proc~add proc~create~13 nvrtc_module%create proc~create_nvrtc_module->proc~create~13 interface~cudageterrorstring_c cudaGetErrorString_c proc~cudageterrorstring->interface~cudageterrorstring_c proc~string_c2f string_c2f proc~cudageterrorstring->proc~string_c2f proc~count_bank_conflicts count_bank_conflicts proc~evaluate_analytical_performance->proc~count_bank_conflicts proc~estimate_bank_conflict_ratio estimate_bank_conflict_ratio proc~evaluate_analytical_performance->proc~estimate_bank_conflict_ratio proc~estimate_coalescing estimate_coalescing proc~evaluate_analytical_performance->proc~estimate_coalescing proc~estimate_occupancy estimate_occupancy proc~evaluate_analytical_performance->proc~estimate_occupancy proc~estimate_memory_pressure estimate_memory_pressure proc~generate_candidates->proc~estimate_memory_pressure proc~estimate_optimal_padding estimate_optimal_padding proc~generate_candidates->proc~estimate_optimal_padding proc~find_valid_combination find_valid_combination proc~generate_candidates->proc~find_valid_combination interface~get_conf_internal get_conf_internal proc~get_conf_configs_to_test->interface~get_conf_internal proc~get_conf_forced_kernel_optimization->interface~get_conf_internal proc~get_conf_kernel_optimization_enabled->interface~get_conf_internal proc~get_conf_log_enabled->interface~get_conf_internal proc~get_conf_measure_iters->interface~get_conf_internal proc~get_conf_measure_warmup_iters->interface~get_conf_internal proc~get_conf_stream->fname proc~get_conf_stream->mpi_abort proc~get_conf_stream->proc~cudageterrorstring interface~cudastreamcreate cudaStreamCreate proc~get_conf_stream->interface~cudastreamcreate proc~get_kernel_instance->interface~to_str proc~get_kernel_instance->proc~get_conf_log_enabled proc~get_kernel_instance->proc~get_kernel_string proc~get_kernel_instance->proc~write_message is_null_ptr is_null_ptr proc~get_kernel_instance->is_null_ptr proc~get nvrtc_module%get proc~get_kernel_instance->proc~get interface~nvtxdomainrangepop_c nvtxDomainRangePop_c proc~pop_nvtx_domain_range->interface~nvtxdomainrangepop_c interface~nvtxdomainrangepushex_c nvtxDomainRangePushEx_c proc~push_nvtx_domain_range->interface~nvtxdomainrangepushex_c proc~astring_f2c astring_f2c proc~push_nvtx_domain_range->proc~astring_f2c proc~create_nvtx_domain create_nvtx_domain proc~push_nvtx_domain_range->proc~create_nvtx_domain mpi_comm_rank mpi_comm_rank proc~write_message->mpi_comm_rank mpi_finalized mpi_finalized proc~write_message->mpi_finalized proc~get_conf_internal_int32 get_conf_internal_int32 interface~get_conf_internal->proc~get_conf_internal_int32 proc~get_conf_internal_logical get_conf_internal_logical interface~get_conf_internal->proc~get_conf_internal_logical proc~check_instance nvrtc_module%check_instance none~check->proc~check_instance proc~check_module nvrtc_module%check_module none~check->proc~check_module proc~add->interface~to_str proc~add->proc~get_conf_log_enabled proc~add->proc~write_message proc~create~9 nvrtc_module_cache%create proc~add->proc~create~9 proc~string_f2c string_f2c proc~astring_f2c->proc~string_f2c proc~create_nvtx_domain->proc~astring_f2c interface~nvtxdomaincreate_c nvtxDomainCreate_c proc~create_nvtx_domain->interface~nvtxdomaincreate_c proc~create~13->fname proc~create~13->mpi_abort proc~create~13->proc~cudageterrorstring proc~create~13->proc~get_conf_log_enabled proc~create~13->proc~pop_nvtx_domain_range proc~create~13->proc~push_nvtx_domain_range proc~create~13->proc~write_message interface~mem_free_host mem_free_host proc~create~13->interface~mem_free_host proc~compile_program compile_program proc~create~13->proc~compile_program proc~destroy~11 nvrtc_module%destroy proc~create~13->proc~destroy~11 proc~get_code get_code proc~create~13->proc~get_code proc~mem_alloc_host mem_alloc_host proc~create~13->proc~mem_alloc_host proc~nvrtcgeterrorstring nvrtcGetErrorString proc~create~13->proc~nvrtcgeterrorstring proc~estimate_bank_conflict_ratio->proc~count_bank_conflicts proc~estimate_optimal_padding->proc~count_bank_conflicts proc~get->fname proc~get->mpi_abort proc~get->proc~cudageterrorstring proc~get->none~check proc~get_mangled_name get_mangled_name proc~get->proc~get_mangled_name interface~is_null_ptr is_null_ptr proc~string_c2f->interface~is_null_ptr interface~is_null_ptr->interface~is_null_ptr proc~is_null_funptr is_null_funptr interface~is_null_ptr->proc~is_null_funptr proc~check_instance->none~check proc~compile_program->fname proc~compile_program->interface~to_str proc~compile_program->mpi_abort proc~compile_program->proc~astring_f2c proc~compile_program->proc~string_c2f proc~compile_program->proc~mem_alloc_host proc~compile_program->proc~nvrtcgeterrorstring proc~destroy_strings destroy_strings proc~compile_program->proc~destroy_strings proc~set_name_expression set_name_expression proc~compile_program->proc~set_name_expression proc~destroy~11->fname proc~destroy~11->mpi_abort proc~destroy~11->proc~cudageterrorstring proc~destroy~11->is_null_ptr proc~destroy~11->proc~nvrtcgeterrorstring proc~get_code->proc~is_unpack_kernel proc~add_line codegen_t%add_line proc~get_code->proc~add_line proc~get_mangled_name->fname proc~get_mangled_name->mpi_abort proc~get_mangled_name->proc~nvrtcgeterrorstring proc~get_name_expression get_name_expression proc~get_mangled_name->proc~get_name_expression interface~aligned_alloc aligned_alloc proc~mem_alloc_host->interface~aligned_alloc proc~nvrtcgeterrorstring->proc~string_c2f proc~destroy_string string%destroy_string proc~destroy_strings->proc~destroy_string proc~get_name_expression->interface~to_str proc~get_name_expression->proc~astring_f2c proc~set_name_expression->fname proc~set_name_expression->mpi_abort proc~set_name_expression->proc~nvrtcgeterrorstring proc~set_name_expression->proc~get_name_expression

Called by

proc~~get_kernel~~CalledByGraph proc~get_kernel get_kernel proc~create~8 kernel_device%create proc~create~8->proc~get_kernel