Machine learning configuration parameters. More...

#include <ml_types.h>

Data Fields
uint32_t	engine_id
	Engine ID to be configured. More...

uint32_t	max_models_created
	Maximum number of models. More...

uint32_t	max_models_loaded
	Maximum number of models loaded. More...

uint64_t	max_model_size
	Maximum model binary size in bytes. More...

odp_ml_compl_mode_t	load_mode_mask
	Load / unload completion modes. More...

odp_ml_compl_mode_t	run_mode_mask
	Run completion modes. More...

Detailed Description

Machine learning configuration parameters.

Examples: odp_ml_perf.c, and odp_ml_run.c.

Definition at line 336 of file api/spec/ml_types.h.

Field Documentation

◆ engine_id

uint32_t odp_ml_config_t::engine_id

Engine ID to be configured.

In a system with multiple ML engines, this parameter selects the engine to be configured. Values in the range 1..odp_ml_num_engines() select a specific engine and the special value of ODP_ML_ENGINE_ANY means that ODP selects the engine to be used.

The default value is ODP_ML_ENGINE_ANY.

Definition at line 347 of file api/spec/ml_types.h.

◆ max_models_created

uint32_t odp_ml_config_t::max_models_created

Maximum number of models.

Application may create and use this many models simultaneously. The default value is 1.

Definition at line 354 of file api/spec/ml_types.h.

◆ max_models_loaded

uint32_t odp_ml_config_t::max_models_loaded

Maximum number of models loaded.

Maximum number of models that the application will keep loaded simultaneously. The default value is 1.

Definition at line 362 of file api/spec/ml_types.h.

◆ max_model_size

uint64_t odp_ml_config_t::max_model_size

Maximum model binary size in bytes.

All model binaries application will pass to odp_ml_model_create() are this size or smaller.

Examples: odp_ml_perf.c, and odp_ml_run.c.

Definition at line 370 of file api/spec/ml_types.h.

◆ load_mode_mask

odp_ml_compl_mode_t odp_ml_config_t::load_mode_mask

Load / unload completion modes.

Mask of completion modes that application will use with model load/unload operations. Multiple modes may be selected, but it is implementation specific if some combinations are not supported. In case of an unsupported combination odp_ml_config() returns failure. Check odp_ml_capability_t.load for supported modes. The default value is 0.

Examples: odp_ml_perf.c, and odp_ml_run.c.

Definition at line 380 of file api/spec/ml_types.h.

◆ run_mode_mask

odp_ml_compl_mode_t odp_ml_config_t::run_mode_mask

Run completion modes.

Mask of completion modes that application will use with model run operations. Multiple modes may be selected, but it is implementation specific if some combinations are not supported. In case of an unsupported combination odp_ml_config() returns failure. Check odp_ml_capability_t.run for supported modes. The default value is 0.

Examples: odp_ml_perf.c, and odp_ml_run.c.

Definition at line 390 of file api/spec/ml_types.h.

The documentation for this struct was generated from the following file:

api/spec/ml_types.h

Data Fields