OrtTensorRTProviderOptionsV2 Struct Reference

Options for the TensorRT provider that are passed to SessionOptionsAppendExecutionProvider_TensorRT_V2. Please note that this struct is similar to OrtTensorRTProviderOptions but only to be used internally. Going forward, new trt provider options are to be supported via this struct and usage of the publicly defined OrtTensorRTProviderOptions will be deprecated over time. User can only get the instance of OrtTensorRTProviderOptionsV2 via CreateTensorRTProviderOptions. More...

#include <tensorrt_provider_options.h>

Public Member Functions
OrtTensorRTProviderOptionsV2 &	operator= (const OrtTensorRTProviderOptionsV2 &other)

Public Attributes
int	device_id {0}
int	has_user_compute_stream {0}
void *	user_compute_stream {nullptr}
int	trt_max_partition_iterations {1000}
int	trt_min_subgraph_size {1}
size_t	trt_max_workspace_size {1 << 30}
int	trt_fp16_enable {0}
int	trt_int8_enable {0}
const char *	trt_int8_calibration_table_name {nullptr}
int	trt_int8_use_native_calibration_table {0}
int	trt_dla_enable {0}
int	trt_dla_core {0}
int	trt_dump_subgraphs {0}
int	trt_engine_cache_enable {0}
const char *	trt_engine_cache_path {nullptr}
int	trt_engine_decryption_enable {0}
const char *	trt_engine_decryption_lib_path {nullptr}
int	trt_force_sequential_engine_build {0}
int	trt_context_memory_sharing_enable {0}
int	trt_layer_norm_fp32_fallback {0}
int	trt_timing_cache_enable {0}
const char *	trt_timing_cache_path {nullptr}
int	trt_force_timing_cache {0}
int	trt_detailed_build_log {0}
int	trt_build_heuristics_enable {0}
int	trt_sparsity_enable {0}
int	trt_builder_optimization_level {3}
int	trt_auxiliary_streams {-1}
const char *	trt_tactic_sources {nullptr}
const char *	trt_extra_plugin_lib_paths {nullptr}
const char *	trt_profile_min_shapes {nullptr}
const char *	trt_profile_max_shapes {nullptr}
const char *	trt_profile_opt_shapes {nullptr}
int	trt_cuda_graph_enable {0}
int	trt_dump_ep_context_model {0}
const char *	trt_ep_context_file_path {nullptr}
int	trt_ep_context_embed_mode {0}
const char *	trt_engine_cache_prefix {nullptr}

Detailed Description

Options for the TensorRT provider that are passed to SessionOptionsAppendExecutionProvider_TensorRT_V2. Please note that this struct is similar to OrtTensorRTProviderOptions but only to be used internally. Going forward, new trt provider options are to be supported via this struct and usage of the publicly defined OrtTensorRTProviderOptions will be deprecated over time. User can only get the instance of OrtTensorRTProviderOptionsV2 via CreateTensorRTProviderOptions.

Definition at line 13 of file tensorrt_provider_options.h.

Member Function Documentation

OrtTensorRTProviderOptionsV2& OrtTensorRTProviderOptionsV2::operator=

(

const OrtTensorRTProviderOptionsV2 &

other

)

Member Data Documentation

int OrtTensorRTProviderOptionsV2::device_id {0}

Definition at line 16 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::has_user_compute_stream {0}

Definition at line 17 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_auxiliary_streams {-1}

Definition at line 43 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_build_heuristics_enable {0}

Definition at line 40 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_builder_optimization_level {3}

Definition at line 42 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_context_memory_sharing_enable {0}

Definition at line 34 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_cuda_graph_enable {0}

Definition at line 50 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_detailed_build_log {0}

Definition at line 39 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_dla_core {0}

Definition at line 27 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_dla_enable {0}

Definition at line 26 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_dump_ep_context_model {0}

Definition at line 68 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_dump_subgraphs {0}

Definition at line 28 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_engine_cache_enable {0}

Definition at line 29 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_engine_cache_path {nullptr}

Definition at line 30 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_engine_cache_prefix {nullptr}

Definition at line 72 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_engine_decryption_enable {0}

Definition at line 31 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_engine_decryption_lib_path {nullptr}

Definition at line 32 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_ep_context_embed_mode {0}

Definition at line 70 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_ep_context_file_path {nullptr}

Definition at line 69 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_extra_plugin_lib_paths {nullptr}

Definition at line 46 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_force_sequential_engine_build {0}

Definition at line 33 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_force_timing_cache {0}

Definition at line 38 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_fp16_enable {0}

Definition at line 22 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_int8_calibration_table_name {nullptr}

Definition at line 24 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_int8_enable {0}

Definition at line 23 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_int8_use_native_calibration_table {0}

Definition at line 25 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_layer_norm_fp32_fallback {0}

Definition at line 35 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_max_partition_iterations {1000}

Definition at line 19 of file tensorrt_provider_options.h.

size_t OrtTensorRTProviderOptionsV2::trt_max_workspace_size {1 << 30}

Definition at line 21 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_min_subgraph_size {1}

Definition at line 20 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_profile_max_shapes {nullptr}

Definition at line 48 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_profile_min_shapes {nullptr}

Definition at line 47 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_profile_opt_shapes {nullptr}

Definition at line 49 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_sparsity_enable {0}

Definition at line 41 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_tactic_sources {nullptr}

Definition at line 44 of file tensorrt_provider_options.h.

int OrtTensorRTProviderOptionsV2::trt_timing_cache_enable {0}

Definition at line 36 of file tensorrt_provider_options.h.

const char* OrtTensorRTProviderOptionsV2::trt_timing_cache_path {nullptr}

Definition at line 37 of file tensorrt_provider_options.h.

void* OrtTensorRTProviderOptionsV2::user_compute_stream {nullptr}

Definition at line 18 of file tensorrt_provider_options.h.

---

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

• onnxruntime/core/providers/tensorrt/tensorrt_provider_options.h