saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-03 03:58:54 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

additional options of NNAPI for ORT_PERF_TOOL (#10351)

Browse source 
* additional options of NNAPI for ORT_PERF_TOOL

* reuse current key '-i'

* fix

* fix

* _MSC_VER won't be defined when build with NDK

* fix

* fix
This commit is contained in:
wejoncy 2022-01-25 02:17:56 +08:00 committed by GitHub
parent 3dfadf9031
commit 5df15c5644
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 27 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
onnxruntime/test/perftest/command_args_parser.cc
View file

@ -65,8 +65,8 @@ namespace perftest {
"\t [Usage]: -e <provider_name> -i '<key1>|<value1> <key2>|<value2>'\n\n"
"\t [Example] [For OpenVINO EP] -e openvino -i \"device_type|CPU_FP32 enable_vpu_fast_compile|true num_of_threads|5 use_compiled_network|true blob_dump_path|\"<path>\"\"\n"
"\t [TensorRT only] [trt_max_partition_iterations]: Maximum iterations for TensorRT parser to get capability.\n"
"\t [TensorRT only] [trt_min_subgraph_size]: Minimum size of TensorRT subgraphs.\n"
"\t [TensorRT only] [trt_max_workspace_size]: Set TensorRT maximum workspace size in byte.\n"
"\t [TensorRT only] [trt_min_subgraph_size]: Minimum size of TensorRT subgraphs.\n"
"\t [TensorRT only] [trt_max_workspace_size]: Set TensorRT maximum workspace size in byte.\n"
"\t [TensorRT only] [trt_fp16_enable]: Enable TensorRT FP16 precision.\n"
"\t [TensorRT only] [trt_int8_enable]: Enable TensorRT INT8 precision.\n"
"\t [TensorRT only] [trt_int8_calibration_table_name]: Specify INT8 calibration table name.\n"
@ -79,6 +79,12 @@ namespace perftest {
"\t [TensorRT only] [trt_force_sequential_engine_build]: Force TensorRT engines to be built sequentially.\n"
"\t [Usage]: -e <provider_name> -i '<key1>|<value1> <key2>|<value2>'\n\n"
"\t [Example] [For TensorRT EP] -e tensorrt -i 'trt_fp16_enable|true trt_int8_enable|true trt_int8_calibration_table_name|calibration.flatbuffers trt_int8_use_native_calibration_table|false trt_force_sequential_engine_build|false'\n"
"\t [NNAPI only] [NNAPI_FLAG_USE_FP16]: Use fp16 relaxation in NNAPI EP..\n"
"\t [NNAPI only] [NNAPI_FLAG_USE_NCHW]: Use the NCHW layout in NNAPI EP.\n"
"\t [NNAPI only] [NNAPI_FLAG_CPU_DISABLED]: Prevent NNAPI from using CPU devices.\n"
"\t [NNAPI only] [NNAPI_FLAG_CPU_ONLY]: Using CPU only in NNAPI EP.\n"
"\t [Usage]: -e <provider_name> -i '<key1> <key2>'\n\n"
"\t [Example] [For NNAPI EP] -e nnapi -i \" NNAPI_FLAG_USE_FP16 NNAPI_FLAG_USE_NCHW NNAPI_FLAG_CPU_DISABLED \"\n"
"\t-h: help\n");
}
#ifdef _WIN32

20
onnxruntime/test/perftest/ort_test_session.cc
View file

@ -323,7 +323,25 @@ OnnxRuntimeTestSession::OnnxRuntimeTestSession(Ort::Env& env, std::random_device
#endif
} else if (provider_name == onnxruntime::kNnapiExecutionProvider) {
#ifdef USE_NNAPI
Ort::ThrowOnError(OrtSessionOptionsAppendExecutionProvider_Nnapi(session_options, 0));
uint32_t nnapi_flags = 0;
std::string ov_string = performance_test_config.run_config.ep_runtime_config_string;
std::istringstream ss(ov_string);
std::string key;
while (ss >> key) {
if (key == "NNAPI_FLAG_USE_FP16") {
nnapi_flags |= NNAPI_FLAG_USE_FP16;
} else if (key == "NNAPI_FLAG_USE_NCHW") {
nnapi_flags |= NNAPI_FLAG_USE_NCHW;
} else if (key == "NNAPI_FLAG_CPU_DISABLED") {
nnapi_flags |= NNAPI_FLAG_CPU_DISABLED;
} else if (key == "NNAPI_FLAG_CPU_ONLY") {
nnapi_flags |= NNAPI_FLAG_CPU_ONLY;
} else if (key.empty()) {
} else {
ORT_THROW("[ERROR] [NNAPI] wrong key type entered. Choose from the following runtime key options that are available for NNAPI. ['NNAPI_FLAG_USE_FP16', 'NNAPI_FLAG_USE_NCHW', 'NNAPI_FLAG_CPU_DISABLED', 'NNAPI_FLAG_CPU_ONLY'] \n");
}
}
Ort::ThrowOnError(OrtSessionOptionsAppendExecutionProvider_Nnapi(session_options, nnapi_flags));
#else
ORT_THROW("NNAPI is not supported in this build\n");
#endif