saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-06-06 00:03:22 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Unified OV compile_model API in OVEP (#20700)

Browse source 
### Description
Have a unified API in OVEP that pass the ONNX graph proto from ORT to OV
for compilation


### Motivation and Context
The earlier implementation used two different flows when onnx model path
is present vs model laoded from memory.
The former directly passed the onnx model path to OV when the graph is
fully supported by EP. While the latter pass the ORT model proto to OV.

This cause a difference in results when ORT optimizations are enabled.
This PR address this issue.
This commit is contained in:
Preetha Veeramalai 2024-05-20 10:20:28 -07:00 committed by GitHub
parent 036fcd93d4
commit ebed2c3785
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 7 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
onnxruntime/core/providers/openvino/backends/basic_backend.cc
View file

@ -90,11 +90,11 @@ BasicBackend::BasicBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
device_config,
subgraph_context_.subgraph_name);
ie_cnn_network_ = exe_network_.Get().get_runtime_model();
} else if (!subgraph_context_.has_dynamic_input_shape &&
global_context_.onnx_model_path_name.find(".onnx") != std::string ::npos) {
} else if (!subgraph_context_.has_dynamic_input_shape) {
// Inputs with static dimenstions
std::string prec_str = (global_context_.precision_str != "ACCURACY") ? global_context_.precision_str : global_context_.model_precision;
exe_network_ = global_context_.ie_core.CompileModel(global_context_.onnx_model_path_name,
const std::string model = model_proto.SerializeAsString();
exe_network_ = global_context_.ie_core.CompileModel(model,
hw_target,
prec_str,
global_context_.cache_dir,

6
onnxruntime/core/providers/openvino/ov_interface.cc
View file

@ -88,7 +88,7 @@ OVExeNetwork OVCore::CompileModel(std::shared_ptr<const OVNetwork>& ie_cnn_netwo
}
}
OVExeNetwork OVCore::CompileModel(const std::string onnx_model_path,
OVExeNetwork OVCore::CompileModel(const std::string& onnx_model,
std::string& hw_target,
std::string precision,
std::string cache_dir,
@ -97,13 +97,13 @@ OVExeNetwork OVCore::CompileModel(const std::string onnx_model_path,
ov::CompiledModel obj;
try {
if (hw_target == "AUTO:GPU,CPU") {
obj = oe.compile_model(onnx_model_path,
obj = oe.compile_model(onnx_model, ov::Tensor(),
"AUTO",
ov::device::priorities("GPU", "CPU"),
ov::device::properties("GPU", {ov::cache_dir(cache_dir),
ov::hint::inference_precision(precision)}));
} else {
obj = oe.compile_model(onnx_model_path, hw_target, device_config);
obj = oe.compile_model(onnx_model, ov::Tensor(), hw_target, device_config);
}
#ifndef NDEBUG
printDebugInfo(obj);

2
onnxruntime/core/providers/openvino/ov_interface.h
View file

@ -44,7 +44,7 @@ class OVCore {
std::string& hw_target,
ov::AnyMap& device_config,
std::string name);
OVExeNetwork CompileModel(const std::string onnx_model_path,
OVExeNetwork CompileModel(const std::string& onnx_model,
std::string& hw_target,
std::string precision,
std::string cache_dir,