Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Ort v1.15.0 ov changes customer a #314

Open
wants to merge 9 commits into
base: ort_v1.15.0_ov_changes
Choose a base branch
from

Conversation

preetha-intel
Copy link

Description

Changes part of Customer A release - 1.15.0

Motivation and Context

Enabled features based on client requirements

This PR is created to review for SDLE.

@@ -88,7 +88,7 @@ CreateOVModel(const ONNX_NAMESPACE::ModelProto& model_proto, const GlobalContext
size_t index = results.size() - 1;

for (auto it = results.rbegin(); it != results.rend(); ++it) {
if (auto const_node = std::dynamic_pointer_cast<ngraph::op::Constant>((*it)->input_value(0).get_node_shared_ptr())) {
if (auto const_node = std::dynamic_pointer_cast<ov::op::v0::Constant>((*it)->input_value(0).get_node_shared_ptr())) {

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are we using OV API 1.0 or 2.0

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These are with API 2.0 as we are supporting only the latest three release of OV.

@@ -57,7 +57,7 @@ BasicBackend::BasicBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
LOGS_DEFAULT(INFO) << log_tag << "Loaded model to the plugin";
} else {
#if defined(OPENVINO_2023_0)
if (subgraph_context.precision != InferenceEngine::Precision::FP16) {
if (!subgraph_context_.has_dynamic_input_shape && dev_prec!="CPU_FP16") {

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Has a bug been raised with OV team for CPU FP 16 Precision

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not yet. Shall raise one.

@@ -257,7 +252,7 @@ void BackendManager::Compute(OrtKernelContext* context) {
}
#endif
bool use_dynamic_backend = true;
if (GetGlobalContext().enable_dynamic_shapes && subgraph_context_.has_dynamic_input_shape &&
if (subgraph_context_.has_dynamic_input_shape &&

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For NPU we still need dynamic backend so maybe this logic wont be suitable going forward

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This logic is only skipping whether enable_dynamic_shapes runtime option is provided. According to this logic, if a model has dynamic shaped input it will be handled with dynamic backend.

std::pair<std::string, ov::Any> device_property;
device_property = std::make_pair("PLUGIN_THROTTLE", "1");
device_config.emplace(ov::device::properties("GPU_CONFIG_KEY", device_property));
// device_config[GPU_CONFIG_KEY(PLUGIN_THROTTLE)] = "1";

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Remove commented code

Copy link

@sfatimar sfatimar left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please look at inline comments

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants