Path for GPU Extension

[rico22]

Hi, everyone. I am studying the tutorial_classification_sample on JupyterLab. It runs on CPU, but I want to run it on GPU. I can see that they use this line (cpu_extension_path = os.path.expanduser("~")+"/inference_engine_samples/intel64/Release/lib/libcpu_extension.so") for the CPU implementation. What should I write for the GPU implementation? Furthermore, where is this 'inference_engine_samples' file JupyterLab? Thank you for your time.

[Hairul_Intel]

Hi rico22,

Thank you for reaching out to us.

For your information, I wasn't able to find "tutorial_classification_sample" from OpenVINO™ Notebooks. Could you provide more information regarding the source repository of your sample?

On another note, I've successfully ran the Hello Image Classification tutorial with GPU from OpenVINO™ Notebooks. To run the tutorial with GPU, you will need to change the following command:

compiled_model = ie.compile_model(model=model, device_name="GPU")

Following the OpenVINO™ Notebooks Installation Guide, you can find the GPU extension path (openvino_intel_gpu_plugin.dll) from the following directory:

openvino_env\Lib\site-packages\openvino\libs

Regards,

Hairul

[rico22]

Hi, Hairul. Thank you for your reply. I am working on the Intel DevCloud platform. Here's the link for the tutorial: https://notebooks.edge.devcloud.intel.com/user/u174279/lab/tree/Reference-samples/iot-devcloud/openvino-dev-latest/tutorials/classification/tutorial_classification_sample.ipynb 

I would like to do inferencing on GPU, but I don't know what to write for GPU implementation. For CPU implementation, this line was written: cpu_extension_path = os.path.expanduser("~")+"/inference_engine_samples/intel64/Release/lib/libcpu_extension.so"

I would also like to know the location of the 'inference_engine_samples' file in Intel DevCloud's JupterLab environment. Thank you.

[JesusE_Intel · Accepted solution]

Hi rico22,

The classification tutorial only supports CPU devices as the inference is running on the same CPU that is running the Jupyter notebook. I recomend taking a look at Reference-samples/iot-devcloud/openvino-dev-latest/developer-samples/python/benchmarkApp-python/BenchmarkApp.ipynb for additional details on how to submit job to a node with a GPU available.

1.2. Introduction

The purpose of this tutorial is to examine a sample application that was created using the Intel® Distribution of Open Visual Inference & Neural Network Optimization (OpenVINO™) toolkit. This tutorial will go step-by-step through the necessary steps to demonstrate object classification on images and batches of images. Classification is performed using a pre-trained network and running it using the Intel® Distribution of OpenVINO™ toolkit Inference Engine. Inference will be executed using the same CPU(s) running this Jupyter* Notebook.

Hope this helps!

Regards,

Jesus

[rico22]

Thank you for your response. I appreciate it. Have a nice day. 

[JesusE_Intel]

If you need any additional information, please submit a new question as this thread will no longer be monitored.

[IshaNCB]

Hi I am getting Error when I am trying to accessing tiny_yolo_v4.ipynb