Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Novice
140 Views

Errors when executing demo using GPU

Hello,  I have installed openvino on my computer and the verification scripts are executed without any errors.  Then I ran the image classification sample application with the Caffe* Squeezenet1.1 model using the following commands:

classification_sample_async.exe -i "C:\Program Files (x86)\IntelSWTools\openvino\deployment_tools\demo\car.png" -m "C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\ir\public\squeezenet1.1\FP16\squeezenet1.1.xml" -d GPU

The following error came:

[ INFO ] InferenceEngine:
        API version ............ 2.1
        Build .................. 2020.3.0-3467-15f2c61a-releases/2020/3
        Description ....... API
[ INFO ] Parsing input parameters
[ INFO ] Parsing input parameters
[ INFO ] Files were added: 1
[ INFO ]     C:\Program Files (x86)\IntelSWTools\openvino\deployment_tools\demo\car.png
[ INFO ] Creating Inference Engine
        GPU
        clDNNPlugin version ......... 2.1
        Build ........... 2020.3.0-3467-15f2c61a-releases/2020/3

[ INFO ] Loading network files
[ INFO ] Preparing input blobs
[ WARNING ] Image is resized from (787, 259) to (227, 227)
[ INFO ] Batch size is 1
[ INFO ] Loading model to the device
[ ERROR ] Error has occured for: pooling:pool1
Best_kernel.empty()(true) should be false
Cannot find a proper kernel with this arguments

 There aren't any errors when I ran the image classification application using CPU (-d CPU).  My environment is like this:

CPU: Intel Xeon E3-1515M V5

GPU: Iris(TM) Pro Graphics P580 (Driver version: 21.20.16.4627)

OS: Windows enterprise 2016 LTSB  (version 1607)

Openvino version: 2020.3.194

Is there anyone can help with my problem? Thanks in advance!

0 Kudos
6 Replies
Highlighted
Moderator
122 Views

Greetings,

First and foremost, you need to ensure the file type of the sample is correct. Take a closer look on the attached photo where I had successfully run it with GPU.

 

 

Sincerely,

Iffa

0 Kudos
Highlighted
Novice
118 Views

Hello Iffa, Thanks for your kind reply. I followed the instruction of the last step (Run the Image Classification Sample Application) in the following link: https://docs.openvinotoolkit.org/latest/openvino_docs_install_guides_installing_openvino_windows.htm... since I've successfully run the verification scripts, I got the .exe file of the sample and I ran the same sample successfully using CPU. So I don't think the file type is the problem. Is it possible that the hardware or the os version don't support the opevino application? but I found nothing in the official documents. Hope you can help further, thank you very much!
0 Kudos
Highlighted
Moderator
113 Views

Ensure that you had carefull followed :https://docs.openvinotoolkit.org/latest/openvino_docs_install_guides_installing_openvino_windows.htm...

 

This should ensure that your GPU is supported & all pre-requisite required for GPU operations are installed.

I'm running Windows 10, i7 with Intel(R) UHD Graphic 620. Check yours and cross check with the tutorial I had attached in the link above.

 

Sincerely,

Iffa

0 Kudos
Highlighted
Novice
108 Views

Hi Iffa: I have re-checked the instruction and my system configuration. My CPU is Intel Xeon E3-1515M V5, GPU is Intel Iris Pro Graphics P580 with driver version 20.20.16.467. Is it possible that the P580 GPU don't support openvino? since I have also successfully ran the program on my i7 CPU and HD660 GPU.
0 Kudos
Highlighted
Moderator
86 Views

Hi @GangXiong,

Thanks for the provided information. According to the hardware components listed under Intel® Processor Graphics section here, the OpenVINO™ toolkit seems to support Intel® Iris® Pro graphics:

  • "Intel Xeon processor with Intel Iris Pro graphics and Intel HD Graphics (excluding the e5 family which does not include graphics)".

With this in mind, please try the following and share the results:

  • Re-run classification sample using model in FP32 precision instead of FP16
    • Simplest way, edit demo_squeezenet_download_convert_run.sh change line #47 to FP32 instead of FP16
  • Run the other verification script with GPU (demo_security_barrier_camera.sh) and share results of both FP16 and FP32
    • Edit demo_security_barrier_camera.sh change line #140 to FP32 instead of FP16

 

Best Regards,

Luis

0 Kudos
Highlighted
Moderator
51 Views

Greetings,


Intel will no longer monitor this thread since we have provided a solution. If you need any additional information from Intel, please submit a new question


Sincerely,

Iffa


0 Kudos