Errors when executing demo using GPU

[GangXiong]

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!

[Iffa_Intel]

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

[GangXiong]

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.html 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!

[Iffa_Intel]

Ensure that you had carefull followed :https://docs.openvinotoolkit.org/latest/openvino_docs_install_guides_installing_openvino_windows.html#Install-GPU

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

[GangXiong]

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.

[Luis_at_Intel]

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

[GangXiong]

Hello Luis, Sorry for the late reply. I tested the solution you provided, it seemed that the problem remained there. I re-run the classification samples using model in FP32 precision, it went well when running on CPU while the error "Best_kernel.empty()(true) should be false, Cannot find a proper kernel with this arguments" would come if it was running on the GPU. Following gives the results: C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\demo>demo_squeezenet_download_convert_run.bat -d GPU target = GPU target_precision = FP32 Python 3.6.5 [setupvars.bat] OpenVINO environment initialized INTEL_OPENVINO_DIR is set to C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194 Python 3.6.5 ECHO is off. Requirement already satisfied: pyyaml in c:\users\admin\appdata\roaming\python\python36\site-packages (from -r C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\demo\..\open_model_zoo\tools\downloader\requirements.in (line 1)) (5.3.1) Requirement already satisfied: requests in c:\users\admin\appdata\roaming\python\python36\site-packages (from -r C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\demo\..\open_model_zoo\tools\downloader\requirements.in (line 2)) (2.18.4) Requirement already satisfied: urllib3<1.23,>=1.21.1 in c:\users\admin\appdata\roaming\python\python36\site-packages (from requests->-r C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\demo\..\open_model_zoo\tools\downloader\requirements.in (line 2)) (1.22) Requirement already satisfied: chardet<3.1.0,>=3.0.2 in c:\users\admin\appdata\roaming\python\python36\site-packages (from requests->-r C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\demo\..\open_model_zoo\tools\downloader\requirements.in (line 2)) (3.0.4) Requirement already satisfied: idna<2.7,>=2.5 in c:\users\admin\appdata\roaming\python\python36\site-packages (from requests->-r C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\demo\..\open_model_zoo\tools\downloader\requirements.in (line 2)) (2.6) Requirement already satisfied: certifi>=2017.4.17 in c:\users\admin\appdata\roaming\python\python36\site-packages (from requests->-r C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\demo\..\open_model_zoo\tools\downloader\requirements.in (line 2)) (2020.6.20) WARNING: You are using pip version 20.2.4; however, version 20.3.1 is available. You should consider upgrading via the 'c:\program files (x86)\python3.6.5\python.exe -m pip install --upgrade pip' command. Download public squeezenet1.1 model python "C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\open_model_zoo\tools\downloader\downloader.py" --name "squeezenet1.1" --output_dir "C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\models" --cache_dir "C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\cache" ################|| Downloading models ||################ ========== Retrieving C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\models\public\squeezenet1.1\squeezenet1.1.prototxt from the cache ========== Retrieving C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\models\public\squeezenet1.1\squeezenet1.1.caffemodel from the cache ################|| Post-processing ||################ ========== Replacing text in C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\models\public\squeezenet1.1\squeezenet1.1.prototxt squeezenet1.1 model downloading completed Waiting for 0 seconds, press a key to continue ... Target folder C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\ir\public\squeezenet1.1\FP32 already exists. Skipping IR generation with Model Optimizer. If you want to convert a model again, remove the entire C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\ir\public\squeezenet1.1\FP32 folder. Waiting for 0 seconds, press a key to continue ... ###############|| Generate VS solution for Inference Engine samples using cmake ||############### Waiting for 0 seconds, press a key to continue ... Creating Visual Studio 16 2019 x64 files in C:\Users\Admin\Documents\Intel\OpenVINO\inference_engine_samples_build... -- Selecting Windows SDK version 10.0.18362.0 to target Windows 10.0.14393. -- The C compiler identification is MSVC 19.27.29111.0 -- The CXX compiler identification is MSVC 19.27.29111.0 -- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.27.29110/bin/Hostx64/x64/cl.exe -- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.27.29110/bin/Hostx64/x64/cl.exe -- works -- Detecting C compiler ABI info -- Detecting C compiler ABI info - done -- Detecting C compile features -- Detecting C compile features - done -- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.27.29110/bin/Hostx64/x64/cl.exe -- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.27.29110/bin/Hostx64/x64/cl.exe -- works -- Detecting CXX compiler ABI info -- Detecting CXX compiler ABI info - done -- Detecting CXX compile features -- Detecting CXX compile features - done -- Found InferenceEngine: C:/Program Files (x86)/IntelSWTools/openvino_2020.3.194/deployment_tools/inference_engine/lib/intel64/Release/inference_engine.lib (Required is at least version "2.1") -- Configuring done -- Generating done -- Build files have been written to: C:/Users/Admin/Documents/Intel/OpenVINO/inference_engine_samples_build Waiting for 0 seconds, press a key to continue ... ###############|| Build Inference Engine samples using MS Visual Studio (MSBuild.exe) ||############### Waiting for 0 seconds, press a key to continue ... "C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Current\Bin\MSBuild.exe" Samples.sln /p:Configuration=Release /t:classification_sample_async /clp:ErrorsOnly /m Microsoft (R) Build Engine version 16.7.0+b89cb5fde for .NET Framework Copyright (C) Microsoft Corporation. All rights reserved. Waiting for 0 seconds, press a key to continue ... ###############|| Run Inference Engine classification sample ||############### Waiting for 0 seconds, press a key to continue ... 1 file(s) copied. classification_sample_async.exe -i "C:\Program Files (x86)\IntelSWTools\openvino_2020.3.194\deployment_tools\demo\car.png" -m "C:\Users\Admin\Documents\Intel\OpenVINO\openvino_models\ir\public\squeezenet1.1\FP32\squeezenet1.1.xml" -d GPU [ 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_2020.3.194\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 Error

[Iffa_Intel]

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