Could not run Myriad X (named UP AI CORE X) on Windows 10

[yamaken]

(I post same question again because the forum is closing.)

https://ncsforum.movidius.com/discussion/1705/about-ncdeviceopen-799-failed-connection-to-device-1-7-2-ma2480-with-error-5#latest

I faced an Error with Windows 10 OS.

I tried running my Myriad X (named UP AI CORE X) with the tutorial below.

https://software.intel.com/en-us/articles/get-started-with-neural-compute-stick

but, at "Step 4. Test the installation", an error occured.

* the device could be seen as Movidius Myriad X on device manager.

* after the error occurd, the device can not be seen.

My OpenVINO toolkit version is "2018 R5.0.1 for Windows"

(Without "-d MYRIAD" option , there is no probrem to run samples.)

Best Regards.

I show a part of the Error

classification_sample.exe -i "C:\Program Files (x86)\IntelSWTools\openvino\deployment_tools\demo\car.png" -m "C:\Users(myname)\Documents\Intel\OpenVINO\openvino_models\ir\FP16\classification\squeezenet\1.1\caffe\squeezenet1.1.xml" -d MYRIAD

[ INFO ] InferenceEngine:

API version ............ 1.6

Build .................. 23224

[ INFO ] Parsing input parameters

[ INFO ] Files were added: 1

[ INFO ] C:\Program Files (x86)\IntelSWTools\openvino\deployment_tools\demo\car.png

[ INFO ] Loading plugin

API version ............ 1.6

Build .................. 23224

Description ....... myriadPlugin

[ INFO ] Loading network files:

C:\Users(myname)\Documents\Intel\OpenVINO\openvino_models\ir\FP16\classification\squeezenet\1.1\caffe\squeezenet1.1.xml

C:\Users(myname)\Documents\Intel\OpenVINO\openvino_models\ir\FP16\classification\squeezenet\1.1\caffe\squeezenet1.1.bin

[ INFO ] Preparing input blobs

[ WARNING ] Image is resized from (787, 259) to (227, 227)

[ INFO ] Batch size is 1

[ INFO ] Preparing output blobs

[ INFO ] Loading model to the plugin

[35mE: [ncAPI] [ 0] ncDeviceOpen:799 Failed connection to device (3.1-ma2480) with error 5[0m

Error

[JesusE_Intel]

Hi yamaken,

Did you get a chance to try the demo with the latest OpenVINO toolkit release 2019 R1.1?

You can download the latest release at the following link:

https://software.intel.com/en-us/openvino-toolkit/choose-download

Regards,

Jesus

[yamaken]

Hi @Intel_Jesus​ 

I upgraded OpenVINO toolkit Version into  release 2019 R1.1 from the link .

("version.txt" in "\openvino\inference_engine\" says its version is below:

Wed Apr 24 18:53:29 MSK 2019

28dfbfdd28954c4dfd2f94403dd8dfc1f411038b

releases_2019_R1.1_InferenceEngine_23780

)

I tested same tutorial , and same error occured...

I put the log as "errorlog.txt" on this board. some part of them written in Japanese :(

I did the demo on command prompt as administrator.

Regards,

yamaken

[yamaken]

I found the sample program running on Windows PowerShell without any error.

"target_precision" on PowerShell is "FP32" against Command Prompt's One is "FP16".

(I put the log as "pslog.txt". )

I want to know why program doesn't ​work on Command Prompt.

Regards,

[Sahira_Intel]

Hi Yamaken,

I ran through a fresh OpenVINO installation on my system running Windows 10, and I did not see any of the above errors. Since you said that the device shows up in the device manager, but then goes away after you run an example, you may have a faulty unit.

In regards to the sample running on PowerShell, since the target precision is FP32, that means that it's running on the CPU. The MYRIAD VPU does not support FP32. It looks like your command to run the sample is .\demo_squeezenet_download_convert_run.bat ?d MYRIAD

The "?" should be -d MYRIAD. Your system likely ignored that parameter and ran on the CPU - hence the successful execution and target precision being FP32.

Please let me know if you have any further questions.

Best Regards,

Sahira

[yamaken]

Hi @Sahira_Intel​ 

As you said, I've run the demo program on CPU... 

Thank you for your pointing out an easy mistake.

I failed to run the program on PowerShell, and the error is same as occured on CommandPrompt.

Are there anything that I shold investigate for what is wrong?  

Something special environment of mine:

• My unit is "UP AI CORE X "​(https://up-board.org/ai-core-x/)
• I convert miniPCIe→PCI-Express using 3rd party unit (https://www.kuroutoshikou.com/product/interface/converter_board/mpcie-pciex1/)

[Sahira_Intel]

Hi Yamaken,

I would suggest contacting AAEON for more information about the compatibility of that 3rd party conversion board - that might be the cause of the error.

Best Regards,

Sahira