Openvino 2022 - using multiple NCS2 asynchrounously

[gaudibert]

Hello,

I'm trying to update my stack to Openvino 2022 to benefit from the latest news. And I'm using this update to gain a little performance by trying to use multiple NCS2 asynchrounously.

However I'm a bit stuck despite reading quite some documentation on your website and the community. I'm finding a lot of examples of this on 2021 version but I can't manage to pass it to 2022 2.0 API.

The closest example I found is this : https://github.com/openvinotoolkit/openvino/tree/master/samples/cpp/classification_sample_async but it doesn't manage multiple devices.

Would you have a very basic example of a model loaded and called by batch on more than 1 NCS2 ? That would be immensely helpful for me.

Best regards,

Greg

[Wan_Intel]

Hi Gaudibert,

Thanks for reaching out to us.

For your information, you can use multiple Intel® Neural Compute Stick 2 asynchronously as shown as follows:

1.   Run Hello Query Device Python Sample script which is located in the following directory:

cd <INSTALL_DIR>\samples\python\hello_query_device

python hello_query_device.py

 

2.   Use Multi-device plugin with the following command:

python classification_sample_async.py -m "<path_to_model>\alexnet.xml" -i "path_to_input_1" "path_to_input_2" -d MULTI:MYRIAD.5.2-ma2480,MYRIAD.5.3-ma2480

For more information, please refer to Running on multiple devices simultaneously and Multi-Device Plugin with the Intel® Neural Compute Stick 2.

Regards,

Wan

[gaudibert]

Hello Wan,

Thank you for your quick reply. That is exactly what I needed.

I should have taken a look inside the openvino repository and not only online, sorry about that !

Cheers,
Greg

[Wan_Intel]

Hi Gaudibert,

Thanks for your question.

This thread will no longer be monitored since we have provided information. 

If you need any additional information from Intel, please submit a new question.

Best regards,

Wan