GNA Plugin: The config “scale factor” causes NN failed offloading to GNA.

[tea6329714]

 

 

[IntelSupport]

Hi tea6329714,

Thanks for reaching out.

We are still checking on this. For now, I observed a similar issue from my replication. I also tried experimenting with the scale value but apparently, only the 1.0 value fail with the script. Is there any difference between the NN1 and NN2 inference scripts?

Regards,

Aznie

[IntelSupport]

Hi tea6329714,

For this Runtime error when the custom Neural Network (NN2 test) config is set to a scale factor of 1.0, we need to escalate to our development team to verify if this is the bug or if there is a layer in NN2 that is yet to be supported by GNA plugin.

Please expect some delay from our development side and we will immediately inform you once we have the answer to this issue you are facing.

Regards,

Aznie

[IntelSupport]

Hi tea6329714,

Thanks for your patience and your validation results. 

We have escalated this information to the development team to verify the limitation in GNA. Once the validation results are verified, we will update our OpenVINO documentation.

We would like to thank you again for the test and results you have conducted.

Regards,

Aznie

[Megat_Intel]

Hi tea6329714,

This thread will no longer be monitored since this issue has been escalated to the development team for bug verification. If you need any additional information from Intel, please submit a new question.

Regards,

Megat