Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.

[X_LINK_ERROR] The inferring error of two similar models.


Hello. I'm studying Deep learning in Republic of Korea.

I have some problems about NCS2.

I've a similar two models that is configured by simple CNN.

The two models are very similar in structure. The only difference is that the number of classes is different.

One model works very well, Both result of inference and time are perfect.

but, the other model doesn't work. Model spends a lot of time finally can't make a result.

If the model can not infer, X_LINK_ERROR is displayed on the terminal. The message displayed is USB_READ_REL_REQ.

Some others said it's a problem of extension USB cable. 

If so, is it not anomalous to produce results for either model?

It is strange that one model produces good results, and the other model consumes time that can not yield any results.

Which part can I suspect?

0 Kudos
2 Replies
New Contributor I

You may suffer the X_LINK_ERROR error even with the so-called "good" model after 18~24 hours long running torture test. It depends on the NCS2 workload from my experience. If you can feed "frame" data as soon as you can (i.e. > 30 fps, async mode, num_request = 2) to increase the NCS2 workload, and then the X_LINK_ERROR might be occurred earlier. If you plugged your NCS2 to a desktop computer, perhaps you can try to plug it directly to the motherboard USB port w/o using USB extension cable. Changing the USB extension cable doesn't work for me and it wastes my money. Using a powerful AC-powered USB hub fixed the X_LINK_ERROR I encountered. Just FYR. Good luck.

0 Kudos

Dear. lilohuang

Unfortunately, My problem does not seem to be related to the cable.

I already use an extension cable that is fully powered.

Also, one model works normally while the other has never worked successfully.

If it was a cable problem, no model would have been successful.

Thank you for your reply

0 Kudos