Hi
I've converted a Yolov7 model using the blob converter tool. When I go to run the new model the system crashes with:
184430102155AB0F00] [1.6] [1.867] [system] [info] Temperatures - Average: 35.17 °C, CSS: 36.06 °C, MSS 35.11 °C, UPA: 34.64 °C, DSS: 34.88 °C
[184430102155AB0F00] [1.6] [1.867] [system] [info] Cpu Usage - LeonOS 83.86%, LeonRT: 56.66%
no cam sync skipping frame 1 right frame is nullptr
no cam sync skipping frame 3 right frame is nullptr
[184430102155AB0F00] [1.6] [3.869] [system] [critical] Fatal error. Please report to developers. Log: 'Fatal error on MSS CPU: trap: 00, address: 00000000' '0'
[184430102155AB0F00] [1.6] [4.052] [system] [info] Memory Usage - DDR: 104.44 / 340.61 MiB, CMX: 2.48 / 2.50 MiB, LeonOS Heap: 55.11 / 77.46 MiB, LeonRT Heap: 34.28 / 41.34 MiB
[184430102155AB0F00] [1.6] [4.052] [system] [info] Temperatures - Average: 35.88 °C, CSS: 36.77 °C, MSS 35.59 °C, UPA: 35.83 °C, DSS: 35.35 °C
[184430102155AB0F00] [1.6] [4.053] [system] [info] Cpu Usage - LeonOS 59.54%, LeonRT: 100.00%
[184430102155AB0F00] [1.6] [5.833] [system] [info] Memory Usage - DDR: 104.44 / 340.61 MiB, CMX: 2.48 / 2.50 MiB, LeonOS Heap: 55.11 / 77.46 MiB, LeonRT Heap: 34.28 / 41.34 MiB
[184430102155AB0F00] [1.6] [5.833] [system] [info] Temperatures - Average: 36.30 °C, CSS: 37.48 °C, MSS 35.83 °C, UPA: 36.53 °C, DSS: 35.35 °C
[184430102155AB0F00] [1.6] [5.833] [system] [info] Cpu Usage - LeonOS 8.18%, LeonRT: 100.00%
terminate called after throwing an instance of 'std::runtime_error'
what(): Communication exception - possible device error/misconfiguration. Original message 'Couldn't read data from stream: 'detections' (X_LINK_ERROR)'
Now if I swap back to a model from the model zoo, it works. again.
I think the issue is that the LeonRT is reaching 100% for some reason and locking up the pipeline. When I update my depthai-core source to the head of develop, I get additional info:
[184430102155AB0F00] [1.6] [5.315] [system] [info] Cpu Usage - LeonOS 15.52%, LeonRT: 100.00%
[184430102155AB0F00] [1.6] [6.906] [system] [critical] Fatal error. Please report to developers. Log: 'Fatal error on MSS CPU: trap: 00, address: 00000000' '0'
[184430102155AB0F00] [1.6] [1718592572.516] [host] [debug] Timesync thread exception caught: Couldn't read data from stream: '__timesync' (X_LINK_ERROR)
[184430102155AB0F00] [1.6] [1718592572.516] [host] [debug] Log thread exception caught: Couldn't read data from stream: '__log' (X_LINK_ERROR)
terminate called after throwing an instance of 'std::runtime_error'
what(): Communication exception - possible device error/misconfiguration. Original message 'Couldn't read data from stream: 'detections' (X_LINK_ERROR)'
I'm using most of the same parameters, only difference is I'm using new anchor boxes that the yolov7 determined for me based on my dataset and my number of classes is not including background class.
Compared to the documentation, the blob converter didn't provide a .json file. I assume that requirement is no longer needed. So I have just uploaded the blob provided.
Let me know if there is any more information needed.