Other Parts Discussed in Thread: AM67A
Tool/software:
Hello,
I'm starting on a project targeting the AM67A. Using the prebuilt wic file, the Edge AI gallery functions as expected. But when I build the edgeai Yocto image from scratch (host: Ubuntu 22.04, building sdk 11.0), the Edge AI gallery examples crash. echo $EDGEAI_VERSION returns 11.0.
Here's a call to one of the out of the box examples and the resulting error:
root@j722s-evm:/opt/edgeai-gst-apps/apps_python# python3 app_edgeai.py ../configs/object_detection.yaml
libtidl_onnxrt_EP loaded 0x381c9c70
Final number of subgraphs created are : 1, - Offloaded Nodes - 283, Total Nodes - 283
APP: Init ... !!!
338.949354 s: MEM: Init ... !!!
338.949438 s: MEM: Initialized DMA HEAP (fd=5) !!!
338.949640 s: MEM: Init ... Done !!!
338.949669 s: IPC: Init ... !!!
339.005974 s: IPC: Init ... Done !!!
REMOTE_SERVICE: Init ... !!!
REMOTE_SERVICE: Init ... Done !!!
339.012924 s: GTC Frequency = 200 MHz
APP: Init ... Done !!!
339.013124 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_ERROR
339.013146 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_WARNING
339.013157 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_INFO
339.014156 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-0
339.014520 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-1
339.014799 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-2
339.015051 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-3
339.015080 s: VX_ZONE_INFO: [tivxInitLocal:202] Initialization Done !!!
339.015119 s: VX_ZONE_INFO: Globally Disabled VX_ZONE_INFO
339.029888 s: VX_ZONE_ERROR: [ownContextSendCmd:1001] Command ack message returned failure cmd_status: -1
339.029936 s: VX_ZONE_ERROR: [ownNodeKernelInit:704] Target kernel, TIVX_CMD_NODE_CREATE failed for node node_84
339.029949 s: VX_ZONE_ERROR: [ownNodeKernelInit:705] Please be sure the target callbacks have been registered for this core
339.029961 s: VX_ZONE_ERROR: [ownNodeKernelInit:706] If the target callbacks have been registered, please ensure no errors are occurring within the create callback of this kernel
339.029976 s: VX_ZONE_ERROR: [ownGraphNodeKernelInit:793] kernel init for node 0, kernel com.ti.tidl:1:2 ... failed !!!
339.030029 s: VX_ZONE_ERROR: [ TIDL subgraph dets ] Node kernel init failed
339.030043 s: VX_ZONE_ERROR: [ TIDL subgraph dets ] Graph verify failed
Traceback (most recent call last):
File "/opt/edgeai-gst-apps/apps_python/app_edgeai.py", line 67, in <module>
main(sys.argv)
File "/opt/edgeai-gst-apps/apps_python/app_edgeai.py", line 46, in main
demo = EdgeAIDemo(config)
^^^^^^^^^^^^^^^^^^
File "/opt/edgeai-gst-apps/apps_python/edge_ai_class.py", line 108, in __init__
model_obj.create_runtime()
File "/usr/lib/python3.12/site-packages/edgeai_dl_inferer.py", line 315, in create_runtime
self.run_time = RunTime(self.artifacts,
^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.12/site-packages/edgeai_dl_inferer.py", line 170, in __init__
self.interpreter = _onnxruntime.InferenceSession(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.12/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 387, in __init__
self._create_inference_session(providers, provider_options, disabled_optimizers)
File "/usr/lib/python3.12/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 439, in _create_inference_session
sess.initialize_session(providers, provider_options, disabled_optimizers)
onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Create state function failed. Return value:-1
I trained a YOLO network using the edgeai-modelmaker repo on branch r11.0 to make sure it wasn't a version compatibility issue - I got about the same error passing an array of 0's into the network. Can you please advise on what may be going on? I see there's now also an SDK 11.01.07 configuration, but wasn't sure if that would work with the r11.0 branch of edgeai-tensorlab.
