dla standalone loadable file inspect

[lix19937]

Is there a tool that can inspect dla standalone loadable file, i want to get inner nodes/layers info ?

I use trtexec load dla loadble file wrong as follow:

[09/04/2024-13:51:41] [I] Engine loaded in 0.0123033 sec.
[09/04/2024-13:51:46] [I] [TRT] Loaded engine size: 0 MiB
[09/04/2024-13:51:47] [E] Error[9]: Cannot deserialize serialized engine built with EngineCapability::kDLA_STANDALONE, use cuDLA APIs instead.
[09/04/2024-13:51:47] [E] Error[4]: [runtime.cpp::deserializeCudaEngine::65] Error Code 4: Internal Error (Engine deserialization failed.)
[09/04/2024-13:51:47] [E] Engine deserialization failed
[09/04/2024-13:51:47] [E] Got invalid engine!
[09/04/2024-13:51:47] [E] Inference set up failed
&&&& FAILED TensorRT.trtexec [TensorRT v8510] # trtexec --verbose --loadEngine=dla_fusion.dla

I know use https://github.com/NVIDIA-AI-IOT/cuDLA-samples/blob/main/src/cudla_context_standalone.cpp#L137 can load file, but i want know more info about dla model.

https://docs.nvidia.com/cuda/archive/11.7.1/cudla-api/index.html#structcudlaTask_1f61768f306610f76fe18ca2ef0bafd78

only get in-out tensors info (shape type)

https://developer.nvidia.com/blog/deploying-yolov5-on-nvidia-jetson-orin-with-cudla-quantization-aware-training-to-inference/
https://zhuanlan.zhihu.com/p/656172954

The same issue see https://github.com/NVIDIA/TensorRT/issues/4106

[lix19937]

@lynettez can you give some advice ?

[lix19937]

@dusty-nv @rbonghi @sujitbiswas @wkelongws

[11061995]

DID THE ISSUE SOLVED?

[lix19937]

NO @11061995