Specify path where nvinfer will write engine files?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

How can I specify where nvinfer will write engine files.?

I have just downloaded and updated my jetson nano to DS5.0GA and it seems that all the sample directories are only writeable by root and therefore nvinfer fails to save the model engine files.

We should be able to specify where these files are written. Either that or we have to change the access attributes to the sample dirs with chmod to allow user writes.

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) Nano, Xavier NX
• DeepStream Version 5.0GA

Wrong location for engine file generation
3

@jasonpgf2a

It seems location of the engine file should be the same as the model file it is generated from.

See in source code nvdsinfer_model_builder.cpp where the engine’s path is simply derived from the model path.

    /* Construct the suggested path for engine file. */
    suggestedPathName =
        modelPath + "_b" + std::to_string(initParams.maxBatchSize) + "_" +
        devId + "_" + networkMode2Str(networkMode) + ".engine";
    return engine;

So, re-configure and move your model to a non-root location may solve your problem.

4

Good idea. Thanks.