Hello all,
I have what is possibly a very simple question with regards to running AI applications on the RZV/2L evaluation board. I have been following the sample applications on the DRP-AI github (https://github.com/renesas-rz/rzv_drp-ai_tvm/tree/main/apps) and managed to successfully create my SDK, create the environment and convert the resnet model they provided and have all the necessary files ready for deploying the model on the evaluation board.
I followed the steps for transferring the model files onto the board and I can see that they were transferred successfully and I can view them on the board. However, when I ran the application I received the following error:
error while loading shared libraries: libstdc++.so.6: cannot open shared object file: No such file or directory
To quickly fix this, I manually copied over libstdc++.so.6 (saved under directory: /opt/poky/3.1.17/sysroots/aarch64-poky-linux/usr/lib/libstdc++.so.6) to the board but then got another error for a different shared library: libgcc_s.so.1.
I see that when I create the SDK it has a file structure which has these shared libraries (under the directory: /opt/poky/3.1.17/sysroots/aarch64-poky-linux/) which looks near identical to the rootfs file structure created when you build an image.
My question is, if you are looking to run AI applications do you need to copy the file structure created by the SDK to the SD card partition which contains the rootfs? as opposed to copying the file structure created when you build an image saved in the directory build/tmp/deploy/images/smarc-rzv2l/