search

NVIDIA / Deep-Learning-Accelerator-SW

NVIDIA DLA-SW, the recipes and tools for running deep learning workloads on NVIDIA DLA cores for inference applications.
Other
178 stars 14 forks source link

ONNX GraphSurgeon is not available on aarch64 platforms #7

Closed leimao closed 1 year ago

leimao commented 1 year ago

ONNX GraphSurgeon is not available on aarch64 platforms and therefore the requirement installation is not completely successful on Xavier/Orin.

nvoliver commented 1 year ago

Thanks, we will clarify in the next update that the Python scripts are intended to be run on host.

leimao commented 1 year ago

Thanks, we will clarify in the next update that the Python scripts are intended to be run on host.

Thanks @nvoliver. In fact, I download the source code of ONNX GraphSurgeon, built it on aarch64 platforms, and it just worked fine. You may just request the TensorRT team to release ONNX GraphSurgeon for aarch64 platforms starting from the next release, which should be something straightforward to do.

nvoliver commented 1 year ago

Updated the instructions in #8.