update your snpe setup docurment ( https://developer.qualcomm.com/docs/snpe/setup.html )
That is too old.
"Currently the SNPE SDK development environment is limited to Ubuntu, specifically version 14.04."
I did install ubuntu 14.04 because this statement, but I failed to copy libatomic.so.1.
ubuntu 14.04 support only aarch64-4.8.x library
ref : https://developer.qualcomm.com/forum/qdn-forums/software/symphony-system...
2. The Caffe2 setup document is not helpful
I did install caffe2 in non-conda env with this document states
ref : https://caffe2.ai/docs/getting-started.html?platform=ubuntu&configuratio...
git clone https://github.com/pytorch/pytorch.git && cd pytorchgit submodule update --init --recursivepython setup.py install
Environment setup for Caffe2Using the SNPE SDK with Caffe2 models requires a valid Caffe2 installation. (see Caffe and Caffe2 Setup.)Go to $SNPE_ROOT and run the following script to setup the SNPE SDK environment. The location of the Caffe2 directory must be provided.source bin/envsetup.sh -f $CAFFE2_DIRwhere $CAFFE2_DIR is the path to the Caffe2 installation.The script sets up the following environment variables:SNPE_ROOT: root directory of the SNPE SDK installationCAFFE2_HOME: root directory of the Caffe2 installation you provided ($CAFFE2_DIR)The script also updates PATH, LD_LIBRARY_PATH, and PYTHONPATH.
3. Did you install TensorFlow in ubuntu 14.04 ?
ref : https://www.tensorflow.org/install/
Install TensorFlowTensorFlow is tested and supported on the following 64-bit systems:Ubuntu 16.04 or laterWindows 7 or latermacOS 10.12.6 (Sierra) or later (no GPU support)Raspbian 9.0 or later
The setup documentation is terrible. Keep getting incompatible packages, too old versions, garbage. I would be simple to just give us a script or conda environment file. The way it is, it's an headache, and just gives Qualcomm a bad image.