Install TensorFlow Python Wheel file on Jetpack 4.2 with Pip3

I recently upgraded Jetpack on TX2 to the brand new version 4.2. The wheel file I used previously on Jetpack 3.3 to install tensorflow-gpu from NVIDIA no longer works because Python is version 3.6 on Jetpack 4.2 Is there a new wheel file? How can I install this? Thanks.

me@me:~/Desktop/phil$ sudo pip3 install tensorflow-1.11.0-cp35-cp35m-linux_aarch64.whl 
[sudo] password for me: 
The directory '/home/me/.cache/pip/http' or its parent directory is not owned by the current user and the cache has been disabled. Please check the permissions and owner of that directory. If executing pip with sudo, you may want sudo's -H flag.
The directory '/home/me/.cache/pip' or its parent directory is not owned by the current user and caching wheels has been disabled. check the permissions and owner of that directory. If executing pip with sudo, you may want sudo's -H flag.
tensorflow-1.11.0-cp35-cp35m-linux_aarch64.whl is not a supported wheel on this platform.

Hi,

You can use this manual, https://docs.nvidia.com/deeplearning/dgx/install-tf-xavier/index.html.

Please use https://developer.download.nvidia.com/compute/redist/jp/v42/tensorflow-gpu/tensorflow_gpu-1.13.1+nv19.3-cp36-cp36m-linux_aarch64.whl (the latest one in https://developer.nvidia.com/embedded/downloads#?search=TensorFlow).

Hi,

Thanks for reporting this.

As vickyy mentioned, the latest package TF-1.13.1+nv19.3 is compatible for Xavier, TX2 and Nano.
We have updated this information in the pinned topic:
https://devtalk.nvidia.com/default/topic/1038957/jetson-tx2/tensorflow-for-jetson-tx2-/

Thanks.

When I follow the instructions from https://docs.nvidia.com/deeplearning/dgx/install-tf-xavier/index.html#troubleshooting, the pip3 install --pre --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v42 tensorflow-gpu TensorFlow install runs for a while and finally end in this error. Any ideas? TIA.

building ‘h5py.defs’ extension
creating build/temp.linux-aarch64-3.6
creating build/temp.linux-aarch64-3.6/tmp
creating build/temp.linux-aarch64-3.6/tmp/pip-build-z6jw4x8b
creating build/temp.linux-aarch64-3.6/tmp/pip-build-z6jw4x8b/h5py
creating build/temp.linux-aarch64-3.6/tmp/pip-build-z6jw4x8b/h5py/h5py
aarch64-linux-gnu-gcc -pthread -DNDEBUG -g -fwrapv -O2 -Wall -g -fstack-protector-strong -Wformat -Werror=format-security -Wdate-time -D_FORTIFY_SOURCE=2 -fPIC -DH5_USE_16_API -I./h5py -I/tmp/pip-build-z6jw4x8b/h5py/lzf -I/opt/local/include -I/usr/local/include -I/home/dennis/.local/lib/python3.6/site-packages/numpy/core/include -I/usr/include/python3.6m -c /tmp/pip-build-z6jw4x8b/h5py/h5py/defs.c -o build/temp.linux-aarch64-3.6/tmp/pip-build-z6jw4x8b/h5py/h5py/defs.o
In file included from /home/dennis/.local/lib/python3.6/site-packages/numpy/core/include/numpy/ndarraytypes.h:1822:0,
from /home/dennis/.local/lib/python3.6/site-packages/numpy/core/include/numpy/ndarrayobject.h:12,
from /home/dennis/.local/lib/python3.6/site-packages/numpy/core/include/numpy/arrayobject.h:4,
from /tmp/pip-build-z6jw4x8b/h5py/h5py/api_compat.h:26,
from /tmp/pip-build-z6jw4x8b/h5py/h5py/defs.c:637:
/home/dennis/.local/lib/python3.6/site-packages/numpy/core/include/numpy/npy_1_7_deprecated_api.h:17:2: warning: #warning "Using deprecated NumPy API, disable it with " “#define NPY_NO_DEPRECATED_API NPY_1_7_API_VERSION” [-Wcpp]
#warning "Using deprecated NumPy API, disable it with "
^~~~~~~
In file included from /tmp/pip-build-z6jw4x8b/h5py/h5py/defs.c:637:0:
/tmp/pip-build-z6jw4x8b/h5py/h5py/api_compat.h:27:10: fatal error: hdf5.h: No such file or directory
#include “hdf5.h”
^~~~~~~~
compilation terminated.
error: command ‘aarch64-linux-gnu-gcc’ failed with exit status 1

----------------------------------------

Command “/usr/bin/python3 -u -c “import setuptools, tokenize;file=’/tmp/pip-build-z6jw4x8b/h5py/setup.py’;f=getattr(tokenize, ‘open’, open)(file);code=f.read().replace(’\r\n’, ‘\n’);f.close();exec(compile(code, file, ‘exec’))” install --record /tmp/pip-8w89eamo-record/install-record.txt --single-version-externally-managed --compile --user --prefix=” failed with error code 1 in /tmp/pip-build-z6jw4x8b/h5py/

[Edit: This was the solution:
sudo apt-get install libhdf5-10
sudo apt-get install libhdf5-serial-dev
sudo apt-get install libhdf5-dev
sudo apt-get install libhdf5-cpp-11
find /usr -name hdf5.h
/usr/include/hdf5/serial/hdf5.h
export CPATH="/usr/include/hdf5/serial/"
]

Hi,

Try to install hdf5 first:

sudo apt-get install libhdf5-serial-dev hdf5-tools

Thanks.

What’s the meaning of nv19.5, nv19.4 nv19.3 ?
Where can I find those corresponding versions?

Thank you
Pei

I’m pretty sure those are different versions of Tensorflow for NVIDIA

Hi,
I met the same issue that I am using JetPack 4.2 and follow exact same steps from https://docs.nvidia.com/deeplearning/frameworks/install-tf-jetson-platform/index.html

And when I try to install the whls, I got errors:

ERROR: tensorflow_gpu-1.13.1+nv19.3-cp36-cp36m-linux_aarch64.whl is not a supported wheel on this platform.

It’s on Xavier. Any hints?

Hi, Klein92

Which command do you execute? Is this one?

sudo pip3 install --pre --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v42 tensorflow-gpu

If yes, could you reboot the system to see if helps?

Thanks.