search

jkjung-avt / jetson_nano

This repository is a collection of scripts/programs I use to set up the software development environment on my Jetson Nano, TX2, and Xavier NX.
MIT License
293 stars 134 forks source link

tensorflow 2.0.0 build error #20

Closed YingUK closed 3 years ago

YingUK commented 3 years ago

Hi, Thanks for sharing your work!

I followed your instruction and tried to install tensorflow 2.0.0 on my jetson nano, however it gave below error after a long time compilation.

tensorflow/python/lib/core/bfloat16.cc:608:60: note:   no known conversion for argument 2 from '<unresolved overloaded function type>' to 'PyUFuncGenericFunction {aka void (*)(char**, const long int*, const long int*, void*)}'
Target //tensorflow/tools/pip_package:build_pip_package failed to build
Use --verbose_failures to see the command lines of failed build steps.
INFO: Elapsed time: 52964.953s, Critical Path: 1102.32s
INFO: 5632 processes: 5632 local.
FAILED: Build did NOT complete successfully

my system setup is: jetpack 4.3 tensorrt 6.0.1 cuda 10.0.326 python 3.7

Could you please suggest how can fix this issue?

jkjung-avt commented 3 years ago

I have seen a similar issue when building tensorflow-2.3.0 on Jetson Xavier NX. That's why I created the tensorflow-2.3.0.patch with fixes for the "PyUFuncGenericFunction" issue.

YingUK commented 3 years ago

Thanks for your reply. I assume this patch cannot fix tensorflow 2.0.0 issue that i mentioned above. Otherwise can I install tf 2.3.0 version in jetpack 4.3 using your script?

jkjung-avt commented 3 years ago

The patch is for 2.3.0. But you should be able to find the same lines of code in tensorflow-2.0.0 source and manually patch the code.

I have not tried to run "install_tensorflow-2.3.0.sh" on JetPack-4.3. It might not work out of the box...

YingUK commented 3 years ago

Hi,

I burned another SD card with JetPack 4.4, and installed tensorflow-2.3.0 successfully, although it took 3 days to complete on Nano. When it was installing scipy==1.4.1 as part of tensorflow wheel, it gave an error"numpy.disutils.system_info.NotFounfError: No lapacl/blas resources found". I fixed this issue based on this discussion. solution is:

sudo apt-get install gfortran libopenblas-dev liblapack-dev
sudo pip install scipy==1.2.1

Hope this will be helpful to who may see this issue in the future. @jkjung-avt , thanks again for your work, I will close this issue for now.