Failed building wheel of h5py

I tried building h5py==3.1.0 in a clean environment using docker, as demonstrated by @AastaLLL. Failed building wheel of h5py - #5 by AastaLLL
Through this, I was able to obtain the whl file.

mkdir ~/data
sudo docker run -it --rm --runtime nvidia -v $HOME/data/:/data/ nvcr.io/nvidia/l4t-base:r32.7.1 /bin/bash

apt-get update
apt-get install -y python3-pip pkg-config
apt-get install -y libhdf5-serial-dev hdf5-tools libhdf5-dev zlib1g-dev zip libjpeg8-dev liblapack-dev libblas-dev gfortran
pip3 install -U pip testresources setuptools
ln -s /usr/include/locale.h /usr/include/xlocale.h
pip3 install --verbose 'protobuf<4' 'Cython<3'
pip3 install numpy==1.19.4 pkgconfig==1.5.5
apt-get install -y git
git clone https://github.com/h5py/h5py.git
cd h5py
git checkout 3.1.0
git config user.email "you@example.com"
git config user.name "Your Name"
git cherry-pick 3bf862daa4ebeb2eeaf3a0491e05f5415c1818e4
H5PY_SETUP_REQUIRES=0 pip3 install . --no-deps --no-build-isolation
H5PY_SETUP_REQUIRES=0 python3 setup.py bdist_wheel
cp dist/h5py-3.1.0-cp36-cp36m-linux_aarch64.whl /data/
exit

ls -l ~/data/h5py-3.1.0-cp36-cp36m-linux_aarch64.whl

Thanks for this reply, it really helped me get on track again, I really appreciate it!

Now I understand some of the versions involved and the base image name. Would you share with me the proper way to build the tensorflow for this version? I am trying to use the latest Jetpack , i.e. Jetpack 4.6.1 or 4.6.4.

I think the repos that I am pointing to by default don’t have the nv22.1 part? What is nv22.1? I am getting the following error:

Removing intermediate container 77e72898829b
—> cb7a53327aa7
Step 27/48 : RUN pip install -U tensorflow==1.15.5+nv22.1 tensorboard==1.15.0 tensorflow-estimator==1.15.1
—> Running in 44f72b2e1ecf
ERROR: Could not find a version that satisfies the requirement tensorflow==1.15.5+nv22.1 (from versions: none)
ERROR: No matching distribution found for tensorflow==1.15.5+nv22.1
The command ‘/bin/sh -c pip install -U tensorflow==1.15.5+nv22.1 tensorboard==1.15.0 tensorflow-estimator==1.15.1’ returned a non-zero code: 1

As there are several sets of instuctions floating around out there, hoping for some clarity from you. Should the instructions be as shown above (from @dusty_nv ):

wget --no-check-certificate https://developer.download.nvidia.com/compute/redist/jp/v461/tensorflow/tensorflow-2.7.0+nv22.1-cp36-cp36m-linux_aarch64.whl
pip3 install --verbose tensorflow-2.7.0+nv22.1-cp36-cp36m-linux_aarch64.whl

— OR –
(from the Dockerfile for jetbot 0.43 release)

ARG TENSORFLOW_URL=https://nvidia.box.com/shared/static/rummpy6q1km1wivomalpkwt2jy28mndf.whl

ARG TENSORFLOW_WHL=tensorflow-1.15.2+nv-cp36-cp36m-linux_aarch64.whl

RUN wget --quiet --show-progress --progress=bar:force:noscroll --no-check-certificate ${TENSORFLOW_URL} -O ${TENSORFLOW_WHL} &&
pip3 install ${TENSORFLOW_WHL} --verbose &&
rm ${TENSORFLOW_WHL}

I am trying to use the jetbot 0.43 release… with 4.6.1 or 4.6.4
jetston Dev kit

—OR —
(thinking out loud)
Should I somehow layer in the correct version of the tensorflow image here on top of the base image?

This topic is about building h5py, so I recommend creating a new topic for discussions outside of this. If you’re looking for JetBot JetPack 4.6.1, there are already places that provide it. I suggest asking on their Discord channel.

https://faboplatform.github.io/JetbotDocs/02.ソフトウェア設定/01.Jetson%20Nano/02.OSアップグレード手順/03.jetbot/

discord channel
https://faboplatform.github.io/EdgeAIDocs/#discord

Thank you!! Works perfect!!

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.