Closed chyidl closed 4 years ago
I second this.
I naively tried the following:
# cd build_tensorflow/
# docker build -t tf-arm -f Dockerfile .
# docker run -it -v /tmp/tensorflow_pkg/:/tmp/tensorflow_pkg/ --env TF_PYTHON_VERSION=3.6 tf-arm ./build_tensorflow.sh ./configs/rpi.conf
(notice the TF_PYTHON_VERSION=3.6
above; that's my only change)
It failed because it could not find a path to python. I forget the exact environment variable's name (something like PYTHON_BIN_PATH
), but it ended up being an empty path thus it wasn't executable. Unfortunately I closed that terminal so I don't have the exact error on hand.
Is it possible to use this repo to build for python 3.6?
Hi, using crosscompiler, gcc linaro don't supports python version >= 3.6.
System information
Distributor ID: Raspbian Description: Raspbian GNU/Linux 9.6 (stretch) Release: 9.6 Codename: stretch gcc: $ gcc -v clang version 6.0.0 (tags/RELEASE_600/final) Target: armv7l-unknown-linux-gnueabihf Thread model: posix InstalledDir: /usr/local/clang_6.0.0/bin Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/4.6 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/4.6.4 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/4.7 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/4.7.3 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/4.8 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/4.8.5 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/4.9 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/4.9.3 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/5 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/5.4.1 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/6 Found candidate GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/6.3.0 Selected GCC installation: /usr/lib/gcc/arm-linux-gnueabihf/6.3.0 Candidate multilib: .;@m32 Selected multilib: .;@m32 Python version 3.6.7 Build label: 0.20.0- (@non-git)
Command triggering issue
$ bazel build -c opt --copt="-mfpu=neon-vfpv4" --copt="-funsafe-math-optimizations" --copt="-ftree-vectorize" --copt="-fomit-frame-pointer" --local_resources 1024,4.0,1.0 --verbose_failures tensorflow/tools/pip_package:build_pip_package --incompatible_remove_native_http_archive=false
Error Logs
+1 to this. I'm working with Odroid XU4 with Ubuntu 18.04, that comes with python3.6 by default.
I have to make lots of efforts to install both tensorflow and opencv, because tensorflow only supports python3.5, but opencv is not available by source on python3.5.
what should be bazel version for building tensorflow 1.12 on raspberry pi 3b i just need compatible bazel version
I've created the dependencies for compiling tag 1.13.1 natively on the odroid xu4. The config is conf/odroidxu4-native.conf, and build dependencies can be found in the Docker.bionic file at https://github.com/mihailescu2m/tensorflow-on-arm
I want use my Raspberry Pi 3B+ to build Tensorflow Version (1.12.0) with bazel version (0.20.0) and Python3.6. but it's take a long time to failed..