TensorFlow：CentOS上的二进制安装错误

Question

I was trying to have TensorFlow installed on my 32-bit CentOS 6.8 linux machine. I have python 3.6 installed with pip 9.0.1. However as I try this command:

pip install https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-0.7.1-cp27-none-linux_x86_64.whl

I got:

tensorflow-0.7.1-cp27-none-linux_x86_64.whl is not a supported wheel on this platform.

The full log info are pasted below:

tensorflow-0.7.1-cp27-none-linux_x86_64.whl is not a supported wheel on this platform.
Exception information:
Traceback (most recent call last):
File "/usr/local/python3/lib/python3.6/site-packages/pip/basecommand.py", line 215, in main status = self.run(options, args)
File "/usr/local/python3/lib/python3.6/site-packages/pip/commands/install.py", line 312, in run wheel_cache
File "/usr/local/python3/lib/python3.6/site-packages/pip/basecommand.py", line 276, in populate_requirement_set wheel_cache=wheel_cache
File "/usr/local/python3/lib/python3.6/site-packages/pip/req/req_set.py", line 245, in add_requirement wheel.filename
pip.exceptions.InstallationError: tensorflow-0.7.1-cp27-none-linux_x86_64.whl is not a supported wheel on this platform.

Not sure what the problem could be. Any suggestion or advice would be greatly appreciated!

---

Update: I have tried the same code on a 64bit centOS 6.5， getting the same error. So maybe I can conclude this has nothing to do with the x86_64 part?

Answer 1

The issue is, you are using py3.6 and 32 bit. pip looks at filenames to figure out what the package is built for:

tensorflow-0.7.1-cp27-none-linux_x86_64.whl

This package name says it is for python 2.7 for a 64 bit linux system. https://www.python.org/dev/peps/pep-0423/

for python 3.6, it is not a guaranteed way, but you may download the pip package for python 3.5 and rename it by replacing all "35"s in the name with "36"

Answer 2

Your machine/os is 32-bit, however tensorflow-0.7.1-cp27-none-linux_x86_64.whl only fit 64-bit machine. And I can't find 32-bit version tensorflow package in tensorflow home page:

https://www.tensorflow.org/get_started/os_setup

It seems the only way to solve it is try a 64-bit machine and 64-bit os.