Install Tensorflow with cuda 3.0 compute compatibility devices

This tutorial will explain you how to setup Tensorflow with cuda 3.0 compute compatibility devices (such as NVIDIA Grid K520, GTX580, GTX650, GTX770, GTX780…)

Basically, You need to use that specific build of Tensorflow if you encounter this kind of error message:

tensorflow/core/common_runtime/gpu/gpu_device.cc:611] Ignoring gpu device (device: 0, name: GRID K520, pci bus id: 0000:00:03.0) with Cuda compute capability 3.0. The minimum required Cuda capability is 3.5.

1 – Install Bazel

First, you need to install Bazel. I recommend to use Bazel 0.1.0, because some other builts raise errors when compiling Tensorflow.

1.1 – Setup Java JDK-8
Bazel requires java jdk-7 or 8. Follow the steps to install it:

Ubuntu (14.04)

$ sudo add-apt-repository ppa:webupd8team/java
$ sudo apt-get update
$ sudo apt-get install oracle-java8-installer

Ubuntu (15.10)

$ sudo apt-get install openjdk-8-jdk

1.2 – Install other dependencies

$ sudo apt-get install pkg-config zip g++ zlib1g-dev unzip

1.3 – Download Bazel 0.1.0
https://github.com/bazelbuild/bazel/releases/download/0.1.0/bazel-0.1.0-installer-linux-x86_64.sh

1.4 – Compile Bazel

$ chmod +x bazel-version-installer-os.sh
$ ./bazel-version-installer-os.sh --user

Bazel executable is located at: $HOME/bin/bazel

2 – Install Tensorflow

2.1 – Install dependencies

$ sudo apt-get install python-numpy swig python-dev wheel git

2.2 – Download Tensorflow 0.7.0

$ git clone --recurse-submodules https://github.com/tensorflow/tensorflow.git -b 0.7.0

2.3 – Compile Tensorflow

Configurate tensorflow built with cuda 3.0 compute capability

$ cd tensorflow
$ TF_UNOFFICIAL_SETTING=1 ./configure

Indicate basic settings according to your config.
For cuda compute compatibility, enter: 3.0

Compile tensorflow

$ $HOME/bin/bazel build -c opt --config=cuda //tensorflow/cc:tutorials_example_trainer
$ bazel-bin/tensorflow/cc/tutorials_example_trainer --use_gpu

After compiling, you should get that output:

000009/000005 lambda = 2.000000 x = [0.894427 -0.447214] y = [1.788854 -0.894427]
000006/000001 lambda = 2.000000 x = [0.894427 -0.447214] y = [1.788854 -0.894427]
000009/000009 lambda = 2.000000 x = [0.894427 -0.447214] y = [1.788854 -0.894427]

3 – Install Python interface

If you are planning to use tensorflow python interface, you first need to create a pip package to install it.

3.1 Create pip package

First step is to use bazel to create pip package

$ $HOME/bin/bazel build -c opt --config=cuda //tensorflow/tools/pip_package:build_pip_package

3.2 Install python tensorflow

We can then use that pip package to install tensorflow in python

$ bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg

# The name of the .whl file will depend on your platform.
$ pip install /tmp/tensorflow_pkg/tensorflow-0.6.0-cp27-none-linux_x86_64.whl

3.3 Test it

To check if everything is working well, you can run that example:

$ cd tensorflow/models/image/mnist
$ python convolutional.py
Successfully downloaded train-images-idx3-ubyte.gz 9912422 bytes.
Successfully downloaded train-labels-idx1-ubyte.gz 28881 bytes.
Successfully downloaded t10k-images-idx3-ubyte.gz 1648877 bytes.
Successfully downloaded t10k-labels-idx1-ubyte.gz 4542 bytes.
Extracting data/train-images-idx3-ubyte.gz
Extracting data/train-labels-idx1-ubyte.gz
Extracting data/t10k-images-idx3-ubyte.gz
Extracting data/t10k-labels-idx1-ubyte.gz
Initialized!
Epoch 0.00
Minibatch loss: 12.054, learning rate: 0.010000
Minibatch error: 90.6%
Validation error: 84.6%
Epoch 0.12
Minibatch loss: 3.285, learning rate: 0.010000
Minibatch error: 6.2%
Validation error: 7.0%
...
...

If you meet any difficulty you can leave a comment!