how to set gpu configuration

[AllenDun]

Environment

If an issue occurs it is helpful to provide several information: Please replace these information below:

info	output
os	uname -a
c++ version	c++ --version
TF version	python -c "import tensorflow as tf; print(tf.__version__);"
TF is working	python -c "import tensorflow as tf; sess=tf.InteractiveSession()" works

You might run the tf_info.sh script in your terminal to gather these information and manually anonymize the output when necessary before pasting it here.

Issue

A clear and concise description of what the bug is will be helpful to resolve issues.

Context:
Please explain the conditions which led you to write this issue, eg. "writing a custom op" or "doing inference on device given a keras model".

Reproduce:
Please explain what you did (exactly), eg., files you modified, or, commands you entered so that I can reproduce if possible.

Output
Please explain what you get as an result or output from the reproduce-steps above, eg. "I get the output/error message ... during the compilation". Please provide the exact output and no interpretation.

Expectation
Please explain what you expected to get, eg. "the inference code x should produce y".

Investigation
If possible, please explain, which steps you did to investigate this issues. What are your findings.

nice work! could you add some codes in your examples to support gpu configurations for multi-gpu environment? It is really bothering when do inference with tensorflow in multi-gpu machine.Thanks

[PatWie]

Have you tried

CUDA_VISIBLE_DEVICES=0 ./path/to/script