name: Performance Issue about: Use this template for reporting a performance issue labels: ‘type:performance’

Please make sure that this is an issue related to performance of TensorFlow. As per our GitHub Policy, we only address code/doc bugs, performance issues, feature requests and build/installation issues on GitHub. tag:performance_template

System information

  • Have I written custom code (as opposed to using a stock example script provided in TensorFlow):
  • OS Platform and Distribution (e.g., Linux Ubuntu 16.04):
  • Mobile device (e.g. iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device:
  • TensorFlow installed from (source or binary): - TensorFlow version (use command below):
  • Python version: - Bazel version (if compiling from source):
  • GCC/Compiler version (if compiling from source):
  • CUDA/cuDNN version: - GPU model and memory:

You can collect some of this information using our environment capture script You can also obtain the TensorFlow version with: 1. TF 1.0: python -c "import tensorflow as tf; print(tf.GIT_VERSION, tf.VERSION)" 2. TF 2.0: python -c "import tensorflow as tf; print(tf.version.GIT_VERSION, tf.version.VERSION)"

Describe the current behavior

Describe the expected behavior

Standalone code to reproduce the issue Provide a reproducible test case that is the bare minimum necessary to generate the problem. If possible, please share a link to Colab/Jupyter/any notebook.

Other info / logs Include any logs or source code that would be helpful to diagnose the problem. If including tracebacks, please include the full traceback. Large logs and files should be attached.