This is a tool to interact with the machine learning models that we have trained to predict the performance and power efficiency of a computer system based on information about its components. We train our models on entries from theTOP500 dataset, a dataset that gives information about the components, performance, and energy usage of top supercomputers systems. The models themselves are from one of several sources:

• Scikit-learn for many general-purpose models

• XGBoost for extreme gradient boosted trees

• LightGBM for light gradient boosted trees

• Tensorflow for custom deep neural networks

The code we use to train the models will be made public in the near future.

1. Toggle what machine learning models are used by clicking the corresponding checkboxes

2. Input the information for the system you want to see a prediction for in the "Prediction Input" section

3. Click the submit button at the bottom of the "Prediction Input" section

4. Graphs displaying each enabled model's prediction for performance and energy efficiency will be displayed after some time.