alexbdg/top500

MINES ParisTech - Analyse of the top500.org site for ML predictions.

TOP500

Analyse of the top500.org site for ML predictions.

Guidelines

On the top500.org site, we can find the ranks and details of the 500 most powerful supercomputers in the world.

To start, have a look at the most recent list and briefly comment on the following question:

• What manufacturers produce the world's largest supercomputers?
• What types of processors do they use?
• What fraction of peak performance is typically achieved for the LINPACK benchmark?

Then, the work to do is:

• Export all the lists (xls format) from 2005 to 2020
• Preprocess your data (clean up, )
• Build a ML model to predict Rpeak, Rmax, Total number of cores and Core speed for the 5 next years
• Discuss results with respect to the Moore Law

Data

• TOP500 dataset

Downloaded from the Top500 website, two lists per years from 2005 to 2019.

Link : https://www.top500.org/

Explanation

The python notebook top500-Multi-Podium.ipynb is divided into 2 parts following the guidelines. The second one, is ruled by the variable head_num that describes the number of best computers to use for predictions.

Results

head_num = 10

Python libraries

os, sys, xlrd, Pandas, Numpy, seaborn, matplotlib, scikit-learn