Welcome

Introduction

Welcome to the High Performance Data Analytics documentation.

This section has been created to show best practices for parallelizing, scaling and optimising your Data Science applications and developping end-to-end AI workloads on MeluXina.

Data Science Development Tools

JupyterLab

Jupyter Lab is one of the most widely used IDEs in the data science community. It is the choice tool for many data scientists when it comes to quick prototyping and exploratory analysis. JupyterLab neatly bundles many functionalities together, enabling collaborative, extensible, and scalable data science.

The following pages will show you how to use JupyterLab on MeluXina, depending on your client environment:

• Using JupyterLab when running from Linux
• Using JupyterLab when running from on Windows
• Converting notebooks to Python code

TensorBoard

• Using TensorBoard on MeluXina

Optimizations

MeluXina is extremely powerful when properly used, with the right tools and optimizations. Speed up the processing of your applications written in Python using Numba and CUDA by following the guides below:

• Acceleration with Numba
• Acceleration with CUDA (WIP)