Building Simulations that Scale

The faster and more reliably your protocol simulations and analyses run, the more efficiently you can iterate on design, and the larger the parameter space and the number of what-if scenarios you can test. In this talk, we’ll discuss how to engineer your simulations and analyses for performance—from data sourcing and pre-processing to executing your simulations in the cloud, post-processing, and visualisation. We’ll also show you how to properly profile your code, identify bottlenecks, and optimise your code, saving you time and headaches!

Some examples where these techniques are specifically useful are large-scale agent-based simulations, Monte Carlo simulations, and parameter sensitivity analysis. radCAD was specifically designed to handle large-scale, long-running (1h+) simulations robustly and we’ll show how you can use some of these features.

By the end of this presentation, you’ll be able to set up your own cloud simulation environment and greatly expand the types of simulations you can execute, with only your imagination as the limit.

Goals

1. Understand the different stages of the modelling and simulation process
2. Appreciate what makes scaling simulations challenging (and important)
3. Walk away with a toolbox of profiling and optimisation tools and techniques

Topics

1. Simulation Workflow
2. Simulation Framework
3. Simulation Profiling
4. Simulation Optimisation
5. Simulations in the Cloud

EthCC[8] — summer 2025

Ethereum Community Conferences & Workshops: 8-9-10-11 July in Brussels