This talk will feature a collection of senior research projects, presented by DigiPen students:
Templatized Lua Binding by Lux Cardell:
In order to integrate Lua scripting into a C++ project, each function
accessible to Lua must have a binding function. For a custom game engine
project, I bound Lua to allow our design team to iterate and design more
efficiently. To this end I wrote a set of templates that generated
binding functions automatically based on the data types the functions
required. These templates were designed to encapsulate the entirety of
the binding process, from popping arguments off the Lua stack to
returning an arbitrary number of values. I used variadic templates to
allow the template to handle any function signature. In order to interface
between dynamic and strict typing, I wrote a generic class that wrapped the basic types from Lua and the user-defined types commonly used throughout the engine.
TypeRT by Michael-Paul Moore:
In order to develop a complex C++ simulation, many collections of data
must be authored to control behavior during execution. Much of this is
driven by the creation of different classes that will each require
similar sets of utilities in order to effectively function. Utilities
such as: Points of Access, serialization, classification, etc. While
engineers can facilitate this functionality on a per class basis, this
is time consuming and can often result in duplicated logic. Furthermore,
while native representation is required during execution, by allowing
the class definitions to reside outside of code we can gain the
• Accessibility to non-engineers without compiler access
• Easily extend feature set on data usage inside and outside the main
• Since these class definitions are outside the toolset, they do not
require code to be built
• Further ensures abstraction from your data from different
To solve these problems we want to utilize a single flexible wrapper
class, referred to as TypeRT, to centralize interactions for game data.
This creates an environment in which a given functionality only needs to
be implemented once and then gained across multiple data types. In the
actual talk details will then be given on the TypeRT implementation.
Process by which existing classes can be ported will be explained.
Transforming and Reducing Workloads: C++17 Parallel Algorithms for Rapid Prototyping by Louis Tan:
C++17 introduced parallel execution policies for algorithms, and a versatile new std::transform_reduce algorithm for taking advantage of the feature. These both allow programmers to spend less time on the implementation details of dispatching parallel workloads, and more time on parallel design through rapid prototyping and iteration. In this talk, we will explore the use case of parallelizing a physics engine for performance, and how std::transform_reduce was used as a catch-all solution to accomplish this.
I’m a fourth-year student in the Bachelor’s of
Science in Computer Science program at DigiPen. I’ve worked on game
engines for the past three years, and in the last year on a physics
project calculating the binding energies of hydrogen in weak magnetic
fields. Additionally, I competed this year in the International
Collegiate Programming Contest, representing DigiPen in the first
division. In the game engine I wrote in my third year, I integrated Lua
scripting to facilitate a team of designers in creating a game. As the
process of writing binding functions for Lua is largely repetitive but
highly function-specific, I found a way to genericize the process of
writing binding libraries using C++ templates.
I am a Senior at DigiPen institute of
technology and currently work as an Associate Software Engineer at
Louis is an avid tinkerer and recent graduate who got his start in C++ developing game mods. He believes in the value of a diverse problem-solving toolkit and has gathered experience in multiple areas not limited to game development, including graphics, physics, reverse engineering and AI. He has also been known to engage in template metaprogramming for amusement, sometimes just to make certain people on the internet very, very angry.