Accelerating Computational Finance Simulations with OpenCL
Master thesis about performance optimization for computational finance models on state-of-the-art hardware.
My name is Michalis and I was an MSc student in TU Delft and a fulltime intern for Ortec Finance. As an engineer with a diverse background ,computer engineering & embedded systems, Ι was a bit hesitant at first, stepping into the finance sector. However, people from Ortec Finance Tech Labs provided me with all the necessary help for rigorous start for my project. The environment was uniquely friendly with people providing support and sharing knowledge in a research centric manner. Also, everyone was keen on discussing and brainstorming on new ideas and putting in the table state of the art technologies. Therefore, it was a great experience on expanding my knowledge on the finance industry, as well as adding new perspectives for the use of HPC hardware.
About the project
During my project, I had the opportunity to combine my previous knowledge in embedded systems with some real world problems, emerging from the financial industry. Therefore, we worked on accelerating computational finance simulation with OpenCL. We explored a number of different multicore platforms, while improving the performance and accuracy of a commercial product offered by Ortec Finance. Overall, it was a well-balanced experience between practical work and research. We managed to submit and present a representative paper of our work in the International Workshop on Multicore Software Engineering in August of 2016, in Grenoble. Overall, Ortec Finance Techlabs is a great environment for people with different backgrounds, who want to use and enhance their skills for solving real world problems with novel means.
Michal his thesis can be found here. In addition, the Europar ’16 contribution Is available in the Conference proceedings or can be downloaded as a chapter at Springer. The earlier working paper can be requested by email: techlabs@-ortec-finance.com.