logo

Discover

Categories
Popular Tags
SV logo
Published on: November 25, 2021 at
How to improve performance of your quant library
2
0
0
How to improve performance of your quant library
Quant libraries are being developed over a number of years, with the focus often being on functionality than performance. New requirements from the business and the regulators arise, making it essential to have easy to read, develop and maintain code. And, at the same time, getting the top performance is essential. In this presentation, we talk about every quant's practical need: maximum re-use of legacy code, efficient memory use, clear code, performance, parallel programming and AAD. We will introduce a novel approach to achieve the best of both worlds: keeping the simplicity of clear Object-Oriented analytics and getting the top performance of optimized Data-Oriented design. We will demonstrate the actual implementation of this approach using a well-known open-source quant library - QuantLib(http://quantlib.org), yielding 150x performance improvement on a single core for xVA calculations. We will also talk about developer tools that can make quant life a lot easier for times when things don't work as expected.
Comments 0