Lunch Lecture: a.s.r.

At a.s.r we have a tens of millions of insurance contracts. Increasingly strict regulations require a.s.r. to calculate financial risk impacts for each contract for thousands of future economic scenarios. A typical simulation run requires a quadrillion (1.000.000.000.000.000) calculations, to be completed in only hours. That’s why we use 10.000 processor cores to do the calculations in parallel. We will show how we have managed to do this and which practical problems and bottlenecks arise when applying calculations on these vast scales: terabytes of data to handle, limited working memory and data structure optimizations.

The link can be found here.

Register now