Investment bank JP Morgan is using Nvidia graphics processors (GPUs) to cost-effectively achieve a 40-fold increase in the performance of risk calculations.
As part of a three-year plan to reduce the computational costs of risk calculation by 75%, JP Morgan's Equity Derivatives Group is migrating from a CPU-based architecture to a hybrid system using Nvidia Tesla M2070 GPUs in its datacentres.
According to Nvidia, using GPUs as companion processors has accelerated application performance by 40 times compared to running them on CPUs. Nvidia says the architecture is also 80% cheaper.
JP Morgan is using the hybrid system to calculate risk across a range of products in a matter of minutes rather than overnight. The GPUs are deployed in multiple datacentres across the bank's global offices via its Compute Backbone infrastructure, allowing the company to distribute processing between tens of global applications. This integration of GPUs into the shared global computational infrastructure has resulted in GPU utilisation rates approaching 70%, 24 hours a day.