NVIDIA’s new Grace Hopper Superchip is revolutionizing Murex’s MX.3 platform by significantly boosting performance while lowering energy consumption, as reported by the NVIDIA Blog.

Managing Risk with MX.3 Powered by Grace Hopper

Post-2008 financial crisis regulations have necessitated stronger risk management frameworks for financial institutions. Pierre Spatz, head of quantitative research at Murex, foresaw the need for more efficient computing solutions. Consequently, Murex has integrated NVIDIA’s advanced technologies like CUDA and GPU-accelerated computing.

Murex recently began testing the Grace Hopper Superchip to improve the performance and energy efficiency of its MX.3 platform, which serves over 60,000 users in 65 countries for trading, risk management, and operations across various asset classes.

The Grace Hopper Superchip is being evaluated for handling X-valuation adjustments (XVA) and other critical calculations. Murex reports that the superchip delivers faster computations and significant power savings, essential for running complex risk models on large datasets.

“On counterparty credit risk workloads like CVA, Grace Hopper is ideal, leveraging a heterogeneous architecture with a mix of CPU and GPU computations,” said Spatz. “It’s the fastest processor for risk calculations and much more power-efficient, making green IT feasible in trading.”

Early tests show the Grace Hopper Superchip can reduce energy consumption by four times and improve performance by seven times compared to traditional CPU-based systems for XVA workloads in MX.3.

Pricing FX Barrier Options in MX.3 with Grace Hopper

Besides risk calculations, Murex has used the Grace Hopper Superchip to price foreign exchange (FX) barrier options using its advanced stochastic local volatility model. Barrier options are derivatives with payoffs dependent on whether the underlying asset’s price crosses a specified threshold during the contract period.

The pricing involves solving a two-dimensional partial differential equation more efficiently on the Arm-based NVIDIA Grace CPU in the GH200. Tests indicate the Grace Hopper Superchip processes these calculations 2.3 times faster than Intel Xeon Gold 6148 systems and is five times more power-efficient per server.

NVIDIA’s advancements in energy-efficient, high-performance computing offer significant benefits for quantitative analytics in the financial sector. Murex’s use of the Grace Hopper Superchip demonstrates the platform’s capabilities in enhancing cost-efficiency and computational speed.