NVIDIA Grace Hopper Superchip Boosts Murex MX.3 Analytics, Cuts Power Use
The financial sector is experiencing a significant technological upgrade as Murex, a Paris-based trading and risk management software provider, integrates NVIDIA's Grace Hopper Superchip into its MX.3 platform. This collaboration aims to deliver enhanced analytical performance and substantial power savings, according to an announcement from NVIDIA Blog.
Managing Risk With MX.3 Driven by Grace Hopper
Following the 2008 financial crisis, financial institutions have increasingly focused on reducing computing expenses while maintaining high-performance risk management systems. The Murex MX.3 platform plays a crucial role in this domain, assisting over 60,000 users in 65 countries with trading, risk, and operations across various asset classes. The platform's capabilities encompass credit and market risk, BASEL capital standards, and x-valuation adjustment (XVA), which includes credit value adjustment (CVA), margin value adjustment, and funding valuation adjustment.
To enhance MX.3's performance, Murex is testing the Grace Hopper Superchip for XVA calculations, market risk calibration, pricing evaluation, and sensitivity and profit and loss calculations. Pierre Spatz, head of quantitative research at Murex, highlighted the chip's benefits: “On counterparty credit risk workloads such as CVA, Grace Hopper is the perfect fit, leveraging a heterogeneous architecture with a unique mix of CPU and GPU computations. On risk calculations, Grace is not only the fastest processor, but also far more power-efficient, making green IT a reality in the trading world.”
The testing phase has shown promising results, with the Grace Hopper Superchip delivering a 4x reduction in energy consumption and a 7x performance improvement compared to traditional CPU-based systems.
Pricing FX Barrier Options in MX.3 With Grace Hopper
Murex has also tested the Grace Hopper Superchip for pricing foreign exchange (FX) barrier options, employing its latest stochastic local volatility model. A barrier option is a derivative with a payoff dependent on whether the underlying asset price reaches or crosses a specified threshold during the option contract's duration. The pricing evaluation utilizes a 2D partial differential equation, which is more cost-effective on the Arm-based NVIDIA Grace CPU in GH200.
The results have been noteworthy, with the Grace Hopper Superchip enabling a 2.3x faster pricing process compared to Intel Xeon Gold 6148. Additionally, the chip offers significant power efficiencies, with a 5x improvement in watts-per-server basis for FX barrier calculations.
According to Murex, NVIDIA’s next-generation accelerated computing platform is driving energy efficiency and cost-saving measures for high-performance computing in quantitative analytics within capital markets. These advancements are expected to set new standards in the financial technology landscape, benefiting banks, asset managers, pension funds, and other financial institutions globally.