According to the NVIDIA blog, NVIDIA’s new Grace Hopper Superchip is transforming the capabilities of Murex’s MX.3 platform, significantly improving performance metrics while reducing energy consumption.
Managing Risk with MX.3 led by Grace Hopper
Following the 2008 financial crisis, regulatory changes led financial institutions to adopt stronger risk management frameworks. Pierre Spatz, Head of Quantitative Research at Murex, expects the demand for more efficient computing solutions to increase. In response, Murex adopted NVIDIA’s cutting-edge technologies, including CUDA and GPU-accelerated computing.
Recently, Murex began testing NVIDIA’s Grace Hopper Superchip, aiming to improve the performance and energy efficiency of the MX.3 platform. MX.3 is used daily by more than 60,000 users in 65 countries for trading, risk management and operations across a variety of asset classes.
The Grace Hopper Superchip is being specifically tested for its ability to handle X Valuation Adjustment (XVA) and other critical calculations. Murex reported that the superchip delivers faster computations and significant power savings, which are critical for agencies that need to run complex risk models on large data sets.
“For counterparty credit risk workloads like CVA, Grace Hopper is the perfect solution that leverages a heterogeneous architecture with a unique mix of CPU and GPU computation,” said Spatz. “When it comes to risk calculations, Grace is not only the fastest processor, but it is also much more power efficient, making green IT a reality for the world of trade.”
Initial tests show that the Grace Hopper Superchip can reduce energy consumption by 4x and improve performance by 7x compared to traditional CPU-based systems when running XVA workloads on MX.3.
Pricing FX Barrier Options on MX.3 with Grace Hopper
In addition to risk calculations, Murex leveraged Grace Hopper Superchip to price foreign exchange (FX) barrier options using advanced stochastic local volatility models. A barrier option is a type of derivative where the payoff depends on whether the price of the underlying asset exceeds a certain threshold during the life of the contract.
Pricing includes more efficient solving of two-dimensional partial differential equations on the GH200’s Arm-based NVIDIA Grace CPU. Testing showed that the Grace Hopper Superchip processed these calculations 2.3 times faster than the Intel Xeon Gold 6148 system. This chip is also 5x more power efficient per server.
NVIDIA’s commitment to energy-efficient, high-performance computing is evident in these advancements, which provide significant benefits for quantitative analysis in the financial sector. Murex’s Grace Hopper Superchip implementation demonstrates the ability of accelerated computing platforms to improve cost-effectiveness and computational speed.
Image source: Shutterstock
. . .
tag