Change Murex MX.3 Performance with NVIDIA Superchip

The ground-up NVIDIA GH200 Grace Hopper Superchip was inspired by HPC and large-scale AI applications

A CPU and GPU coherent memory paradigm for accelerated AI and HPC applications is provided by the NVIDIA GH200 Grace Hopper Superchip

The NVIDIA Grace Hopper architecture combines the versatility of the Grace CPU and the groundbreaking performance of the Hopper GPU into a single superchip

At the centre of the GH200 Grace Hopper Superchip, it can give a total bandwidth of up to 900 gigabytes per second

GPU threads to operate on up to 256 GPUs linked to the switch and high bandwidth access to up to 144 terabytes (TB) of memory

Murex is evaluating Grace Hopper for XVA computations on the MX.3 platform

Grace is the fastest and most power-efficient processor in the trading world, enabling green IT to become a reality in risk calculations