TAIPEI, Taiwan, 24 May, 2022 — ASUS, the leading IT Company in server systems, server motherboards and workstations, today announced plans to adopt the world’s first NVIDIA CPU-powered system designs to deliver new servers powered by the NVIDIA Grace CPU Superchip to address the most demanding HPC, data analytics, digital twin simulations, cloud gaming and hyperscale computing applications. ASUS, a longstanding NVIDIA partner, is known for its R&D capabilities, deep engagement in AI development and consistent product quality. ASUS also plans to support NVIDIA liquid-cooled A100 PCIe GPUs to deliver power efficiency in modern data centers.

One of the first partners to adopt NVIDIA Grace CPU Superchip design

The NVIDIA Grace CPU Superchip is a breakthrough processor delivering the highest performance and twice the memory bandwidth and energy-efficiency of today’s leading server processors. Featuring two chips and up to 144 high-performance Arm® V9 cores with scalable vector extensions and a breakthrough 1 terabyte per second subsystem that consists of LPDDR5x memory, the Grace CPU Superchip can be configured in many ways, from CPU-only to accelerated servers with up to eight NVIDIA Hopper™ GPUs, to address a large diversity of workloads. ASUS is expected to release this new series in the first half of 2023.

Deploy liquid-cooled A100 PCIe GPUs with less energy

Liquid cooling is the next step for modern data centers to deliver better energy efficiency to offset the growing demand for more compute with higher performance. NVIDIA also announced the latest liquid-cooled A100 PCIe GPU for sustainable efficient computing, which is expected to be available this summer. ASUS is working on modifying its current system architecture to support liquid- and air-cooling solutions, and will have servers verified for this new liquid-cooled NVIDIA GPU.