Taipei, Taiwan, March 22, 2022 —ASUS, the leading IT company in server systems, server motherboards and workstations, today announced that ASUS ESC80000A-E11 and ESC4000A-E11 GPU servers will support the latest AMD Instinct™ MI210 accelerator and power discoveries in exascale computing, enabling scientists to tackle humanity’s most pressing challenges – from climate change to vaccine research.
Leveraging the latest 3rd Gen AMD EPYC™ processors and AMD Instinct MI210 (PCIe®) accelerator cards, ESC8000A-E11 and ESC4000A-E11 servers are able to take advantage of the full package of AMD server technologies to deliver the top performance for AI and simulation workloads from the edge to data centers and the cloud.
Advanced ASUS GPU-server design delivers competitive advantage
ESC8000A-E11 in 4U chassis is a dual-socket server based on the latest 3rd Gen AMD EPYC processors and supports up to eight AMD Instinct MI210 accelerators, making it ideal for optimizing AI training, analytics and machine-learning workloads. In addition, the advanced thermal and cooling designs, which include independent airflow tunnels for both the CPU and GPU, enable seriously-efficient heat dissipation. Market-leading features include support for up to four 3000 W 80 PLUS® Titanium redundant power supplies. This configuration delivers up to 96% efficiency, and ensures power is supplied to the system equally from all PSUs – maximizing efficiency and ensuring uninterrupted operation.
“ASUS has a long history of incorporating new technologies with AMD to empower application-optimized server solutions for different workloads. Our innovative GPU server design delivers a competitive advantage in the market, giving our customers the ability to solve the complex challenges much faster than before,” said Robert Chin, General Manager of ASUS Server and Workstation Business Unit. “We will continuously bring more servers integrated with AMD’s advanced innovations to help accelerate the developments in HPC and AI.”
ASUS also offers ESC4000A-E11 in 2U chassis, a single-socket server that is suited for deploying modern AI training clusters at scale with support for up to four AMD Instinct MI210 accelerators. Its flexible design allows graphics, networking and storage expansion and hardware upgrades, including the capability of up to eleven PCIe 4.0 slots for a diverse application, and an easy-access front-panel design that features four NVMe SSD drives and an optional choice of OCP 3.0, an ASUS Hyper M.2 adapter or HBA/RAID card via a PCIe 4.0 slot.
AMD innovations deliver performance leadership
The AMD Instinct MI210 accelerator is built to accelerate deep-learning training, providing an expanded range of mixed-precision capabilities based on the AMD Matrix Core technology, and delivers an outstanding 181 teraflops peak theoretical FP16 and BF16 performance to bring users a powerful platform to fuel the convergence of HPC and AI.
With innovative new double-precision (FP64) Matrix Core capabilities, along with the 3rd Gen AMD Infinity Architecture, AMD is delivering improved performance, efficiency and overall system throughput for HPC and AI – all powered by the latest AMD EPYC CPUs and AMD Instinct MI200-series accelerators.