TAIPEI, Taiwan, March 19, 2024 —ASUS today announced its participation in the upcoming CloudFest, the premier event in the cloud and internet infrastructure space. At booth #C03, ASUS will unveil its latest technological advancements, headlined by the revolutionary ASUS ESC NM1-E1 GPU server. This cutting-edge server, leveraging innovative MGX modular reference architecture from NVIDIA®, is set to redefine the landscape of AI supercomputing, offering unparalleled capabilities and performance.
In its commitment to empowering businesses to build robust generative AI platforms, ASUS presents a comprehensive portfolio of server solutions. Ranging from accessible entry-level models to high-performance GPU servers, along with an array of liquid-cooled rack solutions, ASUS caters to a diverse spectrum of computing needs. With deep expertise in MLPerf benchmarks, the ASUS team excels in refining both hardware and software ecosystems for optimal performance in large-language-model (LLM) training and inferencing, ensuring seamless integration of holistic AI solutions for the demanding realms of AI supercomputing.
Introducing customized AI solutions with ASUS NVIDIA MGX-powered server
The latest ASUS NVIDIA MGX-powered 2U server, ESC NM1-E1, showcases the NVIDIA GH200 Grace Hopper™ Superchip, which offer high performance and efficiency. The NVIDIA Grace CPU includes 72 Arm® Neoverse V9 CPU cores with Scalable Vector Extensions (SVE2) and is powered by NVIDIA NVLink-C2C technology. Integrating with NVIDIA BlueField-3 DPUs and ConnectX-7 network adapters, ASUS MGX-powered servers deliver a blazing data throughput of 400 Gb/s, ideal for enterprise AI development and deployment. Coupled with NVIDIA AI Enterprise, an end-to-end, cloud-native software platform for building and deploying enterprise-grade AI applications, ESC NM1-E1 provides unparalleled flexibility and scalability for AI-driven data centers, HPC, data analytics and NVIDIA Omniverse applications.
Compact GH200 1U server for large-scale AI and HPC
Designed in a 1U and 800 mm chassis, ASUS ESR1-511N-A1 is powered by the NVIDIA GH200 Grace Hopper Superchip, supporting deep-learning (DL) training and inference, and HPC. The latest NVIDIA Superchip, with NVIDIA NVLink® C2C technology, delivers a coherent memory pool, high bandwidth and low latency. Its compact size offers high density, excellent scalability and optimal rack configuration to seamlessly integrate into existing infrastructure.
Elevating AI with the end-to-end NVIDIA HGX H100 eight-GPU server
ASUS ESC N8A-E12 is a robust 7U dual-socket server that harnesses the power of dual AMD EPYC™ 9004 processors and eight NVIDIA H100 Tensor Core GPUs. It's engineered to fuel AI and data-science advancements, offering a unique one-GPU-to-one-NIC configuration for maximal throughput in compute-heavy tasks. With its advanced cooling solutions and innovative components, ESC N8A-E12 stands as a paragon of thermal efficiency, scalability and groundbreaking performance, poised to significantly lower operational costs.
Empowering enterprises with advanced AI infrastructure
Designed to meet the demanding requirements of enterprise AI applications, the ASUS ESC8000A-E12P server integrates top-tier GPUs, swift GPU interconnects and a high-bandwidth fabric. Supporting up to eight dual-slot GPUs, this powerful server offers scalable configurations to tackle diverse workloads, providing options for NVIDIA NVLink Bridge or AMD Infinity Fabric™ Link to enhance performance scaling – ready to excel in both AI and HPC environments, and deliver unmatched performance and scalability.
SupremeRAID server solutions for next-gen workloads
The ASUS RS720A-E12-RS24 server, powered by AMD EPYC 9004 processors, integrates SupremeRAID™ by Graid Technology for remarkable throughput, low latency and exceptional scalability, setting new industry standards. In conjunction with BeeGFS, the parallel file system renowned for its scalability and performance, RS720A-E12-RS24 removes the traditional RAID bottleneck to deliver maximum SSD performance without consuming CPU cycles – a collaborative effort between both ASUS and Graid Technology that harnesses the strengths of each to deliver unparalleled efficiency and reliability in storage solutions.
This powerhouse server is also equipped with 128 Zen 4c cores, up to 4800 MHz DDR5 memory, and supports a 400-watt TDP per socket. Featuring 24 bays for versatile drive options and nine PCIe 5.0 slots, this server offers extensive upgrade potential. With advanced air cooling, remote management capabilities and support for multiple GPUs, it's an ideal solution for demanding AI and HPC applications.
Innovative multi-node solutions with direct-to-chip cooling
The ASUS RS720QA-E12-RS8U server features a four-node design and is optimized for CDN, HCI and cloud applications, enhancing IT flexibility and performance. Its direct liquid cooling efficiently manages high-TDP CPUs, improving data center efficiency. Each node is equipped with robust features, supporting dual processors, extensive memory and multiple connectivity options. Designed for HPC data centers, this server optimizes space and includes direct-to-chip (D2C) cooling solutions, contributing to lower PUE and operational costs, aligning with sustainable energy goals.
ASUS is excited to collaborate with industry partners to provide a full spectrum of server solutions, cooling modules and customized data-center designs, delivering unparalleled efficiency and performance tailored to the evolving needs of the digital landscape.