ASRock Rack Inc., the leading innovative server company, today is set to showcase a comprehensive range of servers for diverse AI workloads catering to scenarios from the edge, on-premises, and to the cloud at booth #1737 at SC 23 held at the Colorado Convention Center in Denver, USA. The event is from November 13th to 16th, and ASRock Rack will feature the following significant highlights:

At SC 23, ASRock Rack will demonstrate the NVIDIA-Qualified 2U4G-GENOA/M3 and 4U8G series GPU server solutions along with the NVIDIA H100 PCIe. The ASRock Rack 4U8G and 4U10G series GPU servers are able to accommodate eight to ten 400 W dual-slot GPU cards and 24 hot-swappable 2.5" drives, designed to deliver exceptional performance for demanding AI workloads deployed in the cloud environment. The 2U4G-GENOA/M3, tailored for lighter workloads, is powered by a single AMD EPYC 9004 series processor and is able to support four 400 W dual-slot GPUs while having additional PCIe and OCP NIC 3.0 slots for expansions.In addition, ASRock Rack will disclose their new range of products incorporating the NVIDIA HGX H100 8-GPU. The 6U8X-EGS2 and 6U8X-GENOA2 are dual-socket servers based on Intel 4th gen Xeon scalable processors and AMD EPYC 9004 series processors respectively, which are designed in 6U height with sufficient airflow to ensure the highest CPU and GPU capability. While having an eight-way configuration, they are also armed with 12 Gen 5 NVMe drive bays and eight to ten PCIe 5.0 x16 slots, as well as 4+4 PSU for full redundancy. The 6U8X-EGS2 and 6U8X-GENOA2 will also be supercharged by the NVIDIA HGX H200 8-GPU, which adopts HBM3e for faster time to results and turbocharges high-performing generative AI and HPC platforms, as well as enables real-time LLM inference. An eight-way HGX H200 provides over 32 petaflops of FP8 deep learning compute and 1.1 TB of aggregate high-bandwidth memory for the highest performance in generative AI and HPC applications.

Visitors to the event will have the opportunity to witness the MECAI server, powered by the NVIDIA GH200 Grace Hopper Superchip. MECAI is a 2U short-depth server based on the NVIDIA MGX reference architecture. This 2U short-depth MGX server has only 450 mm depth, making it ideal for deployment in space-constrained edge environments. With the NVIDIA GH200 Grace Hopper Superchip, which leverages exceptional memory technology and bandwidth to improve throughput, MECAI redefines the possibilities of edge AI, empowering businesses to unlock new levels of efficiency and intelligence in their operations.

ASRock Rack's Groundbreaking Server Systems
ASRock Rack will also unveil its latest GPU systems, including the Arm-based 4U2G-ALTRA powered by Ampere Altra Max and the 2U1G-B650 short-depth server based on AMD Ryzen 7000 series Processors, which are able to accommodate a 3-slot GPU. The company is also showcasing a 1U all-flash storage solution, the 1U24E1S-GENOA, which has 24 hot-swappable E1.S drive bays ready for PCIe 5.0 SSDs powered by a single AMD EPYC 9004 Series Processor. The revolutionary 4U highest-density multi-node server systems, the 4U18N-B550, and the 4U18N-C252 are also in the spotlight.

ASRock Rack's Extensive Lineup of Server Boards
As an innovative leader in the server industry, ASRock Rack has continuously delivered a broad selection of server boards for state-of-the-art server platforms. At its SC 23 booth, ASRock Rack will feature server boards with multiple PCIe 5.0 slots, such as the SIENAD8-2L2T, the BERGAMO2D16-2T, the BEGAMOD8-2T, the GENOAD8X-2T, the SP2C741D16X-2T, and the W790D8UD-1L1N2T/BCM for the increasing utilization of GPUs, accelerators aimed at various demands for today's AI inference, and more. The other highlights are half-width server boards, including the ROMED8HM3, the SIENAD10HM3, the SP2C741D16HM3, and the W790D8HM3, each based on AMD EPYC 8004 series processors, Intel 4th Gen Xeon Scalable, Intel Xeon W-3400, and Xeon W-2400 Processors, best for high-density CPU computing node.