Taipei, Taiwan, June 1st, 2021 –COMPUTEX 2021 Virtual Summit, ASRock Rack announces a wide selection of servers for AI featuring NVIDIA® A100 Tensor Core GPUs.
All the ASRock Rack NVIDIA® A100 GPU-powered servers offer performance solutions for a variety of workloads. NVIDIA-Certified Systems from ASRock Rack deliver the optimized performance, flexibility, and agility required to meet data center workload demands. Powered by NVIDIA® GPUs, the ASRock Rack GPU servers deliver unprecedented acceleration at every scale to power the world's highest-performing elastic data centers for AI, data analytics, and HPC.
The 1U4G-ROME is the 1U GPU server based on A100 PCIe GPU solution with PCIe 4.0 technology. The 1U4G-ROME also features 4 PCIe4.0 x16 dual-slot GPU support and equipped with two additional riser cards for PCIe4.0 x16 low-profile add-in cards, enabling NVIDIA® ConnectX-6 HDR 200Gb/s InfiniBand adapters and Ethernet NICs via the 128 PCIe4.0 lanes from single AMD EPYC 7003 series processor.
The 4U8G-ROME2/2T and 4U10G-ROME2/2T are interconnected with eight and ten NVIDIA® A100 PCIe GPUs. The memory-optimized configuration supports for up to 16+16 DIMM slots and up to 10 dual-slot width PCIe® Gen4 slots in a high-density form factor. These two server solutions feature front access that simplifies the maintenance and management and minimize service down-time to meet customers' demands for compute-intensive workloads.
And the 3U4X-ROME/2T is a 3U GPU server with 3000W redundant PSU server solution. Powered by four NVIDIA® A100 GPUs, the 3U4X-ROME/2T is an AMD EPYC 7003 based that provides the breakthrough performance in this 3U rackmount. Last but not least, the 7U8X-ROME2/2T is a dual AMD EPYC 7003 based AI training system that features eight channel DIMMs and 16x AIC NVMe SSD with eight powerful NVIDIA® A100 GPUs.
Powered by NVIDIA® Ampere architecture-based GPUs, the ASRock Rack servers help customers obtain the advanced level of integration and performance required in modern data centers, surpassing both customer expectations and energy efficiency requirements for HPC and AI workload demands.
To see more, please visit the COMPUTEX 2021 Virtual Summit in ASRock Rack.
Website: https://www.asrockrack.com/
Twitter: https://twitter.com/asrockrack
