Acceleration with 20 times the computing power and A100 GPUs

With OSS 4UV, system developers can scale almost any Gen4 server to increase the performance of compute-intensive applications such as AI training, AI inference, video analysis and XR rendering. For AI inference on INT4 sparse neural networks, the system can achieve peak computing operations of up to 19.9 PetaOPS.

The A100 GPUs can be interconnected via third-generation NVlink that offers 600GB/sec of GPU-to-GPU interconnect. The OSS 4UV also features two PCIe x16 Gen 4 slots that allow the latest generation of network interface controllers (NICs) to transport data throughout the datacenter or edge network.

While the OSS 4UV can connect almost any Gen 4 server to eight A100 PCIe GPUs and two NICs, OSS also offers a complete NVIDIA GPU cloud-ready solution with the OSS 4UV pre-installed in an OSS 2U-EOS-4a Gen 4 server.

“Once again we have delivered unprecedented performance and efficient scalability with the first expansion system that supports the industry’s latest and most powerful GPU accelerators,” said David Raun, CEO of OSS. “Our solution includes a full stack of AI software and pre-trained AI models that are perfect for AI on the Fly applications, allowing our customers to accelerate workloads at every scale.”

About One Stop Systems

One Stop Systems, Inc. (OSS) designs and manufactures innovative specialized high-performance computing modules and systems, including customized servers, compute accelerators, expansion systems, flash storage arrays and Ion Accelerator storage software. These products are used for deep learning, AI, defense, finance and entertainment applications and empower scientists, engineers, creators and other professionals to push the boundaries of their industries.