One Stop Systems doubles HPC datacenter efficiency with rack scale GPU accelerator system offering



One Stop Systems, provider of high performance computing (HPC) GPU accelerators and NVMe flash arrays for HPC applications, expanded on Monday its line of rack scale NVIDIA GPU accelerator products with the introduction of GPUltima-CI.

The GPUltima-CI power-optimized rack can be configured with up to 32 dual Intel Xeon Scalable Architecture compute nodes, 64 network adapters, 48 NVIDIA Volta GPUs, and 32 NVMe drives on a 128Gb PCIe switched fabric, and can support tens of thousands of composable server configurations per rack. Using one or many racks, the OSS solution contains the necessary resources to compose any combination of GPU, NIC and storage resources as may be required in today’s mixed workload data center.

GPUltima-CI features the flexibility of disaggregated composable infrastructure that increases GPU accelerator utilization in mixed workload datacenters. With composable infrastructure, unused GPU, storage and network resources from one application are automatically released to other resource-hungry applications on other server nodes resulting in increased resource utilization.

One Stop designs and manufactures high performance compute accelerators, flash storage arrays and customized servers for deep learning, AI, defense, finance and entertainment applications. OSS utilizes the power of PCI Express, the latest GPU accelerators and NVMe flash cards to build systems, including many industry firsts, for OEMs and government customers. The company’s hardware and Ion Accelerator Software offers performance and scalability. OSS products are available directly, through global distributors, or via its SkyScale cloud services.

“As the most flexible solution for mixed workload datacenters, with GPUltima-CI we once again deliver the most cutting-edge, high-power computing solution for customers worldwide,” said Steve Cooper, CEO of OSS. “GPUltima-CI’s composable infrastructure allows any node in the system access to a multitude of NVIDIA Volta GPUs and expansive storage resources.

“This flexibility will be invaluable to many HPC applications, like AI, deep learning, image processing and scientific modeling. GPUltima-CI makes the datacenter workload-centric, where the hardware truly adapts to the needs of applications, rather than applications trying to adapt to limited hardware.”

“This new OSS system uses the latest NVIDIA Tesla V100 GPUs, by far the most advanced data center GPUs ever built, to accelerate AI, HPC, and graphics,” said Paresh Kharya, Group Product Marketing Manager at NVIDIA. “OSS customers can now fully harness the power of our Volta architecture with GPUltima-CI.”

Advanced PCIe switched fabric forms the critical interconnect between resources in the GPUltima-CI. OSS partner, Liqid, provides the multi-port PCIe switch that features an Intel Xeon processor and runs Liqid’s composable infrastructure software.  

“We’re excited to join OSS in delivering industry-leading GPU and NVMe composable solutions that support a new level of resource availability across the data center,” said Liqid CEO, Sumit Puri. “Our technology platform allows users to manage, scale out, dynamically configure and automate physical, bare-metal server systems. Liqid and OSS enable the ability to treat GPUs as a disaggregated, shared resource for the first time. OSS expansion systems and Liqid Command Center software provide the infrastructure to meet the most demanding of HPC deployments.”

Last month, One Stop introduced a new 2U Ion Accelerator Flash Storage Array that offers flexible capacity while maintaining the high-bandwidth and low-latency pedigree of Ion Accelerator arrays deployed in hundreds of global installations. This OSS shared flash storage array boasts the latest Ion Accelerator 5.0 software, NVMe drives, networking options and dual Intel® Xeon Scalable Processors to support the most demanding applications.

Multiple high-speed networking options allow this array to fit seamlessly into Fibre Channel (up to 32 Gbps), iSCSI (up to 100 Gbps) or InfiniBand (up to 100 Gbps) deployments. The capacity of the Ion Accelerator can be expanded up to 153 terabytes with 24 2.5-inch NVMe drives. High Availability (HA) is achieved using two arrays in a mirrored setup.

The new version 5.0 for the Ion Accelerator software from OSS provides a complete all-flash storage solution that delivers cost-effective, near-native NVMe performance for data-intensive workloads.

Ion Accelerator is designed to deliver consistent ultra-low latency and high bandwidth for the most performance-hungry applications used in information and financial services, healthcare, government, manufacturing, media and entertainment, and other major industries.

Offering bandwidth from a single storage box of 25GB/sec and latency well below 100 µs, Ion software is the ideal solution for delivering on the promise of NVMe flash performance in shared storage. Ion Accelerator has built-in data integrity in the software, with support for various RAID configurations, as well as a HA option that simultaneously accelerates performance for volumes across two systems.

Leave a Reply

WWPI – Covering the best in IT since 1980