One Stop Systems Breakfast Event
Presented at SC16
7:30 AM – 8:00 AM: Registration and Check-In
8:00 AM – 8:10 AM: Steve Cooper
Introducing Magma, a division of One Stop Systems
Steve Cooper will discuss the merger between One Stop Systems and Magma. Both companies are market leaders in PCIe expansion technology used to create high-end compute accelerators and flash storage arrays. Together they become a dominant technology leader of PCIe expansion appliances.
8:10 AM – 8:20 AM: Steve Cooper
Flash Arrays & GPU Compute Accelerators: Direct Attached vs. All-in-One
Steve Cooper will discuss One Stop Systems’ direct attached and all-in-one flash arrays and GPU compute accelerators. He will highlight the major differences between the two product lines and discuss the benefits and applications of each type.
8:20 AM – 8:40 AM: Tim Miller
Direct Attached Accelerators
Compute accelerator expansion systems housing multiple state of the art GPUs provide a flexible option to transform standard servers into ultra-high performance HPC processing nodes. High speed PCIe Gen3 x16 cable connections provide over 12GB of bandwidth per link from the server to the GPUs with total software transparency. One Stop Systems provides a wide range of direct attached compute accelerator options to meet the various needs of customer’s applications and datacenter environment. From power and cost efficient systems that seamless fit in existing data center infrastructure in the EB36xx family to the ultra-dense high capacity of the HDCA family with the industry’s highest teraflop/space solutions. Coupled now with up to 16 of the P100 Pascal GPUs OSS can provide the ultimate in dense HPC performance. Additionally One Stop Systems provides scale out options through its GPUltima to form clusters of these HPC nodes interconnected through high speed RDMA fabrics. This talk will address the capabilities and benefits of the One Stop Systems direct attached compute accelerator products.
8:40 AM – 9:00 AM: Steve Cooper
All-in-one GPU Accelerated Servers for HPC
With the advent of NVLink that provides extreme bandwidth in GPU peer-to-peer applications and more tightly couples some CPUs to the latest SXM2 GPUs, the need for the 4 Pascal with NVLink GPU Accelerated Server (OSS-PASCAL4) and the 8 Pascal with NVLink GPU Accelerated Server (OSS-PASCAL8) will continue to take a leadership role over today’s general purpose servers in HPC applications. GPU centric applications from Deep Learning training to Oil & Gas, Financial modeling, Rendering and general HPC over VDI will demand more GPUs requiring the best-in-class bandwidth, connectivity, power and cooling. This talk will take the user through the OSS GPU Accelerated Servers offering including the NVIDIA DGX-1, the OSS-PASCAL4 and the OSS-Pascal8 server product line.
9:00 AM – 9:20 AM:Jim Ison
All-flash arrays for the Enterprise, Datacenter and Warfighter
Flash appliances have increasingly replaced traditional spinning disks in environments ranging from mobile devices to datacenters and defense vehicles. To many, the transition to an all-flash appliance becomes a simple replacement of 2.5” HDDs with 2.5” SSD using the same hardware infrastructure and software tools originally optimized for the HDD. The OSS approach optimizes the hardware for the environment and optimizes the software for the application using 3 product lines. The Storage Box (SB) provides direct attached 2.5” NVMe support and software transparency for large datasets to an existing server. The Flash Storage Array (FSAe) product line provides direct attached, high-density PCIe NVMe connectivity for the largest data sets at the highest speeds. The Flash Storage Array for Networks product line (FSAn) provides optimized hardware and advanced flash software for the best ROI in any environment by minimizing hardware, support and license costs. This talk walks you through the SB, FSAe and FSAn features and innovations in hardware and software that allow it to scale from 3TB in 1U up to 1PB in 16U.
9:20 AM – 9:30 AM: Q&A