Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


Agenda Posted for ExaComm 2018 Workshop in Frankfurt

The ExaComm 2018 workshop has posted their Speaker Agenda. Held in conjunction with ISC 2018, the Fourth International Workshop on Communication Architectures for HPC, Big Data, Deep Learning and Clouds at Extreme Scale takes place June 28 in Frankfurt. ” The goal of this workshop is to bring together researchers and software/hardware designers from academia, industry and national laboratories who are involved in creating network-based computing solutions for extreme scale architectures. The objectives of this workshop will be to share the experiences of the members of this community and to learn the opportunities and challenges in the design trends for exascale communication architectures.”

Improving Deep Learning scalability on HPE servers with NovuMind: GPU RDMA made easy

Bruno Monnet from HPE gave this talk at the NVIDIA GPU Technology Conference. “Deep Learning demands massive amounts of computational power. Those computation power usually involve heterogeneous computation resources, e.g., GPUs and InfiniBand as installed on HPE Apollo. NovuForce deep learning softwares within the docker image has been optimized for the latest technology like NVIDIA Pascal GPU and infiniband GPUDirect RDMA. This flexibility of the software, combined with the broad GPU servers in HPE portfolio, makes one of the most efficient and scalable solutions.”

Advanced Networking: The Critical Path for HPC, Cloud, Machine Learning and More

Erez Cohen from Mellanox gave this talk at the Swiss HPC Conference. “While InfiniBand, RDMA and GPU-Direct are an HPC mainstay, these advanced networking technologies are increasingly becoming a core differentiator to the data center. In fact, within just a few short years so far, where only a handful of bleeding edge industrial leaders emulated classic HPC disciplines, today almost every commercial market is usurping HPC technologies and disciplines in mass.”

E8 Storage steps up to HPC with InfiniBand Support

Today E8 Storage announced availability of InfiniBand support to its high performance, NVMe storage solutions. The move comes as a direct response to HPC customers that wish to take advantage of the high speed, low latency throughput of InfiniBand for their data hungry applications. E8 Storage support for InfiniBand will be seamless for customers who now have the flexibility to connect via Ethernet or InfiniBand when paired with Mellanox ConnectX InfiniBand/VPI adapters. “Today we demonstrate once again that E8 Storage’s architecture can expand, evolve and always extract the full potential of flash performance,” comments Zivan Ori, co-founder and CEO of E8 Storage. “Partnering with market leaders like Mellanox that deliver the very best network connectivity technology ensures we continue to meet and, frequently, exceed the needs of our HPC customers even in their most demanding environments.”

Accelerating Ceph with RDMA and NVMe-oF

Haodong Tang from Intel gave this talk at the 2018 Open Fabrics Workshop. “Efficient network messenger is critical for today’s scale-out storage systems. Ceph is one of the most popular distributed storage system providing a scalable and reliable object, block and file storage services. As the explosive growth of Big Data continues, there’re strong demands leveraging Ceph build high performance & ultra-low latency storage solution in the cloud and bigdata environment. The traditional TCP/IP cannot satisfy this requirement, but Remote Direct Memory Access (RDMA) can.”

New Types of Memory, their support in Linux, and how to use them via RDMA

Christoph Lameter from Jump Trading LLC gave this talk at the OpenFabrics Workshop. “Recently new types of memory have shown up like HBM (High Bandwidth Memory), Optane, 3DXpoint, NVDIMM, NVME and various “nonvolatile” types memory. This talk gives a brief rundown on what is available and gives some example on how the vendors enable the actual use of this memory in the operating system (f.e. DAX and filesystems) and then show how an application would make use of this memory. In particular then we will be looking at what considerations are important for the use of RDMA to those memory devices.”

NVMe Over Fabrics High performance SSDs Networked for Composable Infrastructure

Rob Davis from Mellanox gave this talk at the 2018 OCP Summit. “There is a new very high performance open source SSD interfaced called NVMe over Fabrics now available to expand the capabilities of networked storage solutions. It is an extension of the local NVMe SSD interface developed a few years ago driven by the need for a faster interface for SSDs. Similar to the way native disk drive SCSI protocol was networked with Fibre Channel 20 years ago, this technology enables NVMe SSDs to be networked and shared with their native protocol. By utilizes ultra-low latency RDMA technology to achieve data sharing across a network without sacrificing the local performance characteristics of NVMe SSDs, true composable infrastructure is now possible.”

Alces Flight: On Demand HPC now Available in the Azure Marketplace

Microsoft Azure customers worldwide now gain access to Alces Flight to take advantage of the scalability, reliability and agility of Azure. With Alces Flight, it is possible for researchers to spin up any size of High-Performance Computing cluster in minutes, providing users with a fully-featured HPC environment that includes thousands of open source applications. 

Sharing High-Performance Interconnects Across Multiple Virtual Machines

Mohan Potheri from VMware gave this talk at the Stanford HPC Conference. “Virtualized devices offer maximum flexibility. This session introduces SR-IOV, explains how it is enabled in VMware vSphere, and provides details of specific use cases that important for machine learning and high-performance computing. It includes performance comparisons that demonstrate the benefits of SR-IOV and information on how to configure and tune these configurations.”

OpenFabrics Alliance Workshop 2018 – An Emphasis on Fabric Community Collaboration

In this special guest feature, Parks Fields and Paul Grun from the OpenFabrics Alliance write that the upcoming OFA Workshop in Boulder is an excellent opportunity to collaborate on the next generation of network fabrics. “Come join the community in Boulder this year to lend your voice to shaping the direction of fabric technology in big ways or small, or perhaps just to listen and learn about the latest trends coming down the pike, or to pick up tips and tricks to make you more effective in your daily job.”