Dell Expands HPC-On-Demand And Server GPU Options at ISC21
Dell Expanding HPC-On-Demand And Server GPU Options, Debuting Omnia Software at ISC21
A few weeks ago, Dell announced it intends to provide some of the most sophisticated supercomputer systems ever built at the upcoming International Science and Technology (ISC21), which will run a number of scientific, engineering and scientific research facilities to support those running research on supercomputers from the world’s first supercomputer center.
At the event in Las Vegas, Dell also announced an even more extreme version of its Omnia HPC-on-demand system, which will support an Intel Xeon processor-based desktop system and is said to run on a GPU or two and more, depending on the configuration. And the system is also said to be capable of using some of the most powerful supercomputers available at that time, including the Blue Gene/L supercomputer in Singapore and a number of other smaller systems.
According to Chris King, senior director of Intel’s Business Systems Division (BSD) for Dell’s HPC division, the omnia system can use as much as 80 to 100 percent of the Blue Gene/L systems’ available Intel Xeon processors and will run the Blue Gene supercomputer in Singapore with almost double the processor-based compute performance when compared to the Blue Gene/L on which it is based.
In addition to running its own system, Dell also is said to be working to get other vendors involved in providing HPC systems as well as providing a complete HPC-on-demand system with all necessary components and software.
There are some interesting take on the Omnia system, which will include Dell’s new ProLiant servers and Intel Xeon processors specifically built for it, and the Blue Gene/L system already running at the Singapore National University. That should be a great test bed for the Omnia system and should help Dell make some strong sales when the system is fully launched with a few months of availability.
The Dell Omnia system will be capable of running Linux and probably Windows too. The BlueGene/L system will be fully supported by Intel.
Omnia: An Open-Source Software Stack for Automation of HPC, AI and Data Analytics
Omnia was developed at Dell Technologies HPC – AI Innovation Lab, in collaboration with Intel and with support from the HPC community. The open-source software is designed to automate the provisioning and management of HPC, AI and data analytics workloads to create a single pool of flexible resources to meet expanding and diverse demands. The Omnia Software stack is an open-source set of Ansible playbooks that allow speeding up the deployment of converged workloads with Kubernetes and Slurm, along with library frameworks, services and applications. Omnia automatically places a software solution based on the use case — for example, HPC simulations, neural networks for AI, or in-memory graphics processing for data analysis — on each server to reduce the time from weeks to minutes.
Nvidia is proud to announce its partnership with Omicron, the world leader in HPC and AI software, to extend Nvidia’s leadership in these areas with Omicron’s leading solutions in software and infrastructure.
Omicron has already proven itself in AI and machine learning to accelerate AI training and enable real-time analytics on the edge. We are also working on a series of AI and HPC solutions that can perform intelligent decision and optimization over data, across a wide variety of applications and use cases.
In collaboration with Nvidia we have designed and implemented new software solutions for the HPC industry, enabling AI and data analytics on large open-source platforms and using Omicron’s large number of software products. These solutions now are available from all major HPC vendors.
Nvidia will deliver software as a service to Omicron, including the Omicron TensorFlow deep learning framework, the Omicron Caffe deep learning framework and Omicron’s own deep learning solution.
This new partnership with Omicron will enable Omicron to deliver deep learning solutions to customers at a lower cost and faster speed than traditional approaches and is expected to deliver hundreds of millions of dollars in annual savings in energy requirements, capital investments and operational expenses. Omicron has a strong competitive edge with a deep focus on AI and machine learning.
“Omicron has shown that we can be successful and disrupt an area of the industry like HPC, that Nvidia can support with the software and engineering expertise of Nvidia employees and the deep expertise of our partners,” said Michael Sorensen, CEO of Omicron.
Nvidia’s software and GPU engineering teams will deliver hardware innovations to help customers perform analytics and the high-performance computing (HPC) necessary to support them. These innovations are in the areas of AI and machine learning, and will also include advances in the areas of machine-learning deep learning, computer vision, computer vision processing and deep reinforcement learning.
“Omicron is one of the leaders in the HPC and AI space and we look forward to building a deeper partnership to accelerate the industry’s progress,” said Martin Braun, Vice President of Omicron Software Solutions.
HPC on demand, VMware and PowerEdge
In parallel, Dell is expanding its HPC on demand offering with VMware support by its partnership with analytics technology vendor R Systems and is offering Nvidia’s A30 and A10 Tensor Core GPUs as accelerator options for PowerEdge R750, R750xa and R7525 servers. That being said, the announcement is mostly about software, Baker says. He says that they have things like PowerProtect, our data protection portfolio and that it is all software. “A lot of what we do in terms of data analytics, such as our Streaming Data Platform, is pure software. A lot is focused on data management and helping our customers extract value from data or the automation of infrastructure, which is what the announcement is largely focused on. Innovation of the last decade and into the next is really focused on automation. The hottest companies in the world are right now the robot process automation vendors, which is software designed to drive automation business processes. The world is being rapidly automated. That includes IT systems. That includes the automobile. That includes all sorts of stuff. And at the heart of it is software.
The combination will push more compute to the edge, and enable a more collaborative environment for the cloud.
On Thursday, February 3, 2017 at 4:00pm PT, Cisco will showcase the next generation IOS Release 15(7) as the industry’s first Hybrid-Enterprise Cloud (HEC) offering that will support hybrid cloud deployment across the enterprise and bring a unique and powerful combination of compute and storage that will transform both the public and private cloud market.
Cisco will also announce a new hybrid-enterprise and public hybrid cloud solution—hybrid cloud on demand. With a new hybrid cloud on demand solution and the introduction of new VMware Storage Spaces and PowerEdge switches, this solution helps the enterprise run multiple clouds on the same hardware. By simplifying the management of complex clouds, the hybrid-enterprise solution extends the capabilities of virtual cloud systems from the core-to-edge cloud.
Hybrid-Enterprise Cloud solutions combine private and public clouds, and allow users to seamlessly migrate their applications between on-premises and public clouds. With the ability to scale private cloud resources and public cloud resources together, users can now run their compute and storage resources as part of a public cloud. This solution provides a much greater level of agility and performance than ever before.
The hybrid-enterprise on demand solution extends the capabilities of virtual cloud systems from the core-to-edge cloud. By integrating the storage cloud with the compute cloud, customers can run compute and storage resources as part of a public cloud, with compute and storage as virtual resources in isolation—the first of its kind solution to be offered by a public cloud vendor. With this new hybrid-enterprise solution, customers can now run a total of eight clouds, while maintaining the flexibility of each type of cloud on its own separate hardware and manage their resources for hybrid deployment.
The hybrid-enterprise solution brings a number of key benefits to customers, including improved performance, capacity, and cost-efficiency:.
Improved performance – The Hybrid-Enterprise solution includes VMware Storage Spaces and PowerEdge storage, offering improved performance with both virtualization and bare metal provisioning. This solution provides all the performance benefits of virtualization, without the management overhead and cost of virtualization.
Storage capacity- On the public cloud, users can now manage their total capacity for on-premises and hybrid cloud deployments with a single solution.