June 22, 2018 —
As the industry leader in High Performance Computing (HPC), HPE understands the complexities you face in adopting, scaling and optimizing HPC and Artificial Intelligence (AI). Leveraging the breadth of our portfolio and partner ecosystem, we address these challenges by providing proven and tested server, storage, and software and services capabilities to help you achieve optimal performance— while maintaining a laser focus on system efficiency, manageability, security and interoperability.
That’s why we’re very excited to be premiering new product offerings at International Supercomputing Conference (ISC) in Frankfurt on June 24-28. Here’s what to look for.
HPE Performance Cluster Manager
We are introducing the new HPE Performance Cluster Manager which delivers a fully integrated system management for all Linux® based HPE high performance computing systems. The software can support installations from a single node theoretically up to 100,000 nodes. HPE Performance Cluster Manager offers:
- Fast system setup from bare metal
- Comprehensive hardware monitoring and management
- Image management and software updates
- Power management
Customers using this system management solution will be able lower their total cost of ownership by reducing the time and resources spent administering HPC systems. The software supports the combined HPE and SGI installed base of customers using HPE Apollo, HPE SGI 8600, HPE ProLiant and HPE Moonshot, SGI ICE X, SGI ICE XA and SGI Rackable systems. The software combines and improves capabilities of two existing cluster management solutions: HPE Insight Cluster Manager and the HPE SGI Management Suite (formerly known as SGI Management Suite).
Hybrid HPC and Infrastructure-as-a-Service
Hybrid HPC is emerging as a key strategy to allow organizations to seamlessly tap into existing on-premises resources as well as scale resources on-demand in the cloud, providing new levels of flexibility, efficiency, and agility. HPE has a number of new programs and technologies to help you deploy a Hybrid HPC infrastructure while continuing to benefit from HPE’s industry-leading combination of products, applications and workload expertise.
HPE is working with Intel to develop an Infrastructure-as-a-Service (IaaS) offering for HPC clouds with Intel Xeon Scalable Processors and Omni-Path Architecture (OPA) as key ingredients. Intel’s innovative HPC foundation, based on the Intel Xeon Scalable processors, includes critical platform innovations in memory, storage and acceleration technologies to address the complex spectrum of diverse HPC workload requirements. Further, given the criticality of HPC fabric in scale out deployments, Intel OPA provides low latency interconnect for scalable performance of multi-node environments.
IaaS is a layer that enables you to to host IaaS services on your own infrastructure in several ways:
- As a private cloud
- On dedicated infrastructure from a third party as a managed private cloud
- On large-scale shared infrastructure from a third party as a public cloud
The HPE IaaS offering integrates traditional HPC components into a standard cloud IaaS framework and empowers HPC cluster provisioning and utilization across all operating environments.
HPE Apollo 6500 Gen10
The new HPE Apollo 6500 Gen10 was introduced at the GPU Technology Conference in March and will now be making its debut at ISC to the HPC community. With eight high-performance NVIDIA V100 Tensor Core GPUs per server, the HPE Apollo 6500 Gen10 System provides superior performance per dollar for HPC and AI workloads—delivering up to 62 TFLOPs double-precision and up to 1 PFLOPs of mixed-precision AI compute. Purpose-built for accelerated computing, this platform features both PCIe and NVIDIA NVLink GPU interconnects, providing the flexibility to suit a wide variety of accelerated computing requirements. GPU support has become mainstream for most HPC applications, including fluid dynamics, structural analysis, visualization, and chemistry and biosciences applications. It can reduce computational times by orders of magnitude.
HPE Apollo 35
The HPE Apollo 35 provides a high density AMD EPYC compute solution with exceptional memory bandwidth and capacity per socket. Offering 4 dual processor servers in 2U, with up to 32 high-performance “Zen” cores and 2TB per socket, as well as 128 PCIe lanes, the Apollo 35 promises to be ideal for memory bandwidth or capacity bound HPC workloads, such as computational fluid dynamics, weather simulation, and oil and gas exploration. The Apollo 35 offers high efficiency through core and memory density, with 2 processors per server, up to 8 processors per system, and up to 80 servers per rack. Similarly, with 2TB DRAM per socket, the Apollo 35 offers up to 4TB DRAM per server, up to 16TB DRAM per system, and up to 320TB DRAM per rack. Combined with support for high-bandwidth network interfaces, the Apollo 35 delivers maximum HPC performance at both a node and rack level.
HPE partners to provide storage offerings to boost performance
WekaIO Matrix is a flash-optimized high-performance storage solution, designed for next-generation workloads with large, diverse data sets and high compute requirements in HPC, HPDA and AI. With a modern design which does not rely on legacy algorithms or assumptions, it leverages a scalable parallel file system with high speed networks to remove data locality as an issue, and integrates all-flash arrays with the scalability and economics of object and cloud storage. WekaIO Matrix became available on the HPE price list in early May 2018 and is now being resold by HPE.
As HPC clusters handle larger volumes of data, storage requirements grow more complex and drive the need for performance boosts across memory and storage. HPE teams with Intel and provides Intel 3D NAND SSDs and Intel Optane SSDs offerings to HPC users. This delivers an industry-leading combination of high throughput, low latency, high Quality of Service (QoS) and ultra-high endurance to break through data access bottlenecks by providing a new data storage tier for HPC workloads.
HPE Data Management Framework
HPD Data Management Framework, with a decade long history, will feature a new distributed architecture near the end of this year, providing enhanced scalability, performance and security. Data workflows will be streamlined and simplified utilizing new metadata capabilities and job scheduler integration functionality. Enhanced integration with Lustre will reduce complexity, while petabyte scale backup and restore capabilities will feature new data versioning and point-in-time restoration capabilities. This upcoming version of HPE Data Management Framework will provide a data management platform capable of optimizing storage infrastructure utilization and minimizing TCO while supporting next generation workloads, such as Deep Learning and genomics, involving data sets with thousands of files.
Join us at ISC
I’ve hit the highlights of what’s going to be on display at the HPE booth E-1030 at ISC 2018. Product experts will be on hand to help you learn more about these innovative new additions to our already industry-leading HPC portfolio. Additionally, we will also be showcasing a number of other key leadership HPC products, solutions and services including: HPE SGI 8600, HPE Apollo 6000, HPE Apollo 2000, HPE Apollo 70. Experts will be on hand to help customers accelerate the AI journey.
Don’t miss our executive presentations:
- Bill Mannel, Vice President & GM, HPC & AI at HPE, speaks at the Exhibitor forum: Tuesday June 26 at 2:20 p.m.
- EngLim Goh, Vice President & CTO, HPC & AI at HPE, speaks along with four others at the Digital Twins session: Tuesday, June 26 from 11:00 a.m. – 12:30 p.m.
- Jean-Luc Assor speaks at the Vendor Showdown: Monday June 25 at 1:05 p.m.
We are looking forward to another successful ISC and the opportunity to again demonstrate how HPE delivers tomorrow’s technology today.
Source: Bill Mannel, HPE blog