What is HPE High Performance Computing
In This Article
For IT professionals and academics alike, the mention of high-performance computing (HPC) may still evoke images of room-sized Cray supercomputers, even as today's HPC offerings are more proportional and relevant--both in terms of physical size and applications--with the business needs and requirements of the modern enterprise.
When it comes to HPE's answer to HPC, its portfolio (which includes the current Cray supercomputer line) offers scale-up and scale-out capabilities for both on-premises, hybrid and public cloud-based environments, as well as purpose-built storage, software and workloads aligned with the use cases of today's organizations.
In the past, deploying on-premises HPC infrastructure was a challenging endeavor that involved high costs, steep learning curves, limited growth and expansion capabilities, and ultimately a low return on investment. For this reason, HPC was typically relegated to academia and government research.
HPE HPC marries supercomputing and the cloud, allowing organizations to capitalize on advanced HPC systems for high performance while capitalizing on elastic, metered cloud services. The product line is designed to offer supercomputing capabilities to organizations alongside the typical benefits of the cloud: agility, ease of management and flexible economic consumption models, to name a few--as well as advanced data analytics, artificial intelligence and machine learning (AI/ML), and deep learning.
By leveraging HPE HPC solutions, organizations can access HPC infrastructure for creating and improving data analytics, powering AI/ML use cases, and optimizing business outcomes with advanced modeling and forecasting capabilities.
How Does HPE High Performance Computing work?
HPE HPC enables organizations to assemble HPE's infrastructure building block components--dense compute resources, high-speed and high-capacity storage, networking, and software--for running and managing their own HPC clusters. The following are some highlights from the HPE HPC portfolio:
HPE Cray Exascale Supercomputers
HPE Cray supercomputers feature multiple processor architectures, various accelerator options and architectures that support forward compatibility with future HPE hardware (e.g., next-generation blades and servers).
HPE Cray supercomputers are available in two configurations:
- HPE Cray EX: the HPE Cray EX features high density and efficiency levels, liquid-cooled cabinetry and direct liquid cooling of all components in its highly dense, blade-optimized configuration.
- HPE Cray Supercomputer: for smaller-scale deployments that only require air cooling, the basic HPE Cray offering features a standard 19" rack configuration, a 2U compute form factor, and HPE Cray software and HPE Slingshot networking.
HPE Apollo Systems
HPE Apollo systems were designed to be plug-and-play, scalable and highly modular for deployment in data centers of all sizes. These efficient, rack-scale offerings give organizations an ideal mix of performance and adaptability and flexibility for a myriad of HPC and AI workloads.
HPE Slingshot is HPE's high-speed, purpose-built supercomputing interconnect that enables high speed and bandwidth, low-latency, large-volume data transfers and operations for HPE. Slingshot supports Ethernet connectivity, sophisticated congestion control and advanced adaptive routing, and other features for meeting high quality of service (QoS) levels.
HPE GreenLake is an edge-to-cloud solution that offers workload-optimized systems in a metered, pay-per-use self-service platform. With GreenLake, organizations can leverage HPC via a cloud service consumption model--even with on-premises hardware--thereby reducing the complexity and costs of managing an HPC environment.
Accelerating digital evolution
HPE's HPC portfolio empowers organizations to innovate freely by offering them the best that HPC has to offer on the market, while at the same time keeping management complexity and cost to a minimum. For example, AI-enabled enterprises can take advantage of HPE HPC's general-purpose graphics processing units (GPGPU) and accelerator solutions for enhancing performance matrix and tensor calculations typical of AI deep learning applications.
Efficient data center solutions
HPE offers HPC solutions that allows for more efficient and streamlined data center operations. For example, the HPE HPC line's high-density systems allow for the placement of more servers into data center racks (when compared to standard enterprise servers). Additionally, its servers use cooling technology that adjusts dynamically with the number of HPC servers in operation, as well as energy-efficient CPUs designed for compute-intensive workloads.
Support for on-premises HPC to the private cloud
HPE's HPC infrastructure was designed to support IT environments of all types and compositions--be it the colocation facility, the edge, or corporate data center. And for enterprises with hybrid multi-cloud strategies, HPE HPC portfolio offerings like the GreenLake edge-to-cloud platform allow for the reserving and running of workloads, on-demand.
Powering high-value infrastructures and app services
HPE HPC portfolio offerings such as the HPE Cray Supercomputers were designed to handle modern use cases such as modeling, simulation, AI/deep learning, and analytics workloads, with advanced capabilities for workload management, container support, and flexible orchestration. And for protecting high-value infrastructures and apps, HPE HPC features built-in silicon root of trust security--an embedded security platform for automatically protecting infrastructure continuously.
HPE offers a wide range of products in its HPC portfolio--to find out which options are the most appropriate for your organization, speak with a WWT expert today or visit our labs to learn about HPE's capabilities.