How to Gain the Performance Needed for AI Applications

Long a staple of science fiction, artificial intelligence (AI) has seen rapid advances in recent years that have brought well within the mainstream. AI has the potential to transform every business sector, and shake up well-established industries with new business models. Research firm Tractica has identified nearly 200 real-world AI use cases and predicts the AI market will see year-over-year growth of more than 64 percent through 2025.

While chipsets such as NVIDIA’s GPUs have accelerated AI advancements, organizations need high-performance computing (HPC) systems to take full advantage of AI. To that end, HPE has introduced three new HPC systems that are optimized for scientific, engineering and big data workloads to deliver faster, more efficient insights.

HPE SGI 8600 System

The first co-created system since HPE acquired SGI, the HPE SGI 8600 is a liquid-cooled petascale system, based on the SGI ICE XA architecture, purpose-built to solve the most complex scientific, engineering and national security challenges. This solution delivers industry’s fastest parallel-processing performance, and scales to more than 10,000 nodes without additional switches using integrated switches and hypercube technology. Based on industry standards, the HPE SGI 8600 also offers more flexibility and choice than competing solutions.

HPE Apollo 6000 Gen10 System 

The next generation of HPE’s large, commercial, air-cooled HPC platform was redesigned to deliver more than 300 teraflops per rack, higher rack-scale efficiency and exceptional price performance. Key new capabilities of this system include:

  • Industry-leading reliability, accessibility, serviceability and manageability
  • Greater application licensing efficiency
  • Reduced latency and higher IOPs performance
  • Reduced power consumption and cooling requirements

HPE Apollo 10 Series

These new cost-effective platforms are optimized for entry-level deep learning and AI applications that are easy to manage and deploy. The HPE Apollo sx40 System is a 1U dual-socket Intel Xeon Gen10 server with support for up to four NVIDIA Tesla SXM2 GPUs with NVLink. The HPE Apollo pc40 System is a 1U dual-socket Intel Xeon Gen10 server with support for up to four PCIe GPU cards.

HPE’s HPC systems are enhanced by a new compute experience designed to address the data-intense needs of AI. The key elements of the HPC experience include:

  • Enhanced firmware security. HPE offers the first industry-standard servers with firmware security built directly into the silicon. Organizations can use HPE HPC systems with confidence, knowing that top-secret and highly confidential information is protected.
  • Increased agility with software-defined HPC capabilities. The new HPE Performance Software Suite for HPC features Core Stack 2.2, a pre-built HPC solution that brings together open-source, third-party and HPE-built components into a single integrated distribution.
  • Simplified management. HPE Insight Cluster Management Utility 8.2 and New HPE SGI Management Suite 3.5 are comprehensive system management solutions offering enhanced provisioning, management and monitoring to simplify operations. Validated, optimized and integrated third-party software solutions for workload management and application software development also enhance agility and productivity.

The business potential for AI is fueling demand for systems that can maximize compute performance, agility, security and flexibility while reducing energy consumption and space utilization. Let Sequel show you how HPE’s HPC systems can help you realize the potential of AI and prepare your organizations for the future.