Hewlett Packard Enterprise (HPE) today launched a new liquid cooled GPU-driven Apollo platform based on SGI ICE architecture, a new collaboration with NVIDIA, and formal support for NVIDIA’s just announced Volta GPU architecture – all as part of a barrage of announcements made at GTC. In short order GTC has become an important launch venue for a broadening spectrum of AI/deep learning-related offerings that leverage accelerator technology. GTC’s growing prominence also showcases, at least for present, NVIDIA’s dominance.

Among HPE announcement highlights are:

• The new HPE SGI 8600 is based on the SGI ICE XA architecture with support for optimal combination of liquid-cooled GPU performance with NVIDIA Tesla GPU accelerators with NVLink interconnect technology. HPE says the new server provides scale and efficiency for complex, largest environments – “up to thousands of nodes with leading power efficiency.”
• Interactive rendering capability from a central datacenter with the HPE Apollo 6500 and NVIDIA Tesla GPUs certified with NVIDIA VCA software.
• Support for NVIDIA’s next generation Tesla GPUs based on its Volta architecture when available in production quantities in the Apollo 2000, Apollo 6500 and Proliant DL380 servers.
• Collaboration with Kinetica (GPU-accelerated and analytics database) to develop a real-time fraud detection product with GPU acceleration. Designed specifically for consumer credit card transaction processing, the new solution, will be demoed at the HPE booth during GTC

“Customers pursuing Deep Learning projects face a variety of challenges including a lack of mature IT infrastructure and technology capabilities leading to poor performance, efficiency and time to value,” said Bill Mannel, vice president and general manager, High Performance Computing and Artificial Intelligence at Hewlett Packard Enterprise in the formal press release. “To address these challenges, HPE is introducing new optimized GPU compute platforms, an enhanced collaboration with NVIDIA and HPE Pointnext Services from the Core Datacenter to the Intelligent Edge.”

The beefed up HPE-NVIDIA collaboration is interesting. The partners say they will “jointly address GPU technology integration and deep learning expertise challenges to accelerate the adoption of technologies that provide real-time insights from massive data volumes.” The combination of HPE’s purchase of SGI and Tokyo Institute of Technology’s choice of SGI for its next TSUBAME machine, which is focused on AI/deep learning and leverages NVIDIA GPUS, was a natural push for a deeper relationship.

“As the artificial intelligence era takes hold, enterprises are increasingly adopting NVIDIA’s GPU computing platform to generate insights from decades of untapped data,” said Ian Buck, General Manager of Accelerated Computing at NVIDIA. “Expanding our collaboration with HPE around deep learning will help enterprises deploy, manage and optimize their GPU computing infrastructure and realize the benefits of AI and deep learning in their business.”

The planned Tokyo supercomputer is one of the largest NVIDIA Tesla P100 GPU based clusters. HPE and NVIDIA say the collaboration will deliver:

• Enhanced Centers of Excellence for benchmarking, code modernization 
and proof of concept initiatives. The locations include Korea, Sydney, Grenoble, Bangalore and Houston
• Early access program for Volta-based NVIDIA Tesla SXM2 GPU systems powered with eight GPUs for selected customers in 4Q 2017

Satoshi Matsuoka, Professor and TSUBAME Leader, Tokyo Institute of Technology described the new machine as a converged world-leading HPC and deep learning platform. “The NVIDIA Tesla P100 SXM2 node solution enables GPU based Deep Learning capability to be scalable to the entire size of our TSUBAME 3.0 system. We look forward to continuing our partnership with HPE to work together on future projects in HPC and Deep Learning,” said Matsuoka.

HPE used the announcement as an opportunity to promote its new technology services organization – Pointnext – introduced in March. Pointnext combined consulting and support organizations under one umbrella. It has three main offerings; advisory services, professional services, and operational services.

Pointnext is described as an “IT services organization built to make Hybrid IT simple and power the Intelligent Edge. As an agile technology partner, we help you to modernize your legacy infrastructure with the flexibility of the cloud, and maximize the value of your connected devices.” HPE reports Pointcast has roughly 25,000 “specialists” in 80 countries.

Also noteworthy is the Kinetica collaboration. Like virtually all HPC technology system providers seeking expansion into the enterprise, HPE is working on “vertical” solutions for particular workflows. Kinetica expertise should help HPE implement such systems.

“We look forward to advancing the Kinetica GPU database with HPE and jointly offering a best-of-breed GPU-accelerated analytics solution that converges Artificial Intelligence and Business Intelligence workloads for financial services as well as for retail, healthcare and other industries,” said Chris Prendergast, vice president of business development and alliances, Kinetica.

Steve Conway, SVP for research, market watcher Hyperion Research, said, “With the need to embed more intelligence and automation into data analytics to address scientific and business challenges, artificial intelligence-based techniques are growing in importance. HPE’s systems and solutions innovations announced today are designed to address key performance and expertise constraints affecting deep learning.”