Behind the scenes @ Labs
Showing results for 
Search instead for 
Did you mean: 

Labs takes a path forward with the U.S. Department of Energy




 By Curt Hopkins, Managing Editor

 HPE announced today that that it has been awarded a significant grant from the U.S. Department of Energy (DOE)  to design a Memory-Driven supercomputing reference architecture that will process over one quintillion (one billion billion) calculations per second.

By contrast, the world’s fastest computer is currently China’s Sunway TaihuLight at 93 petaflops or one thousand million calculations per second.

The grant is being issued under the PathForward program, which is a central component of DOE’s Exascale Computing Project. HPE is one of several companies charged with developing technology that will ultimately lead to a new type of computer, capable of performance that surpasses today’s most powerful supercomputers by an order of magnitude.

This development period runs through 2019, at which point the awardees’ contributions will be assessed. One or two companies will move on to the next stage of this three-stage drive. The goal is to create a functional exascale computer by 2022.

HPE’s part is a collaboration between the High Performance Computing Group and Hewlett Packard Labs. Paolo Faraboschi, an HPE Fellow in Systems Architecture at Labs, shared that Labs will be contributing technologies derived from The Machine research project and based on Memory-Driven Computing principles, including exascale notional design, strategy, and components.

Specifically, Labs will provide a new fabric – the memory fabric protocol that grew out of The Machine project – as the foundation for all messaging and memory traffic in the exascale architecture. With a partner, Labs is planning to design a compute node that has what Faraboschi’s team thinks is the best balance of memory, fabric connections, and FLOPS; and a new way to use Non-volatile Memory for the first tier of storage.

Designing for exascale problems is an exciting challenge for the Labs team. “We haven’t played at this level before, though we have a very strong presence in supercomputing,” said Faraboschi. He added that HPE researchers have the largest number of publications among competing companies.

Our work on The Machine and Memory-Driven Computing put us on the DOE’s radar screen,” said HPE Fellow Paolo Faraboschi. “[This work] established our credibility in this area.

The technologies that are emerging from that effort – NVM, memory fabrics, and others – secured HPE’s inclusion in the project.

 “This is a really exciting time to be involved in supercomputing,” said HPE’s Bill Mannel, president and general manager of High Performance Computing at HPE.

“I’m really looking forward to seeing our work come to reality, solving real problems for the benefit of all of us.”

 Exascale computing will enable a broad set of modeling and simulation applications that are unachievable today, accelerating scientific advancements in everything from the physics of star explosions to precision medicine for cancer.

What Labs brings to the table

  • A new and open memory-semantics system protocol
  • Interconnect topology and high-radix router, optimized for short messages
  • Co-packaged high-performance DRAM for the application working set
  • Fabric-attached nonvolatile memory converging memory and storage
  • Silicon photonics integrating a Silicon Photonic chip with microrings, a CMOS chip with the logic driver, and a fiber-attached device to enable connection to an external single-mode fiber cable

Learn more in this Q&A with HPE's Bill Mannel:

About the Author


Editor, Behind the Scenes @ Labs and Senior Manager, Innovation Marketing

June 18 - 20
Las Vegas, NV
HPE Discover 2019 Las Vegas
Learn about all things Discover 2019 in  Las Vegas, Nevada, June 18-20, 2019
Read more
Read for dates
HPE at 2019 Technology Events
Learn about the technology events where Hewlett Packard Enterprise will have a presence in 2019.
Read more
View all