DOE Primes Pump for Exascale Supercomputers

By Michael Feldman

July 12, 2012

Intel, AMD, NVIDIA, and Whamcloud have been awarded tens of millions of dollars by the US Department of Energy (DOE) to kick-start research and development required to build exascale supercomputers. The work will be performed under the FastForward program, a joint effort run by the DOE Office of Science and the National Nuclear Security Administration (NNSA) that will focus on developing future hardware and software technologies capable of supporting such machines.

The program is being contracted through Lawrence Livermore National Security, LLC as part of a multi-lab consortium that includes Argonne National Laboratory, Lawrence Berkeley National Laboratory, Lawrence Livermore National Laboratory, Los Alamos National Laboratory, Oak Ridge National Laboratory, Pacific Northwest National Laboratory, and Sandia National Laboratories.

Although we’re only six to eight years away from the first exaflops systems, the DOE’s primary exascale program has yet to be funded. (And since this is an election year in the US, such funding will probably not fall into place until 2013.) In the interim, FastForward was devised in order to begin the needed R&D for some of the exascale foundational technologies, in particular, processors, memory and storage.

At least some of the impetus for the program came from the vendors themselves. According to Mark Seager, Intel’s CTO for the company’s High Performance Computing Ecosystem group, the DOE was told by multiple commercial partners that research for the component pieces needed to get underway this year if they hoped to field an exascale machine by 2020. That led to the formation of the program, and apparently there was enough loose change rolling around at the Office of Science and NNSA to fund this more modest effort.

Although all the FastForward subcontracts have yet to be made public, as of today there are four known awards:

  • Intel: $19 million for both processor and memory technologies
  • AMD: $12.6 million for processor and memory technologies
  • NVIDIA: $12 million for processor technology
  • Whamcloud (along with EMC, Cray and HDF Group): Unknown dollar amount for storage and I/O technologies

Although the work is not intended to fund the development of “near-term capabilities” that are already on vendors’ existing product roadmaps, all of this work will be based upon ongoing R&D efforts at these companies. The DOE is fine with this since the commercialization of these technologies is really the only way these government agencies can be assured of cost-effective exascale machines. The FastForward statement of work makes a point of spelling out this arrangement, thusly: “While DOE’s extreme-scale computer requirements are a driving factor, these projects must also exhibit the potential for technology adoption by broader segments of the market outside of DOE supercomputer installations.”

For example, Intel’s FastForward processor work will be based on the company’s MIC (Many Integrated Core) architecture, which the company is initially aiming at the supercomputing market, but with the intent to extend it into big data business applications and beyond. The first MIC product, under the Xeon Phi brand, is scheduled to be launched before the end of 2012, but this initial offering is at least a couple of generations away from supporting exascale-capable machines. According to Seager, a future processor of this kind will need much improved energy efficiency, a revamped memory interface, and higher resiliency.

Although the x86 ISA will be retained, this future MIC architecture will incorporate some “radical approaches” to bring the technology into the exascale realm. To begin with, says Seager, that means reducing its power draw two to three times greater than what would naturally be achieved with transistor shrinkage over the rest of the decade. “It’s a daunting challenge to do better than what Moore’s Law will give you,” Seager told HPCwire.

Fortunately, he says, Intel will be able to leverage its near-threshold voltage circuitry research, some of which was funded under UHPC (Ubiquitous High Performance Computing), DARPA’s now defunct exascale program. Shekhar Borkar, who was the PI for the UHPC work, along with Seager and former IBM’er Al Gara, will be heading up the FastForward work at Intel.

For the exascale memory subcontract, Intel will be leveraging its work with Micron Technology on the Hybrid Memory Cube. The idea is to use similar technology to incorporate 3D stacks of memory chips into the same package as the processor. In-package integration shortens the distance considerably between the processor and the memory, which significantly increases bandwidth and lowers latency. At the same time, cache management is going to be redesigned to optimize the power-performance of memory reads and writes.

Like Intel, AMD will be basing its FastForward processor research on a current design, in this case the company’s APU (Accelerated Processing Unit) product line and the related Heterogeneous Systems Architecture (HSA) standard — that according to Alan Lee, AMD’s corporate vice president for Advanced Research and Development. The current crop of APUs, which integrate CPUs and GPUs on-chip, are aimed at consumer devices, such as laptops, netbooks, and other mobile gear. But AMD has designs on extending its heterogeneous portfolio into the server arena, and the DOE just gave them about 12 million more reasons to do so.

Since AMD first needs to transform their APU into a server design, the chipmaker has a somewhat different, and perhaps longer path to exascale than Intel, which is at least starting with server-ready silicon. On the other hand the MIC architecture is not heterogenous (and may never be), so AMD does have a certain advantage there. “That is the truly unique technology and the strongest one that AMD brings to bear — that we have a world-class CPU and GPU brought together in a single APU,” says Lee.

Lee was less forthcoming about the starting point for the memory research under the FastForward work, other than to say it would be optimizing the technology around its heterogeneous architecture and would involve high-speed interconnects as well as different types and arrangements of memory.

More than anything, Lee sees this R&D work as producing dividends in other areas of AMD’s business. He says the fundamental technologies that the DOE wants for exascale are those the computer industry needs, not just in the future, but right now, referring to the big data domain, in particular. “I expect that a lot of the technology that you see us develop has the potential to make it into a variety of different server products of different genres,” says Lee.

To counterbalance the Intel and AMD work, is NVIDIA, which will be using the company’s Echelon design as the starting point for its FastForward work. Echelon, which was also funded under DARPA’s UHPC program, is based on a future 20-teraflop microprocessor that integrates 128 streaming processors, 8 latency (CPU-type) processors, and 256MB of SRAM memory on-chip. The technology is in line to follow Maxwell, NVIDIA’s GPU architecture scheduled to take the reigns from Kepler in a couple of years. Unlike the Intel and AMD efforts, NVIDIA’s contract is for processor technology only, although the Echelon design also specified an exascale-capable memory subsystem.

While the DOE spread its bets around for the FastForward processor- and memory-based research, there was only one storage subcontract awarded. That went to Whamcloud, who in conjunction with EMC, Cray and HDF Group, got the nod to provide the R&D work for storage and I/O.

The work specifies bringing object storage into the exascale realm, and will be based on the Lustre parallel file system technology. As a result, any development in this area will be open sourced and be available to the Lustre community.

Although the FastForward contracts limit their scope to specific exascale components, rather than complete systems, the research won’t be performed in a complete vacuum. The vendors are expected to work in conjunction with the DOE’s exascale co-design centers, a group that encapsulates various proxy applications, algorithms, and programming models important to the agency. The idea is to align the vendor R&D designs with the DOE’s application needs and expectations, the implication being that these are general enough to apply to a wide range of exascale codes both inside and outside the Energy Department.

All the FastForward contracts have a two-year lifetime, so are slated to expire in 2014. The follow-on DOE work to design and build entire exascale supercomputers are dependent on future budgets. Assuming the feds comes through with the funding, that effort is expected to cost hundreds of millions of dollars over the next several years.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire