Expanding the HPC Ecosystem

By Steve Conway

July 28, 2006

Over the past three years, the Council on Competitiveness has sponsored pioneering studies and conferences on the relationship between HPC and business competitiveness, under the direction of Council Vice President Suzy Tichenor. In January 2006, Bob Graybill, former DARPA HPCS program manager and current division director of USC's Information Sciences Institute, became a senior advisor to the Council. He is helping to guide the Council's HPC Initiative as it works to link together government, academic and business organizations in a “national ecosystem” aimed at advancing innovation and competitiveness through greater use of HPC.

In this exclusive HPCwire interview, Tichenor and Graybill discuss the importance of HPC for businesses and preview information that will be disclosed in more detail at the Council's annual HPC Users Conference on September 7.

HPCwire: When and why did the Council begin looking at the importance of HPC in the private sector?

Tichenor: We've been looking at this for several years, as an outgrowth of our work in innovation. The Council believes that for the U.S. to remain preeminent in global markets, to increase productivity and raise our standard-of-living, we as a nation must become more innovation-based. If work is routine, rule-based, if it can be digitized and reliably codified, there's going to be a low-cost source of labor somewhere in the world to compete for that work and for those jobs.

Our competitive strength is in our ability to be more innovative. So the questions becomes, how do you promote, finance and educate for innovation, and what kind of infrastructure is needed to support this? That's where HPC comes in. We believe there's a need for pervasive access to and use of supercomputing. Three years ago, we launched an initiative to identify how HPC is really being used by businesses, how this is linked to innovation and what challenges prevent wider use of HPC in the business sector.

HPCwire: In a nutshell, what did you find?

Tichenor: First and foremost, we found that for companies that rely on it today, HPC is absolutely essential to business survival. It's not just a “nice to have” tool. We also identified some challenges to more widespread adoption of HPC by business organizations. There's a need for more production-quality application software, for better interfaces, and for more people who know how to use HPC as a production tool. Businesses are telling us that access to talent is a pacing item. This brought us to the issue of education and how to make people more comfortable using HPC.

HPCwire: Was any of this surprising?

Tichenor: Some people who've been immersed in HPC for years have understood the situation, but our studies were the first in-depth, market-based research findings on this topic, and they surprised many people. A lot of thinking in HPC has been focused on how to build better computer systems. The Council is more interested in how these systems can be used most effectively to drive business success and competitiveness.

Unfortunately, HPC is still a niche market within the overall computing market. Our research looked at the full spectrum of HPC users and discovered a bimodal pattern. There is a small group of high-end users and a much larger group of entry-level users, but not many in the middle. We call this gap in the spectrum the “missing middle.” There is another large group of people who are doing technical computing on the desktop, but haven't used HPC and don't understand its benefits. We call this group the “never evers.” The Council is not only trying to address the important needs of high-end HPC users, we are also focusing on how to fill the “missing middle” and how to encourage the “never evers” to adopt HPC for greater competitiveness.

HPCwire: So, how can you fill in the “missing middle” of the market and convince the “never evers” to use HPC?

Tichenor: That's what we're exploring now. We have a tremendous HPC Advisory Committee that has been meeting for a few years now. This is a brain trust of senior executives from the government, academia, private industry, vendors and other key constituencies. Collectively, we see a need to develop mechanisms for reaching out to the “never evers” and entry-level HPC users, so we can expose them more to HPC and grow the market. This will require interesting partnerships that connect the business community with universities, national labs and other parties, not just for access to cycles, but also for access to expertise. The Council and our HPC Advisory Committee want to figure out how to leverage this expertise to provide greater ROI for the country. We will launch pilot programs to help introduce companies to HPC and do many other things that Bob can talk more about.

One reason we're so excited is that we believe HPC is undervalued in many regions of the U.S. Many businesses and other organizations are not aware that we have these HPC assets, including the on-demand services being developed by some vendors, that could help stimulate regional economic development. The Council has a significant program on regional economic development, and we want to see HPC integrated into these regional development plans. HPC can also be a tool for attracting companies to a region.

There are already some interesting partnership models out there. Exploring these models is one of the themes of the Council's September 7 HPC Users Conference in Washington. We recently did two surveys for the NSF and DOE's NNSA, to look at where their HPC-related partnerships with industry have been successful and where the stumbling blocks are. Overall, it turns out that these public-private partnerships have been tremendously successful, yet many of the participating businesses said they were unaware of these valuable HPC resources before starting the program. We need to change that, because these partnerships are a win-win for everyone. The businesses become more competitive through their interactions with the universities and labs and their access to more advanced HPC systems and expertise, and the sponsors advance their own problem solving through techniques they learn from industry.

HPCwire: Are there other areas for expanding HPC usage that you are exploring?

Tichenor: We see strong potential for extending HPC usage through the supply chain, wherever appropriate. This began happening years ago in some more mature HPC markets, such as the automotive sector. In other sectors today, however, large companies use HPC but their suppliers don't.

In our September conference, we'll have a number of companies speaking about this, including Wal-Mart, whose requirements drive product development for many of its suppliers. For example, Wal-Mart might require a large consumer products company to re-think it's packaging, and the consumer products company might then have to meet with its suppliers. Some consumer products companies already use HPC. You've probably heard the example of Procter & Gamble using HPC to redo the manufacturing process for Pringles. We want to explore how these firms can extend that expertise to their suppliers to make the entire supply chain more competitive. There is also an important need for HPC in optimizing the entire supply chain process. This will also be discussed in our conference.

HPCwire: Bob, you had a major impact in shaping the HPC industry at DARPA and then became division director of USC's Information Sciences Institute. How did your connection with the Council happen?

Graybill: Based on my prior experiences, especially at DARPA, I saw an opportunity to help the U.S. private sector exploit HPC more fully for greater competitiveness. As you know, an important goal of DARPA's HPCS program has been to develop a commercially viable HPC system that can deliver breakthrough sustained performance and productivity across a spectrum of national security and other applications, including applications important to industry.

I approached various organizations to gauge their interest in helping me explore this opportunity. The Council was very interested, which wasn't really surprising, given the strong groundwork they had laid through research and discussions during the past three years. During that time the Council did a tremendous job of fact-finding, and this investigative work will be an ongoing effort. The natural progression, however, was to ask, “Now what do we do with all this information?” Based on the Council's findings and the bird's eye view of the HPC industry I acquired at DARPA, it was clear that, at least where industry was concerned, HPC might remain a niche market forever. If that happens, our companies and the country will lose out on a real opportunity to accelerate innovation and competitiveness. We need to actively work to create a national HPC ecosystem that businesses can use.

The Council is an ideal starting point for this initiative. Through their HPC Advisory Committee and their staff, the Council is working with the highest levels of industry, government decision-makers and labs, academia, vendors, and other HPC stakeholders. Equally important, the Council understands the need to make a business case for an HPC ecosystem. Recent studies the Council has undertaken, including the new NSF and NNSA studies that will be discussed for the first time at the Council's HPC Users Conference on September 7, show that there are already some successful models that the business ecosystem could expand on.

HPCwire: What is your role at the Council, Bob?

Graybill: My role is as a senior advisor and my objective is to work with the Council to drive the formation of this ecosystem, which we're calling the National Innovation Collaboration Ecosystem, or NICE for short. NICE will link together the key HPC constituencies to share expertise and thinking, including organizations from government, academia and industry, vendors and others. It will also serve as an information exchange to help businesses gain access to HPC hardware, software, networking resources, and expertise.

The aim of NICE is to boost the global competitiveness of U.S. businesses by creating a collaborative HPC infrastructure that will help our firms transform ideas into usable products. U.S. businesses can't compete globally based on hourly labor rates, as Suzy said; we have to compete through ideas and innovation. Leading U.S. corporations in a variety of industries are already doing this by using HPC for virtual prototyping, “what-if” analyses and other forms of modeling and simulation. We need to make sure all companies, regardless of size, have the kind of ISV software, expertise and other HPC resources they need to remain competitive, and we need to encourage more pervasive use of HPC throughout the supply chain. As part of the NICE initiative, we also need to do more to promote HPC as an important, exciting career path in our high schools and universities. We need to renew that talent stream for the future.

HPCwire: How will you go about organizing the National Innovation Collaboration Ecosystem?

Graybill: Again, the Council provides a strong starting point. They work with many government agencies, universities, and private sector industries. The Council's HPC Advisory Committee will serve as a brain trust and provide valuable oversight. Through my time at DARPA, I have gained considerable experience working with a diverse community in support of a common goal. It's essential to bring together many organizations and to have all the major constituencies involved.

There are six key organizing areas. First, we need to understand and incorporate the dynamics of the market and the users. Second, we need to do industry pilot studies. Third, for the purposes of this ecosystem we need to converge on a backbone infrastructure for high performance computing and communication that's standards-based to the extent possible. After that, the next step is to create an HPC Innovation Service Portal where businesses, especially the “never-evers” and entry-level HPC users, can access HPC expertise, cycles, etc. Some of these companies don't need to use HPC every day and can't justify creating their own infrastructures.

The fifth key element is robust applications software that's scalable and ready to use. This will be challenging to accomplish, but the Council's studies have shown that improving applications software is extremely important for industry, and a large majority of the ISVs and businesses that were surveyed are willing to partner with outside organizations to improve the software. We will be a catalyst to help make this happen.

Last and definitely not least, the NICE ecosystem needs to focus on training and education. We need to reach out to universities and high schools, to help create new generations of students who are excited about careers in HPC and are prepared to help advance the HPC industry and U.S. competitiveness.

HPCwire: How would the ecosystem be funded and managed for the longer term?

Graybill: Various players have strong interest in different parts of the ecosystem. The idea is to get them to lead in their areas of interest. Where long-term management of the ecosystem is concerned, we'll need to see along the way where that would best reside. The Council's job and my job is to get the ball rolling and stay involved. For that to happen, all six of the areas I described need to move forward in parallel. All of us who've been connected to the HPC industry for a while realize that moving hardware forward faster than software is problematic, and vice versa. All of the elements must work and evolve together.

Again, we're not starting from scratch. The Council has done a lot of related work through its conferences and studies, and we're aware of some public-private sector programs in the U.S. that could serve as effective models for what we're aiming to do on a larger scale.

HPCwire: How unusual is it to have a formal initiative to help meet the private sector's HPC needs? Are other countries doing this?

Graybill: To our knowledge, no other country is taking an approach to helping drive private-sector innovation and competitiveness that's as holistic and comprehensive as ours. Piecemeal approaches don't work. You can't focus just on innovation or software or cycles. You have to move all the elements forward together. As the Council's HPC Advisory Committee said, it's a hard problem but we know of no other way of solving it.

HPCwire: Can you tell me more about the HPC Advisory Committee's role vis-a-vis the NICE initiative?

Graybill: We view the HPC Advisory Committee as a brain trust from both the business and technical perspective. We want them engaged at the business level, and we would like the key technical people from their organizations to be involved actively in our workshops and other activities.

HPCwire: In his State of the Union address, President Bush proposed substantially increasing funding for supercomputing and basic science. Work done by the Council and by groups like HECRTF that you, Bob, were heavily involved with, helped make HPC a higher priority for Congress and the Administration. To what extent would this increased funding help the private sector?

Graybill: The increased investment proposed in the President's American Competitiveness Initiative for basic research in the physical sciences and engineering will help enhance U.S. innovation capacity and stimulate the breakthroughs that drive new product development, economic growth and competitiveness. We need these investments because our competition is not standing still. Other countries have also recognized the linkages between increased innovation and competitive gain, and are making their own investments. If we stand still, we will fall behind. The HPCS program has done a great job of focusing on productivity and on sustained performance. In the future, it is critical that the U.S focus less on peak flops and more on the whole HPC ecosystem in order to accelerate private and public sector innovation.

Tichenor: The government investment in HPC helps the business sector in several ways. First, programs like HPCS provide critical cost-sharing opportunities to advance HPC R&D. When you have a market as small as HPC, it's hard for vendors to garner enough revenue to make these major R&D investments alone. And these investments are very important because high-end users have unsolved problems that require us to keep pushing the technology envelope. Petascale problems aren't limited to government scientific research. We've published a number of case studies that confirm that industry also has problems needing petascale computing. They exist in the oil industry, the automotive and aerospace industries, and elsewhere.

Second, the government's investment in purchasing HPC systems is also crucial. It not only helps the government meet its mission-critical requirements, it also provides an important revenue stream to the HPC vendors so they can invest more in R&D for future-generation HPC systems. Additionally, the government is usually the first to purchase the most advanced systems. As aggressive users with highly complex problems to solve, they push and prove out the technology, providing valuable information back to the hardware and software developers. The developers use this information to make more useable and affordable products, enabling wider adoption of this technology across the private sector. This in turn helps to grow the market and increase our competitiveness. And the healthy cycle continues.

HPCwire: To what extent do businesses have access to the really big government systems?

Tichenor: More and more. Last year, for example, DOE's INCITE program was opened up for the first time to participation by the business community. This program is extremely valuable, because it provides access to some of the most powerful supercomputers in the country, systems that industry cannot afford to purchase at this time. Four companies, in addition to a number of universities, passed the rigorous DOE selection process and received large allocations of time on these advanced DOE systems. INCITE really helps the U.S. to leverage some of its largest HPC assets for an additional competitive lift to the country.

HPCwire: This may sound like a softball question, but why is it important for the government and academic community to learn about the HPC needs of businesses?

Graybill: They have more in common than they sometimes realize. It takes much too long for an initial idea to get developed by a university, picked up by a lab and eventually enter industry. To reduce the cycle time, we need to get all these parties engaged in a collaborative environment.

Tichenor: The HPC needs of government, academia and business are interrelated and interdependent. All of these parties are ultimately in this together, so they need to share perspectives and progress. Businesses are often solving problems that are similar to those that government researchers are tackling, so there are opportunities to cross-pollinate. Government and university researchers have lessons they can share with business about doing high-end work in a research environment, and businesses can share some things with government and academic people about applying research in real-world production environments. As Bob mentioned earlier, our national security is inextricably linked to our economic strength. Increasingly, our economic strength will be tied to our ability to out-innovate and out-compute, and HPC is critical for this.

HPCwire: There is a general perception that the U.S. is globally dominant in HPC technology and applications. What is your assessment of U.S. HPC competitiveness today? Are we in danger of falling behind other countries?

Graybill: The U.S. is the overall leader today, but the recent example of the Earth Simulator, and ambitious petascale plans by several nations, remind us that we need to remain vigilant and committed to HPC leadership. It's very encouraging to see U.S. petascale initiatives proceeding at DARPA HPCS, DOE and the NSF. None of these initiatives is aimed at advancing HPC technology for its own sake, however. The goal is to improve our ability to solve problems for government, science and industry. In that context, as Suzy mentioned, we as a nation still face serious challenges in HPC applications software, renewing the talent stream, etc. The NICE ecosystem will help keep HPC technology and human resources aligned with the problems, so the U.S. can remain a leader in both HPC technology and problem solving.

Tichenor: As Bob mentioned earlier, standing still is falling behind. And Japan not only has aggressive development plans, they also recognize the importance of using HPC across industry. Supercomputing is at the top of the list of that country's top 10 science goals, and government documents indicate the linkage between these goals and the international competitiveness of its industries.

This is an important indicator that the competitiveness game is changing. It's not enough to just make the most powerful computers. Competitive advantage comes from using them to solve complex industrial problems that permit companies to achieve and maintain leadership in the highly competitive global marketplace.

HPCwire: On September 7, the Council will hold its third annual HPC Users Conference in Washington. What will you focus on this year?

Tichenor: This year's conference will be particularly interesting because it will set the stage for the next phase of the Council's HPC work. The conference will move this work from a research-and-planning mode to an implementation mode. It will really set the stage for the NICE initiative. The dialogue won't be as much about what the problems are, as about the path forward, about the pilot programs and investment programs and the best business models that are needed to keep HPC healthy in the U.S. We'll talk about the roles of the public and private sectors, and how to make HPC a win-win for the country.

We're going to look at the collaborative models that have worked well and at where adjustments need to be made. We need to leverage the country's tremendous investment in HPC facilities and expertise, and look at how to drive HPC more broadly across the private sector.

As we've tried to do in past, we'll bring people to the conference that you don't typically hear speak at HPC conferences. These people have important insights into how we can best leverage this technology for competitive advantage.

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