The U.S. Presidential Election – What’s at stake for HPC and Exascale?

Print Friendly, PDF & Email

Article by Mike Bernhardt and Doug Black

This question generated a flood of responses — and most asked to remain anonymous. It appears that politics sometimes conflicts with free speech, especially if you work for an organization that relies on federal funding. So when you see an unattributed comment, we are simply honoring the respondent’s request.

Thanks to everyone who contributed to this issue. We can’t possibly use all the comments we received, and quite frankly, some of the responses, as much as they made us laugh, really wouldn’t be appropriate to print.

Why is this discussion important?

While there are tens of thousands of people working within the HPC and emerging exascale ecosystem, these discussions often lose sight of the millions of people who benefit from applied research and breakthroughs in areas such as drug design, automobile and aircraft design and testing, weather forecasting and disaster prediction, military and defense, and finance and economics – the areas that rely heavily on high performance computing. There truly is so much at stake. The impact of HPC and exascale affects millions of people on a daily basis.

As we reviewed the responses from the community, we placed them in several categories based on the emotional tone of the response. We did not pre-determine these categories, we developed them as we read the responses. In a general sense, here’s how the responses stacked up on an emotional scale.

Frustration 54%
Fear 29%
Hope 11%
Anger 6%

The first of our anonymous quotes:

“As a government funded research organization, we have been operating without a real budget for more than three years now. We live quarter-to-quarter never knowing if we will have a budget to even make payroll during the next three month period. We were told we had a 3-year budget through our Federal funding agency, but then after program award, were told we’d have to go back and fight for each quarterly increment. We spend as much time preparing justification reports and fighting for survival as we do actually trying to advance scientific research. It was frustrating a year ago. Now it’s just ridiculous. How can researchers and computational scientists give 100% effort toward deep thinking around various technology issues when in the back of their minds, they are continually wondering if they will even be employed in a few months. Politicians on both sides of the fence say they support HPC, but it appears to be nothing more than lip service. We are just not seeing the commitment to the kind of long term strategy needed to drive exascale – or for that matter, even advanced HPC research. We are all at the mercy of what the hardware vendors can build – what they believe will drive product sales in the near term. HPC innovation and exascale development require strong, unified Federal funding, and this goes hand in hand with economic recovery. And, without adequate funding and a strong commitment to technology leadership, I fear the U.S. may not even finish in this world-changing race.”

Will the U.S. presidential election have an impact on HPC and exascale?
According to 38% of our respondents, the answer is No.

“The problem is deeper than the agenda of either presidential candidate. The problem comes from the lack of congressional commitment to very difficult and long-term research, and the fact that the science and technology leadership fails to connect the dots and recognize that economic recovery could very well be fueled by HPC innovation.”

But perhaps the community as a whole has not done enough to present the case for why exascale is so vitally important.

According to IDC vice president Earl Joseph, “I think that regardless of the elections, the case for exascale needs to be made in clear and simple terms. Otherwise funding will be slow and limited. The case needs to be very clear on what exascale will provide as end results, like specific new discoveries, economic growth, job retention, etc. The key is to have specific examples that congress on both sides will agree are important.

Yet, a number of influential community leaders believe there is adequate bi-partisan support for HPC and exascale research to stabilize the exascale funding issue long-term. They don’t think the issue is with Congress. They point to a different problem holding the U.S. back. This group points the finger at ineffective leadership for nurturing computational science and technology growth from the Office of Science and Technology Policy (OSTP). They also say there is disagreement and disconnect when it comes to priorities and funding goals among the decision makers at OSTP, OMB and DOE.

Lots of finger pointing and no one in charge.” Obama explicitly mentioned exascale a few years ago. But that initiative seems to have evaporated. Is anything being done? Sure, as we’ve pointed out in past issues, there are a number of [disjointed] innovative research projects underway. But how do they tie together? The future of HPC and exascale appears to be a horse race without a track. (We will explore this further in a future issue.)

The counterargument to this view of course, if in fact this is the case, is that a change in the White House might be a good thing. Bringing in a new president would result in new leadership at many levels, very likely working its way down to the various science and technology leaders who some think represent the bottleneck in U.S. progress. But with that kind of change also comes a very big risk. Science and technology research could be cut even more as a part of budget reductions.

If the current Administration remains in office, many readers feel reelection* will be interpreted as an electoral stamp of approval for the leaders doing a ‘good enough’ job, and no change is necessary.

(Note. Since our own editors were arguing over this, we offer this side note free of charge. “reelection” is the choice of the Chicago Manual of style, and “re-election” is the choice of the AP style guide. And there you have it.)

The President is not the answer, but he’s certainly part of the equation. The President can’t get us moving at a more steady pace toward exascale without the support of Congress. But the President has to believe in the “need” and has to encourage legislation and funding that will enable the necessary progress. So, in my opinion, it doesn’t matter who gets elected, the problem exists and must be addressed, and there is no evidence that either party recognizes the underlying importance of what a national exascale initiative could do for this country.

IDC research vice president, Steve Conway, agrees that unless the community does a better job of educating members of Congress on why we need an exascale initiative, the presidential election won’t have much of an impact. “In the vector era, $30 million bought the biggest supercomputer available. Today’s largest supercomputers cost $400-500 million and an early exascale computer might cost $1 billion or more. I expect governments will increasingly demand to know more about the returns they can expect from these major investments, especially if global economic conditions remain challenging. National and regional security threats in this decade of exascale development will be hard pressed to justify the required funding levels by themselves. The HPC community will need to describe potential tangible returns to science, industry, society and the economy.

Will the U.S. presidential election have an impact on HPC and exascale?

62% of our readers believe the answer to this question is Yes.

While a majority of our readers believe the presidential election will have an impact on HPC and exascale, opinions vary widely on whether that impact will be positive or negative. What the U.S. does to drive HPC and exascale technology innovation will have an indirect influence on the heavily funded exascale research programs in Europe, Russia, India, China, and Japan.

  • While every country would like to have bragging rights to owning the first exascale system, we all recognize that getting beyond a benchmark system and into the world of real applications will take a monumental effort with every country sharing information. It’s expected that the U.S. will drive much of the research that will lead to new interconnects, new power options, and new approaches to HPC silicon. We have many American partners in our research efforts. If their side of the research slows down or goes away, global progress will suffer.
  • The U.S. Presidential election could have a profound impact on HPC and exascale research. If the next administration steps up the game and puts some serious funding behind high performance computing, we could see significant advances in many computational areas. American industry could realize a tremendous spike in terms of competitiveness driven by new technology.
  • “With a loud enough drumbeat from the community, perhaps the Obama administration can get some things moving. To date, they have struggled to create forward momentum for Science and Engineering innovation. So the risk is more of the same – a continued lack of progress. Under a Romney administration, our fear is that Science and Engineering research investments could be cut in an attempt to balance the budget – thus setting this nation even farther behind.”

Could this be the end of Science and Engineering leadership for the U.S.?
The Exascale Report certainly does not preach doom and gloom. But we do want to serve our mission of responsible journalism and representing the voice of the emerging exascale community. The fact is, we’ve been having this discussion and talking about the very serious situation being created by not investing adequately in scientific and technical computing research, like exascale, for more than eight years. And while some progress is made with our two and three year funded research programs, the overall relationship among stakeholders in government, industry and academia is not thriving. We are a nation at risk.

Here are the words of David Baltimore, (President of Caltech from 1997 to 2006), from an article that appeared in the Los Angeles Times in 2004. . At the time, Baltimore discussed what he felt were some alarming trends impacting U.S. technology leadership.

We no longer have a lock on technology. Europe is increasingly competitive, and Asia has the potential to blow us out of the water.

Line break inserted here to give you time to think about that comment. That was eight years ago.

“In the last 20 years, many of the students in American universities who majored in the sciences and engineering came from Asia. Today, significant numbers are staying in Asia because the schooling there is so improved, and because we have made it harder to study here. And Asian scientists who have been successful here are returning home. None of this is lost on the governments of, say, India and China, which are putting huge sums into modernizing their science infrastructure and universities.”

Baltimore’s point is that the U.S. was not keeping pace and was going down a path that could lead to the demise of the nation’s technological leadership. Baltimore closed his 2004 commentary with the following words that still ring true today: “So the cascade could begin: If America becomes a less affluent society, we will see a diminution in support for the research that is critical to our future. There are already clouds on the horizon: because of the deficit, federal budgets will get tighter and science funding is likely to suffer. The economic recovery is generating too few jobs. Silicon Valley still has lots of vacant space. The venture capital industry is scared and conservative.”

So, where are we today, eight years later? Well, it depends on when you read this article. The claim to having the world’s fastest supercomputer is a game of leap frog. As of this writing, the U.S. has reclaimed the No. 1 position, for the first time in three years. But the competition is fierce among the U.S. Japan, and China as the race towards exascale drives more and more interest in holding on to these “world’s fastest” bragging rights.

And we offer this comment from one of our readers in a .gov organization. “While these other nations are stepping up their investments in science and engineering research, I fear the U.S. is about to drop out of the race.”

If there was ever a time to state the need for U.S. leadership in HPC and future exascale research, this is the time. Be sure to read our featured anonymous response, “The U.S. presidential election has left HPC research an orphan.

So, what’s the bottom line? Does the presidential election really matter to the future of HPC and exascale?

I like to think it doesn’t matter. I like to think that either candidate will see the importance of HPC and exascale to driving our economy forward.

All I know is we can’t take another four years of operating without a solid budget commitment. I think most of us are willing to risk losing it all in hope of seeing some change. So yes, it matters very much.

Clearly Science, technology, and engineering will be impacted. We can only hope that the next four years gives us an Administration that understands what is at risk and also sees the importance of science and technology research to rebuilding the economy.

And here’s an interesting perspective.

If we had billions of dollars to throw at exascale today, we wouldn’t know what to do with it – where to spend the money. What’s lacking? We are clearly lacking a long-term plan, but maybe what’s lacking even more is committed leadership in the right places. This discussion should not be about bragging rights or timeframes. It should be a discussion of how do we build a foundation for generations of science and technology leadership. When you paint that picture, you can’t finish the canvas without a long-term HPC and eventual exascale strategy – and commitment.

What do you think? We’d love to hear from you.

Download a PDF version of this article: “The U.S. Presidential Election – What’s at stake for HPC and Exascale?”

For related stories, visit The Exascale Report Archives.

Comments

  1. ThomasThurston says
  2. Mike Bernhardt says
  3. MikeB-guest says
  4. MikeB-guest says
  5. MikeB-guest says