On July 12, 2017, the U.S. federal budget for its Exascale Computing Initiative (ECI) took its next step forward. On that day, the full Appropriations Committee of the House of Representatives voted to accept the recommendations of its Energy and Water Appropriations Subcommittee for Fiscal Year 2018 (FY18) budget for a variety of agencies that includes the Department of Energy (DOE). Part of the DOE funding is for the United States effort to develop an exascale computer. Just to be clear, the U.S. government is not defining exascale as 10^18 floating point operations per second (Flops) on the Linpack benchmark. Rather, as defined by the National Strategic Computing Initiative (NSCI) an exascale is a “computing systems at least 50 times faster than the nation’s most powerful supercomputers in use today.”
The news for exascale that was released on July 12 was not terribly dramatic, but helped to expose a slight mystery. But first, the numbers. As reported earlier, for the Department of Energy (DOE) the President’s FY18 budget proposed to spend $508 million on Exascale Computing Initiative (ECI). This number was divided between the DOE Office of Science (SC) and the semi-autonomous National Nuclear Security Administration (NNSA). The request for the NNSA was a total of $183 million with the bulk ($161 million) going to the Advanced Simulation and Computing (ASC) program. The remainder ($22 million) would go to building physical infrastructure for the exascale systems. The House Appropriations Committee provided the full budget for the ASC program with very little comment. This means that the House appropriated NNSA with the full $183 million for exascale.
On the SC side of things, the program funding exascale is the Office of Advanced Scientific Computing Research (ASCR) and the situation is a bit more complicated. The President’s budget proposed that ASCR would get $347 million in support of ECI. The request number was split into two parts. The first part of the request was $197 million for Exascale Computing Project (ECP). The House Appropriations Committee gave $170 million to ECP, which is down $27 million. The other part of the SC funding request ($150 million) was for the two leadership computing facilities (one at Oak Ridge and the other at Argonne national laboratories) to prepare them for the exascale systems. The report language does not explicitly address the $150 million for facilities, but based on the overall ASCR numbers, it looks like they got nearly the full requested amount.
But, here is where the mystery pops up. Part of the text of the House markup report says, “The Committee is concerned that the deployment plan for an exascale machine has undergone major changes without an appropriately defined cost and performance baseline.” The report goes on to require the DOE to provide both Houses of Congress an updated baseline within 90 days of enactment of the Appropriations Act. The report does not provide any further information about the “major changes” it is referring to.
However, in the original President’s SC budget request, on page 25 of the introduction there are two short sentences that say, “The ALCF [Argonne Leadership Computing Facility] upgrade project will shift toward an advanced architecture, particularly well-suited for machine learning applications capable of more than an exaflop performance when delivered. This will impact site preparations and requires significant new non-recurring engineering efforts with the vendor to develop features that meet ECI requirements and that are architecturally diverse from the OLCF [Oak-ridge Leadership Computing Facility] exascale system.” It appears that these sentences in the request are part of the House’s Appropriator’s concerns about “major changes.” Another interesting note is that the announced Argonne 180 petaflops computer code-named Aurora is not mentioned in the request. At this point, any speculation about the meaning these words is useless and any conclusions probably wrong.
The good news is that the House Appropriations Committee supported the request for funding for the exascale program. Once again, this is rather amazing in light of the other significant budget cuts that are being made. These include the House agreeing with the President to zero out the funding for Advanced Research Project Agency – Energy (ARPA-E) and nearly cutting the funding for the Energy Efficiency and Renewable Energy (EERE) program by 50 percent. There is still a long way for the FY18 budget to become enacted in law, but the developments of last week help to shed light on the trajectory. Things are looking good for the U.S. exascale programs and that is great news for our country’s national and economic security.
Coming next – we hear from the Senate Appropriators!
About the Author
Alex Larzelere is a senior fellow at the U.S. Council on Competitiveness, the president of Larzelere & Associates Consulting and HPCwire’s policy editor. He is currently a technologist, speaker and author on a number of disruptive technologies that include: advanced modeling and simulation; high performance computing; artificial intelligence; the Internet of Things; and additive manufacturing. Alex’s career has included time in federal service (working closely with DOE national labs), private industry, and as founder of a small business. Throughout that time, he led programs that implemented the use of cutting edge advanced computing technologies to enable high resolution, multi-physics simulations of complex physical systems. Alex is the author of “Delivering Insight: The History of the Accelerated Strategic Computing Initiative (ASCI).”