A new report from the Council on Competitiveness (Council) explores how US government investment in HPC benefits America’s industrial and economic competitiveness. The Council on Competitiveness, with funding from the US Department of Energy, engaged Intersect360 Research to interview HPC-using US organizations across a wide range of industries.
Over a six month period, Intersect360 conducted 14 in-depth interviews and collected 101 comprehensive online surveys. The findings were published and released today in a 75-page report, entitled “Solve. The Exascale Effect: the Benefits of Supercomputing Investment for U.S. Industry.”
The Council, which is known for popularizing the sentiment “to outcompute is to outcompete,” views HPC as instrumental to the progress of the United States in science, engineering and business. The report reiterates the value that HPC brings to the collective table and identifies key areas of targeted investment that would provide the greatest benefits.
Key findings:
Two-thirds of US companies that use HPC say that “increasing performance of computational models is a matter of competitive survival.”
More than one-third of US industry representatives surveyed claim their most demanding high performance computing applications could utilize 1,000-fold increases in computing capability over the next five years.
One of the comments speaks directly to the difficulty humans have conceptualizing big numbers. “I tend to think in terms of 100x,” wrote one respondent. “However, if I view today’s compute power to what we used for modeling and simulation just a couple of decades ago, I have to believe that 1,000x will be needed. I just can’t get my head around this yet.”
Software scalability was cited as the most significant limiting factor to achieve the next 10-fold improvement in performance, and it was given as the second most significant limiting factor to reach a 1,000-fold improvement. Cost of hardware was the number one impediment to 1,000-fold improvement, and making the case to management also ranked as a significant limitation.
There is strong agreement that government investment in leading-edge HPC benefits US companies and industries, but often in non-quantifiable ways. Respondents also note that the links between government and industry need to be strengthened.
Several of the comments and responses from long-form interviews centered on the idea of trickle-down or the ripple effect, where once elite-class technologies find their way into commodity systems. The biggest promise of exascale computing, viewed this way, is affordable petascale computing.
In the words of one commenter, “The fastest benefit that exascale is going to have is that the petaflop will become extremely inexpensive, so that every midsize company will be able to have a petaflop supercomputer. But industry doesn’t adopt the leading edge of HPC; they’re usually several ticks behind.”
One of the strongest refrains in the report was the pressing need for application software that will exploit hardware scalability. Recognizing that different users have different needs, the report suggests several program models for government consideration. These are targeted programs for different segments of the commercial space, broken down into in-house software; ISVs; open source; and entry-level HPC. With regard to entry-level adopters, the report contends “the ability to integrate HPC into the workflow is a bigger challenge than scalability.”