Ken Kennedy

The New York Times has a nicely-written obituary for computing pioneer Ken Kennedy, penned by John Markoff. Here’s a snippet:

A member of the generation of researchers who were the first to have access to modern supercomputers, Mr. Kennedy spearheaded early work on software programs known as parallelizing compilers, systems that can automatically spread workloads among a large number of processors, vastly speeding calculations.
Early computers were based on a single processor that would perform the steps of a software program sequentially. But in the 1970s and 1980s researchers began to look for ways to increase computing speed by harnessing tens, hundreds and even thousands of processors, in much the fashion that adding lanes to a freeway will allow more traffic to flow.
The challenge that such systems presented was the need to create programming tools that would hide the interdependencies and complexity from the scientists and engineers who wanted to use the machines as problem-solving tools.
“These compilers made it possible for mere mortals to write advanced programs,” said Edward Lazowska, the Bill and Melinda Gates professor of computer science at the University of Washington in Seattle. “Ken was the No. 1 person in parallel compiling.” (Parallel compilers are software programs that translate programmers’ language-oriented instructions into numeric codes that control computer operation.)
The software technology he developed has served as the foundation for successive generations of scientists and engineers who developed advanced simulations, including weather and climate prediction and the model of automobile collisions. Moreover, the fruits of his technology are now rapidly reaching broad consumer audiences both through the latest generations of personal computers and through videogame players, which now come equipped with parallel processors.

Kennedy also played an important role on the first incarnation of the President’s Information Technology Advisory Committee (PITAC), which put together the 1999 Information Technology Research: Investing in our Future report. The strong, well-supported recommendations in that report helped pave the way for a dramatic expansion of the federal government’s support for computer science research. Kennedy was also a co-PI on CRA’s Computing Community Consortium proposal, which was ultimately successful.
I was privileged to have a few interactions with Kennedy over the six years or so I’ve been at CRA and was always impressed with his grasp of policy and his willingness to do more than was necessary in service of the field.
Update: Chuck Koelbel from Rice passed along these additional details:

A memorial service for Dr. Kennedy will take place at First Presbyterian Church, 5300 Main Street, Houston, on Thursday, February 15 at 3pm. In lieu of flowers, the family suggest gifts be made to Rice University, Ken Kennedy Memorial Fund. Checks may be mailed to Rice University MS-81, P.O. Box 1892, Houston, TX, 77251-1892. To contribute online, visit giving.rice.edu, click “Make a Gift Online”, choose “Designation-Other, and type “Ken Kennedy Memorial Fund” in the Special Instruction box.

 

As we noted yesterday, the National Science Foundation does quite well in the President’s FY 2008 Budget Request, slated to grow 6.8 percent over FY 2007 (or nearly $409 million) to $6.4 billion. That growth rate would continue NSF on the 10-year “doubling” trajectory originally set by the Administration as part of last year’s American Competitiveness Initiative. The news for the computing community that is so heavily reliant on NSF is equally good — both the Computing and Information Science and Engineering directorate and the Office of Cyberinfrastructure would see big gains in the President’s plan. Here are the details (brace for charts):

First, the macro level view of the agency:


National Science Foundation
(in millions of dollars)
FY06
Actual
FY07
Budget Request
FY08
Request
$ Change vs
FY07 Request
% Change vs
FY07 Request (%)
Research and Related Activities $4,449.25 $4,765.95 $5,131.69 $365.74 7.7%
Education and Human Resources $700.26 $716.22 $750.60 $34.48 4.8%
MREFC $233.81 $240.45 $244.74 $4.29 1.8%
Agency Operations and Award Management $247.06 $281.82 $285.59 $3.77 1.3%
National Science Board $3.94 $3.91 $4.03 $0.12 3.1%
Office of the Inspector General $11.47 $11.86 $12.35 $0.49 4.1%
Total NSF $5,645.79 $6,020.21 $6,429.00 $408.79 6.8%

As you can see, the great bulk of the Administration’s planned increase is aimed at the Research and Related Activities (R&RA) account, home of NSF’s research directorates. The agency’s education efforts — in the Education and Human Resources directorate — would also see an increase, though not nearly as robust as R&RA. Of particular interest to those of us in the computing community, NSF is using some of the increase provided by ACI on a new NSF-wide initiative called “Cyber-enabled Discovery and Innovation” (CDI) that aims to “broaden the Nation’s capability for innovation by developing a new generation of computationally based discovery concepts and tools to deal with complex, data-rich and interacting systems.” The $52 million initiative will be led by NSF’s CISE directorate (who will control $20 million of the funding), with participation from Engineering, Mathematics and Physical Science, Social, Behavioral and Economic science, Cyberinfrastructure, International Science, and EHR. The agency appears to have big plans for the initative, projecting out-year funding growing to $250 million in FY 2012 (with CISE hopefully maintaining a proportional share).

Drilling down a little further, here’s the Directorate by Directorate breakdown within R&RA:


NSF Research and Related Activities
Directorate Budgets
(in millions of dollars)
FY06
Actual
FY07
Budget Request
FY08
Request
$ Change vs
FY07 Request
% Change vs
FY07 Request (%)
Biological Sciences $580.90 $607.85 $633.00 $25.15 4.1%
Computer and Information Science and Engineering $496.35 $526.69 $574.00 $47.31 9.0%
Engineering $585.46 $628.55 $683.30 $54.75 8.7%
Geosciences $703.95 $744.85 $792.00 $47.15 6.3%
Mathematical and Physical Sciences $1,086.61 $1,150.30 $1,253.00 $102.70 8.9%
Social, Behavioral and Economic Sciences $201.23 $213.76 $222.00 $8.24 3.9%
Office of Cyberinfrastructure $127.14 $182.42 $200.00 $17.58 9.6%
Office of International Science and Engineering $42.61 $40.61 $45.00 $4.39 10.8%
Office of Polar Programs $390.54 $438.10 $464.90 $26.80 6.1%
Integrative Activities1 $233.30 $231.37 $263.00 $31.63 13.7%
U.S. Arctic Research Commission $1.17 $1.45 $1.49 $0.04 2.8%
Total, Research and Related Activities $4,449.25 $4,765.95 $5,131.69 $365.74 7.7%

The increases weren’t evenly distributed throughout the directorates. Of the research directorates, CISE would see the largest percentage increase. In fact, the 9 percent requested growth rate is the largest for the directorate in seven years. Here’s how CISE plans to spend the funding:


NSF CISE Directorate Funding
(in millions of dollars)
FY06
Actual
FY07
Budget Request
FY08
Request
$ Change vs
FY07 Request
% Change vs
FY07 Request (%)
Computing and Communication Foundation $105.30 $122.82 $149.15 $26.33 21.4%
Computer and Network Systems $141.07 $162.98 $191.98 $29.00 17.8%
Information and Intelligent Systems $103.78 $119.30 $154.63 $35.33 29.6%
Information Technology Research $146.20 $121.59 $78.24 -$43.35 -35.7%
Total, CISE $496.35 $526.69 $574.00 $47.31 9.0%

Deborah Crawford, the acting AD for CISE, highlighted a number of new programs the new funding — and funding freed up as the ITR program comes to an end — would allow the directorate to pursue. First is an emphasis on “Discovery Research for Innovation,” which includes these new efforts:

  • High Risk, High Return Research ($50 million) — “Seeking Big Ideas in support of Grand Vision.” Programs in the area will focus on fundamental questions in computing, larger projects, and try to exploit the potential of emerging technologies.

  • Cyber-enabled Discovery and Innovation ($20 million) — As detailed above, CISE will lead this NSF-wide effort, focusing on promoting computational thinking for problem solving.

In addition, CISE will continue its support for planning of the Global Environment for Networking Innovations with $20 million in funding for pre-construction planning. GENI was also included in the Foundation’s 2007 Facility plan as the first “Horizon” project — a step away from “Readiness Stage,” which would allow for extensive pre-construction planning. GENI is one of 10 projects listed as “Horizon” projects. (There’s just one project in the Readiness Stage in FY 2008 — the Advanced Technology Solar Telescope — and just one that is listed as a possible new start in FY 2008 (the Advanced Laser Interferometer Gravitational Wave Observatory)).

The budget also includes statistics on the number of awards and the funding rate estimates for the directorate in FY 06, 07 and 08. The directorate estimates it will fund fewer research grants in FY 07 than it did in FY 06 (950 in FY07 vs 1,003 in FY06), with a corresponding drop in funding rate (18 percent in FY07 vs 22 percent in FY06). For FY08, the directorate expects the number of research grants to grow to 1,000 and the funding rate to rise a bit to 20 percent.

Despite those figures, this is, overall, a very promising start for computing at NSF in FY 2008 — which, given NSF’s role in funding 87 percent of academic basic research in computing, makes it a good start for the field.

 

As stated in a previous post about the FY08 Budget Request, Department of Energy’s Office of Science did well with a $296 million, or 7 percent, increase over the FY07 request. The Advanced Scientific Computing Research (ASCR) request is $340.2 million, an increase of $21.5 million or 6.8 percent.
The ASCR has three overarching programs: Research in applied mathematics and computer science with a request of $82.8 million up from $69.6 million in the FY07 request; Scientific Discovery through Advanced Computing (SciDAC) with a request of $56.3 million up from $56.1 million in the FY07 request; and High-performance computing and network facilities and testbeds with a request of $201.1 million up from $193 million in the FY07 request.
Overall these are good numbers for computing, and science, and certainly help to make up for some of the recent lean years. As always, we’ll keep you posted as we learn more and as the budget process goes forward on Capitol Hill.