tim Burrell of microsoft’s secure Windows Initiative describing the Phoenix compiler and automated vulnerability finding at the eusec West conference, may 2008, London, u.K.

 

recommendations:

Enable the creation of compiler research centers. There are few large compiler research groups anywhere in the world today. At universities, such groups typically consist of a senior researcher and a few students, and their projects tend to be short term, usually only as long as a Ph.D. project. Meanwhile, the few industrial groups studying advanced compiler technology tend to focus on near-term solutions, even as the compilers, the programs they translate and analyze, and the target execution environments have increased in complexity. The result is that too much of today’s compiler research focuses on narrow problems, ignoring the opportunity to develop revolutionary strategies requiring long-term commitment for their development and evaluation.

In those areas of compiler research seeing diminishing returns today from incremental approaches (such as program analysis and optimization), researchers must attempt radical new solutions that are likely to be lengthy and involved. The compiler research community (university and industry) must work together to develop a few large projects or centers where long-term research projects with the support of a stable staff are carried out. Industry and funding agencies must work together to stimulate and create opportunities for the initiation of these centers. Joint in-dustry/government funding must support the ongoing evolution and maintenance of the software.

Create significant university/industry/ government partnerships for the development of infrastructure and the gathering of benchmarks while funding individual projects at universities and other research

centers. Implementation and experimentation are central to the compiler area. New techniques and tools, as well as new implementations of known approaches, can be meaningfully evaluated only after they are incorporated into industrial-strength compiler infrastructures with state-of-the-art optimizations and code-generation strategies. The absence of widely accepted compiler research platforms has hindered research efforts in compiler technology, design of new programming languages, and development of optimizations for new machine architectures. The development of complete compilers requires an immense effort typically beyond the ability of a single research group. Source code is now available for the GNU compiler and for other compilers developed by industry (such as IBM’s Java Jikes compiler, Intel’s Open Research Compiler, and Microsoft’s Phoenix framework). They may yet evolve into the desired infrastructure, but none currently meets all the needs of the community.

Experimental studies require benchmarks that are representative of the most important applications at the time of the evaluation, meaning the process of gathering representative programs is a permanent process. Numerous efforts have sought to gather benchmarks, but the collections tend to be limited and are often complicated by doubts as to how representative they truly are. A serious difficulty is that many widely used programs are proprietary. In domains where open source applications might represent proprietary software, it would suffice for the purpose of evaluating compilers to make use of open source versions that

 

66 CommunICatIons of the aCm | feBRuaRY 2009 | vol. 52 | No. 2

outperform their proprietary counterparts; representative input data sets must accompany any set of benchmarks. Also desirable is a description of the algorithms and data structures to enable research in new programming languages and extensions that may be better suited for expressing the computation.

Federal agencies and industry must collaborate to establish the necessary funding opportunities and incentives that will move compiler specialists at universities and industry to address the research infrastructure and benchmark challenges. Funding opportunities should also be made available for smaller academic studies and development efforts. The computer science community has achieved notable success in developing novel compiler infrastructures, including: compiler front-end development; program-analysis frameworks for research in compilers; verification tools, security applications, and software-engineering tools; and virtual-machine frameworks for both mainstream and special-purpose languages. Even if the GNU and industry-developed compilers were a useful infrastructure for research, development of new robust, easy-to-use infrastructures by the researchers who need them are critical for future advances. Not only is it important to support such research projects, the research community must recognize their academic merit.

Develop methodologies and repositories that enable the comparison of methods and reproducibility of results. The reproducibility of results is critical for comparing the strategies, machines, languages, and problem domains. Too much of the information available to-

Photogra Ph by ryo hirosa Wa