Guidelines for Writing System Software and Tools Requirements
for Parallel and Clustered Computers

A collaborative effort involving both users and developers of system software

These documents present the findings of a national task force convened in 1995, on the recommendation of the Second Pasadena Workshop on System Software and Tools for High-Performance Computing Environments. The group was charged with establishing the basic requirements for a software infrastructure to support parallel applications development. Those requirements must be representative of the high-performance computing (HPC) user community as a whole. It must also be possible to implement the requirements in a robust and consistent fashion across a full range of parallel and clustered computing platforms. In the interests of expediency, the requirements must be implementable without significant new technological development.

A prime motivation for the effort was the belief that a "standard" set of system software and tools is essential to the continued growth of the HPC industry. HPC providers must be able to anticipate requirements in order to deliver a robust working environment early in the cycle of new system releases. If multiple vendors implement the environment quickly, users will derive significant benefits by simply having access to a robust environment that is consistent across multiple HPC platforms. The availability of such an environment will also encourage more development of parallel software by independent software vendors.

The task force effort was endorsed by several national groups, including the Parallel Tools Consortium and the Scientific and Engineering Computing Working Group of the National Coordinating Office for HPCC. Over sixty representatives from major user sites and commercial software vendors participated in the three meetings and extended email discussions.

The results are presented as a series of guidelines for preparing system software and tools requirements for inclusion in Requests for Proposals (RFPs). Specific recommendations have been grouped into two general categories:

These recommendations should not be construed as a dis-incentive for HPC providers to offer additional tools or software meeting special needs. Rather, they establish the minimal level of support that is needed by application developers. Some agencies will find it necessary to impose additional requirements (e.g., additional security features, reliability constraints, the availability of specialized libraries) on the RFP process. Suggestions are included for how this should be accomplished as well.

It is also important to note that user and system requirements evolve continuously. The guidelines laid out in these documents reflect current concerns in the user community, as well as the current level of HPC technology. They are intended to be in effect for just two years. Participants strongly urge that a second task force be established in mid-1997 to review the capabilities. Thorough revision is expected to be necessary, and should take into account not just technological advances, but the experiences of agencies that have applied the guidelines in their procurement procedures.

Associated Documents

A hard-copy version of the first four items has been prepared as a technical report. The bibliographic citation for this is:
Cherri M. Pancake, Bruce Blaylock, and Robert Ferraro, "Guidelines for Writing System Software and Tools Requirements for Parallel and Clustered Computers," Technical Report 95-80-11, Department of Computer Science, Oregon State University, November, 1995.
To obtain a copy, contact

Last revision November 22, 1995

Comments to Cherri Pancake,