Introduction

The unavailability of comprehensive, user-oriented programming environments is one of the foremost obstacles to the routine application of parallel, high performance computing technology. Most critical is the need for advances in integrated parallel debugging, performance evaluation, and program visualization tools. New tools must be developed to address the following parallel programming productivity requirements:

(R1)

A user (program-level) view. Past tool development has been dominated by efforts directed at the execution level (e.g., efficient implementation of monitoring). Consequently, tool users are given little support for ``translating'' program-level semantics to and from low-level execution measurements and runtime data.

(R2)

Support for high-level, parallel programming languages. The development of advanced parallel languages (e.g., HPF [4] and pC++ [8]) further separates the user from execution-time reality because of complex transformations and optimizations that take place between the layers of language abstraction, runtime paradigm, and execution environment.

(R3)

Integration with compilers and runtime systems. The majority of debugging and performance analysis tools have been developed independent of parallel languages and runtime systems, resulting in poor reuse of base-level technology, incompatibilities in tool functionality, and interface inconsistencies in the user environment.

(R4)

Portability, extensibility, and retargetability. Users of portable languages need a consistent program development and analysis environment across multiple execution platforms. The tools should be extensible, so their functionality can be increased to accommodate new language or runtime system features. Support for retargetability allows the tool design to be easily reused for different parallel languages and system environments.

(R5)

Usability. Implementing a high-level, portable, integrated tool does not automatically result in an easy-to-use tool. In the past, less emphasis was put on well-designed interfaces which led to very powerful but poorly used program analysis tools.

We have designed and developed a parallel program analysis environment, (TAU, Tuning and Analysis Utilities), for a parallel, object-oriented language system, pC++. In this paper, we describe the design and show how programming productivity requirements for pC++ are addressed in its implementation. Our goal is not to propose itself as a general purpose solution to parallel program analysis. Rather, our goal is to demonstrate the potential benefits of a new development strategy for program analysis tools, one that promotes meeting specific analysis requirements over providing general purpose functionality.

The pC++ language system and the Sage++ restructuring toolkit that forms the basis of the pC++ compiler are briefly described in §2. From this description, we develop a general model of pC++ program observability and analysis. The environment for pC++, presented in §3 to §5 is based on this model and reflects the critical program analysis requirements of the pC++ language. We outline the capabilities of the tools, showing specific instances of their application and explaining how they are implemented in the pC++ programming environment.