Organizers

• Barton Miller (University of Wisconsin - Madison) "bart" AT "cs.wisc.edu"
• John Mellor-Crummey (Rice University) "johnmc" AT "cs.rice.edu"

Abstract

Multipetaflop systems for open science are being installed at Argonne National Laboratory (ANL) and Oak Ridge National Laboratory (ORNL). Compared to systems of the past, these machines represent a dramatic increase in scale and complexity. Making effective use of systems at this scale will require performance tools that help application developers by pinpointing and explaining application performance bottlenecks. For ANL's Mira system, the principal challenge will be exploiting threading effectively on each node's 16-core, 4-way multithreaded processors. For ORNL's Titan system, the principal challenge will be exploiting hybrid parallelism across a 16-core processor and a GPU.

Providing tools that make it possible to address these problems is requiring increasingly sophisticated methods for instrumentation, measurement, analysis and modeling of application performance. Meeting these challenges at scale will require tools with unprecedented capabilities.

The goal of this workshop is to bring together tools researchers to discuss challenges of performance analysis and debugging on emerging petascale systems, review ongoing research, and work together to tackle the challenges of performance analysis on petascale systems together as a community. Attendees at the workshop will be tools developers in the areas of tools infrastructure, debugging, performance instrumentation, measurement, analysis, and visualization. We expect that about a third of the workshop will consist of opportunities to work together in smaller groups to make progress on tackling problems of mutual interest.

Specific aims of the workshop include

• Identifying performance tool capabilities needed to analyze the spectrum of issues that arise with emerging extreme scale systems.
• Discussing emerging capabilities of research and commercial tools in this space.
• Identifying common needs, functionality, and opportunities for sharing infrastructure.
• Discussing design aspects of sharable components.
• Developing standardized interfaces to facilitate sharing.

Final Program