Future Technologies Group
Berkeley Lab Computing Sciences

Group Members

Current Projects

Past Projects

Related Sites

    

SciDAC Performance Engineering Research Institute (PERI)

PERI is an Integrated Software Infrastructure Center (ISIC) for enabling improved of high-end computer performance. In particular, the Center is developing a science for understanding performance of scientific applications on high-end computer systems, and is also developing engineering strategies for improving performance on these systems. The project integrates several active efforts in the high performance computing (HPC) community and is forging alliances with application scientists working on DOE Office of Science (SC) missions to ensure that the resulting techniques and tools are truly useful to end users.

The project focuses on how one can best execute a specific application on a given platform. In particular, we seek to
  1. Understand the key factors in applications that affect performance.
  2. Understand the key factors in computer systems that affect performance.
  3. Develop models that accurately predict performance of applications on systems.
  4. Develop an enabling infrastructure of tools for performance monitoring, modeling and optimization.
  5. Validate these ideas and infrastructure via close collaboration with DOE SC and other application owners.
  6. Transfer the technology to end-users.

This activity focuses on high performance computing systems (i.e., large distributed memory parallel systems; large shared memory systems; and large cluster systems). We also focus on representative scientific applications and problems of interest to the SC, initially those areas emphasized in the SciDAC application projects. Our objective is to characterize the realistic performance bounds of HPC applications and systems, understand why these limits exist, determine what can be done to more closely approach these limits, and predict what can be expected on future applications and systems. To that end, we will develop benchmarks, models, analytical techniques, and tools.