The Global Unified Parallel File System (GUPFS) project is a multiple-phase, five-year project at the National Energy Research Scientific Computing (NERSC) Center to provide a scalable, high performance, high bandwidth, shared file system for all the NERSC production computing and support systems. The primary purpose of the GUPFS project is to make it easier to conduct advanced scientific research using the NERSC systems. This is to be accomplished through the use of a shared file system providing a unified file namespace, operating on consolidated shared storage that is directly accessed by all the NERSC production computing and support systems. During its first year, FY 2002, the GUPFS project focused on identifying, testing, and evaluating existing and emerging shared/cluster file system, SAN fabric, and storage technologies; identifying NERSC user input/output (I/O) requirements, methods, and mechanisms; and developing appropriate benchmarking methodologies and benchmark codes for a parallel environment. This report presents the activities and progress of the GUPFS project during its first year, the results of the evaluations conducted, and plans for near-term and longer-term investigations.

Adaptive mesh refinement (AMR) is a powerful technique that reduces the resources necessary to solve otherwise in-tractable problems in computational science. The AMR strategy solves the problem on a relatively coarse grid, and dynamically refines it in regions requiring higher resolution. However, AMR codes tend to be far more complicated than their uniform grid counterparts due to the software infrastructure necessary to dynamically manage the hierarchical grid framework. Despite this complexity, it is generally believed that future multi-scale applications will increasingly rely on adaptive methods to study problems at unprecedented scale and resolution. Recently, a new generation of parallel-vector architectures have become available that promise to achieve extremely high sustained performance for a wide range of applications, and are the foundation of many leadership-class computing systems worldwide. It is therefore imperative to understand the tradeoffs between conventional scalar and parallel-vector platforms for solving AMR-based calculations. In this paper, we examine the HyperCLaw AMR framework to compare and contrast performance on the Cray X1E, IBM Power3 and Power5, and SGI Altix. To the best of our knowledge, this is the first work that investigates and characterizes the performance of an AMR calculation on modern parallel-vector systems.