High performance computing has become the third pillar of
science and technology, providing the superior computational
capability required for dramatic advances in fields such as
DNA analysis, drug design and astrophysics.  However, during
the past decade, progress has been impeded by a growing lack
of adequate language and tool support.  In today's
dominating programming paradigm, users are forced to adopt a
low-level programming style similar to assembly language if
they want to fully exploit the capabilities of parallel
machines.  This leads to high-cost software production and
error-prone programs that are difficult to write, reuse, and
maintain.  Emerging peta-scale architectures with hundreds
of thousands of processors, and applications of growing size
and complexity will further aggravate this problem.

This talk will discuss the state-of-the-art in programming
paradigms for high performance computing and identify the
challenges posed by future architectures and their
applications.  We will discuss the requirements for
high-productivity programming languages that represent a
viable compromise between the dual goals of human
productivity and target code efficiency, and outline the
related research challenges for effective programming
environments supporting high-level language abstractions.
Specifically, we will provide an overview of the new
programming language Chapel developed in the
High-Productivity-Computing-Systems (HPCS) project Cascade.
Chapel is a modern object-oriented language providing
support for generic programming, collections and iterators,
and explicit parallelism in conjunction with user-defined
data distributions.

The final part of the presentation will focus on the
definition, use, and implementation of data distributions in
Chapel.  The main challenge is to expose the user to enough
level of detail regarding parallel execution to grant
effective communication of problem-specific knowledge, while
concealing the unproductive details related to low-level
parallel programming such as communication, synchronization,
and the explicit distinction between local and remote memory
accesses.  Rather than offering a fixed set of built-in
distributions, Chapel provides a distribution class
interface which allows the explicit specification of the
mapping of elements in a collection to units of uniform
memory access, the control of the arrangement of elements
within such units, the definition of sequential and parallel
iteration over collections, and the specification of
allocation policies.  The result is a concise
high-productivity programming model that separates
algorithms from data representation and enables reuse of
distributions, allocation policies, and specialized data
structures.