Previous: Introduction Up: Introduction Next: This Tutorial
The CC++ Programming Language
Is parallel programming difficult? Many programmers complain
that architecture dependencies, confusing notations,
difficult correctness verification, and burdensome overhead make parallel
programming seem tedious and arduous. These barriers
cast doubt on the practicality of parallel programming, despite
its many potential benefits.
Compositional C++ (CC++) was designed
to alleviate the frustrations of parallel programming
by adding a few simple
extensions to the sequential language C++. It is a strict
superset of the C++ language so any valid C or C++
program is a valid CC++ program. Conversly,
many classes of parallel CC++ programs can be simply
rewritten as equivalent sequential programs. For these
classes of programs, the developement path can be very
similar to that of the equivalent sequential program.
This compatibility with the C and C++ languages
facilitates the transition to the task of parallel
programming for users knowledgeable in those languages.
CC++ extends C++ with the following eight constructs:
- par
- blocks enclose statements that are executed in parallel.
- parfor
- denotes a loop whose iterations are executed in parallel.
- spawn
- statements create a new thread of
control executing in parallel with the spawning thread.
- sync
- data items are used for synchronization.
- atomic
- functions control the level of interleaving of actions
composed in parallel.
- processor objects
- define the distribution of a computation.
- global
- pointers link distributed parts of the computation.
- data transfer
- functions describes how information is transmitted
between address spaces.
Despite the simplicity and the small number of extensions, the conjunction
of these constructs, when combined with C++, results in an extremely
rich and powerful parallel programming notation.
The richness of this language is reflected in its suitability
to a wide spectrum of applications. In fact, CC++ integrates
many seemingly disparate fields:
- Sequential and parallel programming
- Parallel blocks
are notationally similar to sequential blocks.
- Shared and distributed memory models
-
CC++ can be used on shared or distributed memory architectures
as well as across networks which may be heterogenous.
- Granularity
-
CC++ can be used in computations involving a variety of
granularities, ranging from fine-grain parallelism with frequent
synchronization between small threads to
coarse-grain parallelism with sparse synchronization between large,
distributed processes.
- Task and data parallelism
- Task and data parallel applications
can be expressed in CC++, as well as programs that combine the two.
- Synchronization techniques
-
The synchronization mechanisms
provided by CC++ are powerful enough to express any of the traditional
imperative synchronization and communication paradigms.
All of the object-oriented features of C++ are preserved
in CC++. These features (especially generic classes and
inheritance mechanisms) promote the reuse of code, structure,
and verification arguments.
Thus, CC++ provides an excellent framework for the developement
of libraries and for the use of software templates in program construction.
This reuse is especially important and useful in the
context of parallel programming.