An overview of Compositional C++ (CC++)
Guowei Huang and Aixiang Yao

• What is CC++ ? CC++ is a strict superset of the C++ language in that way any valid C or C++ program that does not use a CC++ keyword is also a valid CC++ program. It was designed to alleviate the frustrations of parallel programming by extending the C++ by the following six basic abstractions: thread, sync variable, atomic function, processor object, global pointer, transfer function.

• What does CC++ compiler do? And how to install a CC++ compiler? CC++ compiler is a translator, it translates CC++ code into C++ code with calls to the threads library. Version 4.0 (beta release) of CC++ compiler is available, but a successful installation requires a restricted combination of host operating system, C++ compiler, and threads packages.

• How does CC++ work ? Like any other parallel programming language, CC++ provides the mechanism by which parallel threads of control are created.

• Six Basic abstractions implemented by the CC++ extensions

  • Processor object is a collection of data and computation that defines a single address space. (declared by keyword global)

  • global pointer is a mechanism for linking together processor objects.

  • Thread is a mechanism for specifying concurrent execution. Threads are created independently from processor objects, and more than one thread can execute in a processor object. (created by using keywords par, parfor and spawn)

  • sync variable is used to synchronize thread executions. (identified by keyword sync)

  • Atomic function is a mechanism used to control the interleaving of threads executing in the same processor object. (identified by keyword atomic)

  • Transfer function allow arbitrary data structures to be transferred between processor objects as arguments to remote procedure calls. (identified by the predefined type CCVoid)

• Key words of CC++ :

  • par blocks enclose statements that are executed in parallel. The most basic mechanism for creating parallel threads of control in CC++ is the parallel block. A parallel block looks just like a compound statement in C or C++ with the keyword par.

  • parfor denotes a loop whose iterations are executed in parallel.

  • spawn statement creates a new thread of control executing in parallel with the spawning thread.

  • sync data items are used for synchronization. Synchronization is achieved by making a consumer requiring data (attempt to read data) from a channel block until a producer makes data avaiable.

  • atomic functions control the level of interleaving of actions composed in parallel. To achieve mutual exclusion, an object can be declared atomic. This ensures that even if multiple producers attempt to access (to write, or to append) the same object concurrently, the actual operations will occur in some sequential order.

  • global is used to declare a processor object or to declare a global pointer to objects in processor object. To access a data structure or to perform computation (using a remote procedure call, or RPC) in another processor object, a global pointer must be used.

  • proc_t can be used to construct a placement structure with a specified processor name. proc_t is an implementation-defined type that specifies where to place a processor object and where to find its definition.

  • CCVoid is used to declare a transfer function. data transfer functions describe how data is transfered between address spaces (processor objects).

• Summary
  CC++ provides basic mechanisms that can be used to implement a variety of different parallel program structures. 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. Also all of the object-oriented features of C++ are preserved in CC++ . These features (especially generic classes and inheritance mechanisms) promote the the reuse of code, strucure.

• References
  • Designing and Building Parallel Programs, by Ian Foster, http://www.mcs.anl.gov/dbpp/.

  • A tutorial for CC++, by Paul A.G. Sivilotti, and Peter A. Calin, http://cs.caltech.edu/~paolo/ccpp/tutorial/tutorial.html.