Every parallel programming language must address certain issues, either explicitly or
implicitly. There must be a way to create parallel processes, and there must be a way to coordinate
the activities of these processes. Sometimes the processes work on their own data and do not
interact. But when processes exchange results, they must communicate and synchronize with each
other. Communication and synchronization can be accomplished by sharing variables or by message
passing.
There are two methods of synchronization: synchronization for precedence and synchronization for
mutual exclusion. Precedence synchronization guarantees that one event does not begin until another
even has finished. Mutual exclusion synchronization guarantees that only one process at a time
enters a critical section of code where a data structure to be shared is manipulated.
PVM is a software package that permits a heterogeneous collection of Unix
computers hooked together by a network to be used as a single large parallel computer. Thus large
computational problems can be solved more cost effectively by using the aggregate power and memory
of many computers. The software is very portable. The source, which is available free thru netlib, has
been compiled on everything from laptops to CRAYs.
PVM enables users to exploit their existing computer hardware to solve much larger problems at
minimal additional cost. Hundreds of sites around the world are using PVM to solve important scientific,
industrial, and medical problems in addition to PVM's use as an educational tool to teach parallel programming.
With tens of thousands of users, PVM has become the de facto standard for distributed computing
world-wide.
Occam |
|
"The occam programming language is designed to express concurrent algorithms and their implementation
on a network of processing components.
Occam enables an application to be described as a collection of processes, where the processes
execute concurrently, and communicate with each other through channels. Each process in such an
application describes the behaviour of a particular aspect of the implementation, and each channel
describes a connection between two processes. This approach has two important consequences. Firstly,
it gives the program a clearly defined and simple structure. Secondly, it allows the application to
exploit the performance of a system which consists of many parts.
Occam is developed at INMOS Limited, now SGS-THOMSON Microelectronics Limited in the United
Kingdom. The development of the INMOS transputer, a family of devices which place one or more
microcomputers on a single chip, has been closely related to occam, its design and implementation.
The transputer reflects the occam architectural model, and may be considered an occam machine."
(from Occam 2.1 Reference Manual, May 12, 1995)
Though occam was designed for multitransputer system now there are compilers for other platforms.
KROC (Kent Retargetable Occam Compiler)
KROC is a portable OCCAM compiler system that enables OCCAM to run on
non-transputer platforms (Sun SPARC running SunOS 4.1.3U1 or Solaris 2.5; DEC Alpha
running OSF1/3.0). KROC works
by translating code produced by an INMOS OCCAM Toolset compiler into the native
assembler for the target architecture and linking in a small (< 2K bytes) kernel that
provides the process scheduling and message-passing functionality of the transputer
micro-code.
Release of KROC is
available
Sisal
|  |
Mission The objectives of the Sisal Language Project are to develop high-performance functional
compilers and
runtime systems to simplify the process of writing scientific programs on parallel supercomputers
and to help
programmers develop functional scientific applications.
Impact Functional languages such as Sisal provide a low-cost approach to developing parallel
computing applications
that still offer high performance and portability. This is of major significance, as parallel
software costs are projected to
top $1 trillion. Despite the commercial availability of multiprocessor computer systems, the number
of parallel scientific
and commercial applications in production use today remains small. It is estimated that the
worldwide cost of
developing software for sequential machines will reach $450 billion in 1995. Even if only a quarter
of the software is
parallelized, it would cost at least an additional $550 billion. Functional programming can reduce
this increased cost
while still providing high performance and portability.
Functional languages, such as Sisal, promote the construction of correct parallel programs by
isolating the programmer from the complexities of parallel processing. Based on the principles of
mathematics, Sisal exposes implicit parallelism through data independence and guarantees determinate
results.
Click here for more info on Sisal's mathematical
foundations.
