COMP 422 Parallel Computing Resources

Cilk

Cilk is an algorithmic multithreaded language. Cilk enables programmers to concentrate on structuring programs to expose parallelism and exploit locality. The responsibility of scheduling the computation to run efficiently is left to Cilk's runtime system, which takes care of details such as load balancing and communication. Unlike other multithreaded languages, Cilk is algorithmic in that its runtime system guarantees efficient and predictable performance.

Links

Cilk Home
An Introduction to Cilk
Cilk Language Reference Manual
Cilk Implementation Description
Scheduling Computations by Work Stealing
Detecting Data Races in Fork-Join Programs

MPI

The Message Passing Interface (MPI) is a standard communication library specification for message passing. MPI is the standard implementation technology for portable, scalable, high performance parallel programs.

Links

MPI Home
MPI 1.1 Standard
MPI Tutorials
Manual Pages for Commands and Library Procedures
MPICH MPI Implementation Home
Book: Using MPI - Portable Parallel Programming with the Message Passing Interface, 2nd Edition

Unified Parallel C (UPC)

UPC is an extension of the C programming language designed for high performance computing on large-scale parallel machines. UPC provides a uniform programming model for both shared and distributed memory hardware. The programmer is presented with a single shared, partitioned address space, where variables may be directly read and written by any processor, although each variable is physically associated with a single processor. UPC uses a Single Program Multiple Data (SPMD) model of computation in which the amount of parallelism is fixed at program startup time, typically with a single thread of execution per processor.

Links

UPC Home
UPC Tutorials
Berkeley UPC Implementation Home
Berkeley UPC Documentation
Book: UPC: Distributed Shared Memory Parallel Programming

Co-array Fortran (CAF)

CAF is an extension of the Fortran 90 programming language designed for high performance computing on large-scale parallel machines. CAF provides a uniform programming model for both shared and distributed memory hardware. CAF supports the abstraction of a two-level global address space in which data is either local or remote. Shared variables, known as co-arrays, may be directly read and written by any processor. The syntax for accessing only locally-resident global data is distinct from that for accessing global data that may reside anywhere, which enables programmers to explicitly manage locality for performance.

Links

Co-array Fortran Home
CAF Language Overview
Rice CAF Implementation Home