Research Of Parallel Algorithm Based On OpenMP And MPI

Now the demand for computing power and large-scale scientific and engineering computing for parallel computing is on the growing, the demand promotes the development of high-performance computers. The current single-processor can't meet the needs of large-scale computing problems, parallel computing is the supporting technology that the only way to solve this kind of problem. The current parallel programming environment are message passing, shared memory and data parallel. Some of the inherent parallelism in the practical application is to produce a source of parallel machines, so the application of parallelism in the application of parallel computers is the key problem, parallel algorithm is the basis for parallel program development, and therefore parallel algorithm is important.OpenMP application programming interface is a parallel programming standard on a shared memory multi-processors, it is a widespread interest because it is a programming model easy to use. OpenMP is not a language, it is a instruction set extended with some new language, such as Fortran and C/C++. OpenMP application programming interface standard defines a set of extended C/C++and Fortran instructions. Unlike the previous API, such as POSIX threads and MPI, OpenMP is conducive to incrementally parallelize serial programs, programmer can increase the parallel instruction in the cycle part of the program. OpenMP uses the Fork-Join parallel execution model. ACA is a algorithm that inspired by the foraging behavior of ant colony, the algorithm reflects the biological inspiration and represents the many problems in computer science. This paper presents a ant colony optimization based on OpenMP.MPI is a message passing standard, it simplifies the development of applications and libraries, this standard defines the core semantics and syntax. For the majority of users, it is very useful to messaging program using Fortran77 and C. It provides communication for a set of concurrent processes in the program, the reason why we attract the message passing model because of its wide range of portability and communication. MPI is a library, therefore, the use of MPI and the language must be combined. MPI supports FORTRAN and C/C++. Eight queens problem is how to place the eight queens in the 8×8 grid of chess, they do not attack each other and it is a typical representative of backtracking algorithms. This paper proposes a mechanism based on MPI to solve the eight queens problem.Part 1 of this paper is introduction, simply describes the research background and significance of the topic, reviews the status of related research at home and abroad, gives the main research content and organizational structure of the paper. Part 2 focuses on the history of parallel computers, architecture and the classification and development of parallel algorithms. Part 3 introduces the OpenMP parallel programming model and ant colony algorithm, and proposes a parallel ant colony algorithm based on OpenMP, then do the experiment and analysis the results. Part 4 describes MPI and Eight Queens Algorithm, proposes a parallel MPI-based Eight Queens Algorithm, then do the experiment and analysis the results. Part 5 is conclusion and outlook, presents the problems during the study and the future research.