1. Distributed computing
Distributed computing, this research mainly study how to calculate dispersion system, decentralized system is a set of computer, linked through the computer network and communication after the formation of the system, the need for a large number of calculation of engineering data partition into small pieces, by multiple computers calculate respectively, after the upload operation result, incorporating results unified data conclusion science,
At present common distributed computing projects often use computer idle capacity in the tens of millions of volunteers from all over the world, and the data transmission through the Internet,
Distributed computing, compared with other algorithm has the following advantages:
1, scarce resources can be Shared,
2, can be on multiple computers by distributed computing load balance calculation,
3, can put the program on the computer is most suited to running it, among them, the sharing of rare resources and balance load is one of the core of the computer distributed computing,

Parallel computing
2.Parallel computing or parallel computing is relative to the serial computing, parallel computing is used at the same time a variety of computing resources to solve the problems of the calculation process, and to implement parallel computing, computing resources should include a fitted with multiprocessor parallel processing computer, a computer connected to the network proprietary number, or a combination of use, the main purpose of the parallel computing is rapidly resolve the calculation problem of large and complex,
Parallel computing can be divided into parallel time and space in parallel, time parallel pipeline technology, namely spatial parallel use of multiple processors perform concurrent computing, the current research is mainly parallel problem of the space, designer's point of view, to program and algorithm can be divided into the data parallel and task parallel parallel computing, data parallel to resolve the big tasks into several subtasks, same simple to handle than tasks in parallel,
Space parallel to the production of two kinds of parallel computers, according to Michael Flynn (fehling classification) is divided into single instruction multiple data (SIMD) flows in multiple instruction stream and data stream (MIMD), and the commonly used serial machine, also known as single instruction stream and data stream (SISD), MIMD kind machine and can be divided into five categories: common parallel vector processor (PVP), a symmetric multiprocessor (SMP), massively parallel processor (MPP), workstation fleet (COW), a distributed Shared memory processor (DSM)

3. The difference between the parallel computing and distributed computing
Parallel computing and distributed computing are used in parallel to achieve higher performance, large tasks into smaller tasks, simply put, if processing unit Shared memory, is known as parallel computing, and vice is distributed computing, also some people think that distributed computing is a special case of the parallel computing,
But distributed workunits have independence between each other, a task on the result of the package is not returned or handling errors as a result, the next task package deal with little impact, therefore, a distributed real-time demand is not high, and allowed error in calculation (because each computing tasks to several participants, to compare the results after upload the results to the server, then the large difference of the results of verification,
Distributed to deal with the problem of general model is based on the "search", the so-called "looking for", is the equivalent of exhaustive method! To try to every possible results, generally from 0 ~ N (a number) is one of the tests, until we find the required as a result, in fact, in order to easy to one-time to detect the correct results, we assume that the result is a special form, in this type of search, we may be lucky to find the answer from the start; That may not be enough lucky to finally find the answer, it is very fair,
So to speak, a parallel program there is a lot of contact between the task of parallel processing package, and parallel computing of each task is necessary, no waste of segmentation, each work is to deal with, and the calculation results influence each other, requires that each of the results to be absolutely right, and to try to be on time synchronization, and distributed to don't deal with many tasks, there are a lot of useless data blocks, so distributed computing speed although soon, but the real "efficiency" is the low low again, may have been looking for, but always can't find, may also be found from the start; And the parallel processing is different, its number of workunits are relatively limited, should be possible in a limited time,
Distributed coding generally used c + + (also useful JAVA, but is not the mainstream), basic need not the MPI interface, parallel computing using MPI or OpenMP,

4. Grid computing
Grid computing is a kind of distributed computing, also is a kind of cluster computing is very related technology, if we say that a job is distributed, then, must be involved in the work is not just a computer, but a computer network, obviously this "ants move mountain" on the way will have strong data processing ability, the essence of grid computing is the combination and sharing resources and ensure the system security,
Grid computing through the use of a large number of heterogeneous computer unused resources (week of CPU and disk storage), as embedded in the distributed telecommunication infrastructure in a virtual computer cluster, in order to solve large-scale computational problems provide a model, the focus of the grid computing on the ability to support the management domain calculation, which makes it with the traditional computer cluster, or different from the traditional distributed computing, grid computing aims to solve for any single supercomputer is still difficult to solve bigger problems, and at the same time keep the flexibility to solve the problem of smaller, so, grid computing provides a multi-user environment,

5. Cloud computing
New concept of cloud computing is a new start, it is not just a computing concepts, such as computer and operation service concept, it is a distributed computing, parallel computing and grid computing, or the commercial implementation of the concept of cloud computing is not only including distributed computing, also including distributed storage and distributed cache, distributed storage, including distributed file storage and distributed data storage,