Looking forward to the next generation of languages: Fortress, chapel or x10)

Link: http://blog.csdn.net/phoenixsh

These three languages are DARPA-sponsored HPCS (high productivity vity computer system), respectively undertaken by Sun, Cray, and IBM. HPCS were originally a high performance computer system and a project launched by DARPA to promote the next-generation software development platform. Later, as multi-core became a climate, concurrent programming was recognized as an extreme difficulty, so I began to emphasizeProgramThe employee's work efficiency was changed.

These three Languages aim at the future multi-core and even multi-core (? Although there are many differences in details on the manycore platform, they are consistent in the following aspects:

The goal is to increase the programmer's work efficiency by 10 times on the basis of maintaining C/Fortran running efficiency;
Designed for a wide range of platforms from large hosts to desktop systems, it is not only limited to science and engineering computing, but also to the mainstream commercial computing fields;
Try to make it easier for programmers to explore different levels of parallelism;
Use advanced technologies to reduce the burden on programmers (for example, type inference, lockless parallel support ).

Chapel claim that it can be separatedAlgorithmThe specification and data layout can be used to debug programs on workstations or even PCs, and then run in a large distributed and highly concurrent environment. I don't know how to do it. Which of the following experts will study it?

Interestingly, sun's fortress was designed as a substitute for FORTRAN (incompatible), while IBM's x10 (temporary name) was an extension based on Java, added many concurrency features. It is clear that their respective business strategies are under consideration.

For sun, it may feel that its Java has been successful enough in the field of general software, but Java has never been competent for large-scale scientific computing tasks, which makes sun worried. Taking this opportunity to develop fortress, the goal is to replace Fortran's dominance in the field of scientific computing. To achieve this, Sun has brought "formula tranlation" to its fullest, claiming that it would allow scientists and engineers to program with mathematical formulas. For example, you can use the Greek letters that mathematicians like as variables in the program, and connect the two numeric variables together to represent multiplication, And you can directly write the score and index. Like this:

 

Probably DARPA thinks this thing may be good, but it is not general enough, so sun is not able to enter the next round. However, sun is still developing fortress. Sun said that no language is suitable for all occasions. The implication is that I should use these two languages.

IBM doesn't have to say that it has earned a full disk in Java, so it is closer to Java than sun does to Java. Haha, joke. In fact, IBM has invested a lot of manpower and material resources in Java. For example, there is a project that studies the real-time Garbage Collector, and the control algorithms used in it are very creative. The results have been able to control a small unmanned helicopter in real time. I personally think x10 will have a promising future.

Sorry, I didn't even mention the idea about how these languages support thousands or even thousands of cores? -- I don't know either of these languages. Interested friends can refer to this articleArticleGet started. At the same time, forget the war between Java and C/C ++. It should be a past tense.