Among the decades, very many programming languages has been evolved. We list some most widely used in the context of HPC here.
Low-Level programming languages
- Machine code: That are 10001001011110 your computer can understand. You likely not.
- Assembler: human-readable (to somehow extent; many says 'for special humans'), but still very low-level. Allows the programmer to get the very last possible performance crumbs from the hardware, but in turn is hardware-dependent and very tedious to program. Highly-tuned libraries often contain ASM parts, often making this software less-portable. You should not start at assembler level when programming an application.
High-level programming languages
... are readable by humans. The source code is a text file, to be translated to an executable file by a compiler (we do not want to introduce the definition bush about compiled and interpreted languages here). In addition to human-readablity these programs are portable, you just need to recompile your code on another hardware using a compiler understanding the same language. Note that being standard-compliant for both the code and the compiler greatly enhance the chances of porting to be a smooth, enjoyable procedure.
- Fortran, the 1st wide-used high-level programming, used in HPC since 1954 and will likely be used for a long time, as there are still many Fortran projects and tons of libraries-canot-give-up around. Fortran handles multi-dimensional arrays very comfortably. Due to some restrictions in the memory handling the compiler has more optimisation freedom leading to quite well-performing code out of the box, usually.
- C and C++ are related languages with the second being object-oriented, with a lot of programming comfort. “C programs are the fastest” is usually not true. As C has very few syntax limitations, this often gives the compiler a hard time optimizing. The passing of pointers is in most cases the biggest handicap. You can rid this, but need to take care.
- JAVA is a widely used programming language, but due to performance issues in most cases not suited for true high performance computing.
Script / interpreted languages
- Shell helps you to script many calls to a single program. The batch scripts are programmed in one of Shell languages.
- Python evolves to very powerful scientific computing language nowadays, especially for those prepared to abadon MATLAB from the compute pipeline.
- Perl is another scripting language with very powerful handling of regular expressions.
- MATLAB is a powerful numerical computing environment with same-called programming language.
Books an Links
- Brian W. Kernighan, Dennis M. Ritchie: Programmieren in C - Carl Hanser Verlag - 1990. ISBN-13: 978-3446154971 (in German)
- Brian W. Kernighan, Dennis M. Ritchie: The C Programming Language - Prentice Hall; 2nd edition (April 1, 1988). ISBN-13: 978-0131103627
- Henry S. Warren: Hacker's Delight - Addison-Wesley Professional; 2nd revised edition (25. September 2012). ISBN-13: 978-0321842688 Webpage: hackersdelight.org/ (Archived)
- Ulrich Breymann: Der C++ Programmierer - Hanser Verlag - 2017. ISBN-13: 978-3446448841 (in German)
- online C++ reference
- William H. Press: Numerical Recipes 3rd Edition: The Art of Scientific Computing - Cambridge University Press - 2007. ISBN-13: 978-0521880688 Webpage: numerical.recipes/ (The original nr.com two-letter domain gone by the wind.)
- Stephen Chapman: Fortran for Scientists and Engineers - McGraw-Hill Education. This book was the Fortran book for many years; the actual 4ed - 2017 seem to be overprised.
- Michael Metcalf, John Reid, Malcolm Cohen: Modern Fortran Explained: Incorporating Fortran 2018 - ISBN-13: 978-0198811893. Latest edition of Fortran books by those autors.
- Walter S. Brainerd: Guide to Fortran 2008 Programming - 2015 - ISBN-13: 978-1447167587. Also available in older editions for older Fortran standards.
- Fortran Standards Documents