The average (and most companies would suggest ideal) computer user has very little knowledge of what goes on just in turning on their computers. Most needs center around opening of a browser and surfing the web. Computer science and software engineers, as well as the computer “geek” understand that a lot goes on in under the hood. Bill Gates money was the product of wanting to do more than turn on a computer. Steve Jobs maximized Unix to create a “perfect” surfing experience. As it happens, if you are a daring soul, you can wipe the drive of your computer, install Ubuntu (Linux , yes it very close to Unix) derived, and get ready to have headaches because even though they are on version Saucy Salamander, even us highly evolved don’t want to find a compiler nor write eighty percent of the code that that will let us use our ipods. This paper paper involve the use of programming languages and ease of use. The article by Loh, titled “The Ideal HPC Programming Language” is an investigation into how HPC computing benchmarks needed to be examined and then improved upon. That is the simplest explanation.
Imagine that you are a programmer and wished to create software based on the organization you saw fit. This was the question that Andrew Loh asked himself. Sun Microsystems, the company that sponsored this project, has all the ups and downs of most other dot.com, tech industries, meaning that their products were not just physical items that could be touched but also software solutions that could be utilized by other programmers. They suffered halting programs that were some of the first to attempt with parallelism, which is essentially complete software and hardware integration (I would argue that Apple now comes the closest to this conceptually) Their business model is part of the world’s new paradigm. In other words, like Facebook, you can use it but you can’t place Facebook in a shopping bag. Sun Microsystems currently deals in servers and workstations that are often used by companies and usually not by everyday consumers. So when they asked the question that was asked at the beginning of the paragraph the researchers wanted to know what users would do if there were no limitations such as computer systems, compilers, or strict language configurations. These “sample” programs, without constants and knowing (or caring whether or not they were actually functional) opened the eyes of not just Sun Microsystems but also other companies. Whiteout any constrains, programmers designed what they needed to be productive, and overwhelmingly, these sample looked very much like FORTRAN. It is important to remember that these subjects were already program language savvy, but it speaks volumes that many would write something that resembles FORTRAN, a programming language that was developed in 1950’s. They realized that they had hit upon something very important and potentially profitable. Sun rewrote many HPC benchmarks were given feedback which consisted of a reduction in the lines of code needed and reduction in redundancy, a straight forward code. Of the course biggest critique is obvious, which is the inability of compilers that created fully operational programs. Therefore, if this wasn’t fixed, it would render code changes meaningless. Sun Microsystems, needed solutions. The article contributes that the need to get rid of large amounts of source code was bigger that parallelism and hardware performance. The computational gurus needed to be concise. The computations needed to be made elegant. E=mc2 is an excellent, and succinct way to describe energy. Mass is easily attained, and c, the speed of light, has a static constant. This type of simplicity will take time as well as cooperation. As the article states from”programming in the small” to programs that are collaborative, not competitive.
Use your promo and get a custom paper on
"The Ideal HPC Programming Language".
As programmable hopefully becomes simpler, FORTRAN will do more than do-loops, which are used for doing simple counting but have to be daisy chained to be effective in complicated calculations. They are effective but not for HPC usage. As the article states “Loops might be nested, or loop iterations might be unbalanced.” (Loh, 2010). They will be unbalanced.
FORTRAN is array rich. Arrays follow certain programmable patterns and FORTRAN has stepped up their complexity. Without the increase in arrays, they will be unable to create a simpler language that instead of using 1-d arrays, can use 3-d matrices and therefore simplify algorithms. FORTRAN is the benchmark for HRC computing ( Hu,2000).
- Hu, Y. C., Jin, G., Johnsson, S. L., Kehagias, D., & Shalaby, N. (2000). Hpfbench: A High Performance Fortran Benchmark Suite. ACM Transactions on Mathematical Software, 26(1), 99. Retrieved from http://go.galegroup.com/
- Millennial Fortran. (1999, November). Mechanical Engineering-CIME, 121(11), 28. Retrieved from http://go.galegroup.com/
