A new system, PaSH, has been developed to make computer programs run faster while also delivering accuracy in the Unix shell. The Unix shell is a 50-year-old programming environment that is still widely used today.
The new system boosts the speeds of programs by splitting the components into pieces and allow them to run simultaneously on multiple computer processors. Doing so allows programs to carry out tasks a lot more quickly.
With the new system, it will enable the high number of people who use it. Biologists, data scientists, engineers, and economists can speed up their programs without fear of getting the wrong results.
Aside from faster programs, the system will make it easier for programmers to develop tools that biologists, engineers, and others use. There is no more need to make any special adjustments to the program commands to enable automatic, error-free parallelization.
Also Read: Sphero Mini-Car Robot Teaches Kids About the Basics of Computer Programming - Isn't It Too Cute?
Solving an Old Problem
PaSh is focused on program or scripts in the Unix shell. One of the main problems researchers encounter for years is the correct and automatic parallelization of shell scripts. Even if the Unix shel allows for multilanguage scripts, they are still challenging to parallelize via the traditional methods.
This is because parallelizing a program is often tricky because some parts of it rely on others. Therefore, when one gets the order wrong, the whole program fails.
PaSh solves these issues through a preprocessing step that inserts simple annotations onto program components that is viable to be parallelized. After that, the system will attempt to parallelize the components of the script while the program is running.
Therefore, this dodges another problem in shell programming and effectively speeds up a lot of the components compared to traditional methods that try to conduct parallelization in advance.
It also accelerates program and ensures it returns accurate results. When the system arrives at a program component that can't be parallelized, it will run the original version to avoid causing an error.
The system has been tested on hundreds of scripts and it didn't break a single one. It was able to run programs six times faster compared to unparallelized scripts. Moreover, it boosted script speeds, making it the first to show this type of correct transformation in its fullest form.
In the future, the computer scientists want to use the system to address distribution problems in which it will divide a program to run on several computers instead of having many processors in one computer. They are also looking to improve the annotation scheme to make it more user-friendly and be able to better describe complex program components.
Related Article: This Technology Will Allow You To Control Robots With BrainWaves And Hand Gestures
This article is owned by TechTimes
Written by April Fowell
