Parallel Programming with OpenMP
OpenMP Hands On
To use OpenMP in your own programs, you need a computer with more than one CPU, or a multi-core CPU and an OpenMP-capable compiler. GNU compilers later than version 4.2 support OpenMP. Also, the Sun compiler for Linux is free [2], and the Intel Compiler is free for non-commercial use [3].
Listing 5 shows an OpenMP version of the classic Hello World program. To enable OpenMP, set -fopenmp when launching GCC. Listing 8 shows the commands for building the program along with the output.
Listing 5
Hello, World
Listing 8
Building Hello World
$ gcc -Wall -fopenmp helloworld.c $ export OMP_NUM_THREADS=4 [...] $ ./a.out Hello World from thread 3 Hello World from thread 0 Hello World from thread 1 Hello World from thread 2 There are 4 threads
If you are using the Sun compiler, the compiler option is -xopenmp. With the Intel compiler, the option is -openmp. The Intel compiler even notifies the programmer if something has been parallelized (Listing 9).
Listing 9
Notification
Benefits?
For an example of a performance boost with OpenMP, I'll look at a test that calculates pi [4] with the use of Gregory Leibniz's formula (Listing 7 and Figure 5). This method is by no means the most efficient for calculating pi; however, the goal here is not to be efficient but to get the CPUs to work hard.
Listing 7
Calculating Pi
Parallelizing the for() loop with OpenMP does optimize performance (Listing 6). The program runs twice as fast with two CPUs than with one, in that more or less the whole calculation can be parallelized.
Listing 6
Parallel Pi
$ gcc -Wall -fopenmp -o pi-openmp pi-openmp.c $ export OMP_NUM_THREADS=1 ; time ./pi-openmp Pi = 3.141593 real 0m31.435s user 0m31.430s sys 0m0.004s $ export OMP_NUM_THREADS=2 ; time ./pi-openmp Pi = 3.141593 real 0m15.792s user 0m31.414s sys 0m0.012s
If you monitor the program with the top tool, you will see that the two CPUs really are working hard and that the pi-openmp program really does use 200 percent CPU power.
This effect will not be quite as pronounced for some problems, in which case, you might need to resort to serial execution for a large proportion of the program. Of course, your two CPUs will not be a big help in such a case, and the performance boost will be less significant. Amdahl's Law [5] (see the "Amdahl's Law" box for an explanation) applies here.
Amdahl's Law
"Speedup" describes the factor by which a program can be accelerated with parallelization. In an ideal case, program execution with N processors would take just 1/N of the time required by a serial program. This ideal case is known as linear speedup. In the real world, linear speedup often is impossible to achieve because some parts of a program do not particularly lend themselves to parallelization.
Given a part of a program that supports parallelization, P (thus, 1 – P is the non-parallelizable part), and the number of processors available, N, the maximum speedup is calculated by the formula in Figure 6.
If the serial part of the program (1-P) is 1/4, the speedup cannot be greater than 4 – no matter how many processors you use.
Glossar
SMP
Symmetric multi-processor system. All of the machine's CPUs can access the shared main memory – in contrast to cluster systems, in which separate machines exchange data over the wire. OpenMP is suitable for parallel programming on SMP systems.
Thread
One popular definition of thread is a "lightweight process." A Unix process has a separate memory area and various resources are assigned to it – such as environmental variables, network connections, or device access. A thread shares memory and certain other resources with other threads in a process. This reduces the management overhead compared with processes, and facilitates switching between threads. Pressing Shift+H in the top tool enables and disables the thread display.
Infos
- OpenMP homepage: http://www.openmp.org
- Sun compiler: http://developers.sun.com/sunstudio/
- Intel compiler: http://www.intel.com/cd/software/products/asmo-na/eng/compilers/clin/
- Calculating pi (Wikipedia): http://en.wikipedia.org/wiki/Computing_Pi
- Amdahl's law (Wikipedia): http://en.wikipedia.org/wiki/Amdahl's_law
« Previous 1 2 3 4
Buy this article as PDF
(incl. VAT)
Buy Linux Magazine
News
-
Another New Linux Laptop has Arrived
Slimbook has released a monster of a Linux gaming laptop.
-
Mozilla VPN Now Available for Linux
The promised subscription-based VPN service from Mozilla is now available for the Linux platform.
-
Wayland and New App Menu Coming to KDE
The 2021 roadmap for the KDE desktop environment includes some exciting features and improvements.
-
Deepin 20.1 has Arrived
Debian-based Deepin 20.1 has been released with some interesting new features.
-
CloudLinux Commits Over 1 Million Dollars to CentOS Replacement
An open source, drop-in replacement for CentOS is on its way.
-
Linux Mint 20.1 Beta has Been Released
The first beta of Linux Mint, Ulyssa, is now available for downloading.
-
Manjaro Linux 20.2 has Been Unleashed
The latest iteration of Manjaro Linux has been released with a few interesting new features.
-
Patreon Project Looks to Bring Linux to Apple Silicon
Developer Hector Martin has created a patreon page to fund his work on developing a port of Linux for Apple Silicon Macs.
-
A New Chrome OS-Like Ubuntu Remix is Now Available
Ubuntu Web looks to be your Chrome OS alternative.
-
System76 Refreshes the Galago Pro Laptop
Linux hardware maker has revamped one of their most popular laptops.