Resources for “Real-Time and Performance Improvements in the 2.6 Linux Kernel”

April 25th, 2005 by William von Hagen in

Resources for the print article.

“Realfeel Test of the Preemptible Kernel Patch” by Andrew Webber: www.linuxjournal.com/article/6405

Real-Time Patches and Projects

2.4 Pre-emptible Linux Kernel Patch: kpreempt.sourceforge.net

2.6 Real-Time and Preemption Patch from Ingo Molnar: redhat.com/~mingo/realtime-preempt

MontaVista Real-Time Project: source.mvista.com/linux_2_6_RT.html

RTAI Project: www.rtai.org

DIAMPM RTAI (Real-Time Application Interface): www.aero.polimi.it/~rtai

Benchmarks

realfeel Benchmark: brain.mcmaster.ca/~hahn/realfeel.c

__________________________

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Time interval

On June 2nd, 2005 Anonymous (not verified) says:

Figure 3 is reported as a 1 min. run time.

The end of the test period, around 3500, shows a increase in jitter. Has a longer than 1 min. run been tested? What do the results above 1 min. show?

Am I seeing a bug?

On May 17th, 2005 John Major (not verified) says:

Great article. Clears up that nagging "I know this is good but how does it help?" I had about preemtable kernel. When I look at the jitter results in figure three I have two related questions. Why is the distribution so balanced in the vertical direction (rather than skewed mostly to the positive), and why are almost all the positive data points balanced almost exactly by a negative point. Could this be a bug in Realfeel where a 'delayed' interrupt incorrectly causes a successive accurate interrupt to be reported as 'early' by the same amount? You can also see a few points in figures one and two where they exceed the general noise.