Re: [sheflug] Latency of context switch

[Alex Hudson <home [at] xxxxxxxxxxxxxx>]

On Thu, 2006-04-13 at 07:15 -0700, Chew Kim Hoe Jason wrote:
> Thanks... I understand that, I am trying to compare embedded linux
> with other RTOS. Therefore I am trying to find some benchmarking on
> this. I am being a little general as not to pose too many restriction
> in order to get some feedback :o) I am actually using DIGNSYS
> SDK675001-EK1 with Oki ML675001 chip on it, running on uClinux kernel
> 2.4.22.

Hmm. I doubt that you'll find any published figures which are
particularly relevant to your case. It's going to vary massively from
chip to chip and version to version.

One of the major changes in more recent Linux is the use of different
schedulers: these can affect the context switch latency massively: 

	http://cs.nmu.edu/~randy/Research/Papers/Scheduler/

If you're able, it might be worth doing the tests yourself with
something like LMbench:

	http://www.bitmover.com/lmbench/

Cheers,

Alex.



___________________________________________________________________

Sheffield Linux User's Group -
http://www.sheflug.co.uk/mailfaq.html

  GNU the choice of a complete generation.