Economics of Commercial Kernels and Other System Software
You have two choices. Hire a programmer to create an in-house kernel that would end up costing you about $80,000 to $100,000 (approximately 10 man months) or buy a commercial kernel for about $6,000 to $9,000. Which makes choice more sense? Which choice do you think your boss would appreciate. The bottom line is that a quality kernel with quality support is a good business decision.
Preemptive Virtues
So you have decided to use Multitasking. But, which approach will you choose? You have three basic methods of multitasking, round robin, time-slice, and preemptive model. I have read articles discussing the first 2 choices, but not the preemptive model, hence the reason for my article below..
RTOS task switching - an example implementation
FreeRTOS is an open source mini real time scheduler (or RTOS kernel). Following a brief introduction to some multitasking concepts, this article describes how FreeRTOS operates. The AVR microcontroller port along with the GCC tools are used to present an example implementation that includes source code.
Testing and Measuring the TAMS 3011
The eCos embedded operating system offers an alternative to UNIX-style operating systems for development work. <a href=http://www.ibm.com/developerworks/power/library/pa-tams2/?ca=dgr-lnxw961TestTAMSP2>This article</a> examines how its architecture influences the development process, building a sample application and exploring the differences in architecture between eCos and UNIX.
Unbound Stacks and Stoppable Tasks
Unbound stacks and stoppable tasks have been an integral part of smx for quite a while and many of our applications have benefited from them. They are good tools to have. They make a contribution to the evolving theory of optimum embedded system kernels.
Why Use a Commercial Kernel?
The President of a small company recently asked me why he
should buy a commercial kernel instead of having one done
in-house. This started me thinking. How do I answer when
someone says "Our application is to simple for a kernel", or
"It will only take 1-2 weeks", or "The learning curve is to
long"? The following paper is my answer.
Files
