Searching a realtime OS booting as fast as possible

Searching a realtime OS booting as fast as possible

Post by Markus Ber » Thu, 11 Dec 2003 15:11:31


Hey NG,

I am searching for a hard realtime OS that boots as fast as possible,
let say in about 500ms. After this boot sequence I would like to
perform stuff like copying data from one memory location to another.
I found RTLinux but I don like that license and I don believe that
it is already ported to another processor than the PPC405GP.
Is there anybody out there who can give me a hint?

Thanks in advance

Markus
 
 
 

Searching a realtime OS booting as fast as possible

Post by Peter Pete » Thu, 11 Dec 2003 16:26:15

Hi Markus,


You should let us know for what kind of target you need this OS.

For 32-bit x86, On Time RTOS-32 ( http://www.yqcomputer.com/ ) used on BIOSSless
targets can boot in less than 500ms. The FAQ page
http://www.yqcomputer.com/
requirements.

Peter

 
 
 

Searching a realtime OS booting as fast as possible

Post by Markus Ber » Thu, 11 Dec 2003 17:28:50

Hi Peter,

sorry I forgot to mention that I am not bound to a specific target.
Hence there are no constraints to bear in mind.

Markus
 
 
 

Searching a realtime OS booting as fast as possible

Post by Pat For » Thu, 11 Dec 2003 21:42:36


Qnx 4.XX could be set up for this, try an email to qnx ( jwall is a good
contact)
 
 
 

Searching a realtime OS booting as fast as possible

Post by Paul Keina » Fri, 12 Dec 2003 00:12:33

On Wed, 10 Dec 2003 07:11:31 +0100, Markus Bernauer



Is it really necessary to bootstrap the system as fast as possible or
do you only need to be ready to operate as soon as possible after the
power is applied ?

In the old core memory days, the power fail interrupt simply stored
the volatile internal CPU and peripheral registers into the
non-volatile stack in the core. At power on, these registers were
loaded back core and the system was ready to run in a few machine
instructions after power up as if nothing had happened.

Modern implementations could use battery powered RAM or various
hibernation tricks such as storing the contents of the RAM and CPU
registers into a flash card. Of course the power supply must be able
to maintain the voltages for some time between the power fail
interrupt and actual disappearing of the power in order to be able to
save the memory contents.

Using some hibernation tricks to suspend the processing is usually
much faster than making a full bootstrap sequence. So in principle, a
system needs to be bootstrapped only once, but after that, power
failures and even component replacement can be done without a full
reboot if the system has been designed that way.

Paul
 
 
 

Searching a realtime OS booting as fast as possible

Post by prenom_nom » Mon, 29 Dec 2003 22:44:27


I think most dedicated RTOSes should meet this requirement on modern
CPUs assuming you are not instantiating bootloads of C++ objects upon
startup.