On Wed, 10 Mar 1999, Paul Clyne wrote:
> What is the speed tradeoff for each method ?.
Speed isn't the issue here. Well... only a little. Because when
you have no modules there is no time lost to load/unload some module. And
it's safer.
This is about memory usage. A module like the sound module is
pretty large. And if you don't use sound all the time than it will be a
puls to make it module. For example when I listen to a CD I have to
insert the sound module to change the volume settings. Than I unload it.
And when running with only 8M of ram it can make a difference.
> I have a P100 with 32M memory and 50M or so swap {which hardly seems to
> get used} but the systems _feels_ slow.
According to the latest standards you system is slow. Compared
with my computer your system is very fast.
> Is a smaller Kernel (ie using modules) with the auto modle loader (what's
> it called again) running in the background going to operate faster than the
Actually if you use kerneld the changes are smaller. kerneld uses
itself resources and time. kerneld is very useful for a beginner. After
some time, I feel like I have more power without kerneld. Things like I
can let a user mount floppies. But what if I don't load the fat and the
vfat module? I made scripts that insert/remove the needed modules. So I
gain some resources this way. And I preffer to know what's going on.
> same items made into a monlithic one ?. Or is the currently 'inactive'
A monolithic one is faster and safer. It's faster because you
don't load or unload modules and because you have one precess less
(kerneld). But it eats more memory to put all in one piece.
> parts of a monolithic kernel going to suck more CPU cycles than the module
> loader would use.
The unused parts are exactly like that: unused. They just take up
the space. There is no CPU involved, nor swap...
Raider
--
``Liberate tu-temet ex inferis''
