Re: MiNT scheduling

[Mario Becroft <mb@tos.pl.net>]

On Wed, 12 Nov 1997, Konrad Kokoszkiewicz wrote:

> Actually QED's key response is too slow even without gcc running, just the
> usuall MiNTOS stuff makes the computer slow. There's no such problem in 
> normal MultiTOS setup (i.e. an AES without MiNTOS). I have no idea how to
> solve it.

I don't know, on my TT I don't notice any slow-down while running MiNTOS.
Normally QED reacts instantly, I never noticed it to be slow at all. It's
only when other processor-intensive processes are running that the
response slows down.

I have found, however, that the problem becomes not so bad when I add
"SLICES=1" to my mint.cnf.

While this seems a bit kludgy, perhaps it would help this problem if the
AES or the app that has focus could somehow be given a drastically higher
priority just when mouse clicks or keypresses occur.

But to process the key- or mouse-click should usually require a minimal
amount of processor time. For it to take a long time to respond, seems to
suggest that the background processes must be given a really huge amount
of the processortime. Surely if the foreground app could even be allowed
10% of such time whenever a key- or mouse-click occurs it should be able
to respond entirely quickly enough. Does this make sense?

Or is the answer to have smaller timeslices?

MiNT is supposed to have "adaptive prioritisation." What does this
actually mean? Just how does MiNT really distribute processortime?

Either way, there must be some solution, since this problem is not
nearly so bad on some other operating systems.

-- 
Mario Becroft                        Auckland, New Zealand
mb@tos.pl.net                 http://www.pl.net/user/mario      |\__/,|   (`\
Tariland, Atari Support in New Zealand                        _.|o o  |_   ) )
tariland@tos.pl.net  http://www.pl.net/user/mario/tariland --(((---(((--------