[SqNOS]
Re: ioRelinquishProcessor, timer, power consumption and IRQs
Gerardo Richarte
gera at corest.com
Thu Jul 20 15:08:13 UTC 2006
Craig said:
> Is there some way you can wake up only if the IRQ is indeed associated
>with a delay, network packet, mouse-move, etc.? The rate of such
>interrupts would be far less than 2kHz, so you could conceivably use
>even less power.
>
>
Not really, unless I reproduce a big deal of IRQ handling code in
ioRelinquishProcess, as I'll need to distinguish which IRQ awaked the
hlt. It may be as simple as just checking if it was a timer IRQ, and if
it's not really wake up. Ok, after all, it may not be too complicated.
Now, in the case it's a timer IRQ I'll need to see if the time for
the expiration of next Delay has come or not. Is there an easy way to do
it? (I mean, easy like "already done").
> As it stands, I'd be happy with 2kHz or 1kHz, but probably no less than
>that. I see no drawbacks to this, and it's nice and simple.
>
>
yeah, I like the simplicity too, like ioRelinwuishProcess does
"hlt", what else were you expecting? :-)
however, I'm experiencing a weird effect that I associate to this
change (but could be something else). When I double click on a word,
lets say in workspace, the highlighting of the word takes a lot of time
(more than a second), however, I subjectively feel improved every other
spect of responsiveness... any idea?
gera
More information about the SqueakNOS
mailing list