20-Мар-2005 19:35 [EMAIL PROTECTED] (Eric Auer) wrote to

EA> - POWER has an "check for key"-flood detector. If a program looks for
EA>   keyboard events more than N times per second, then some throttle
EA>   mechanism kicks in which sleeps M of each 8 timer ticks. Values are
EA>   documented for IBM PC DOS POWER. My problem with that: It would make
EA>   FDAPM more *invasive* to timing if I add support for the function.

     Yes, this may be invasive, but correct implementation should be useful.
As I already report, many program do dumb pooling without any extra "release
time slice" or HLT supplement. As result, when I run two (or more) Volcov
Commander under W98, CPU usage raises to 100%, notwithstanding which
idle-forcing program I run.

EA>   Maybe it would be an idea to write a tool which controls ACPI throttle
EA>   based on keyboard-check flood ratio in some way? I mean, instead of
EA>   just halting the whole system for M of each 8 timer ticks (a timer
EA>   tick is 0.054 seconds, relatively long, you WILL notice the stop-and-
EA>   go behaviour), reduce the CPU clock to M times 1/8th of maximum in a
EA>   smooth way.

     But far not all CPUs/motherboards, especially old or non-mobile support

EA>   be? Something like "if keyboard is checked more than 1000 times per
EA>   timer tick, then the system is probably bored and can be slowed down,

     There shouldn't be used constant factor, or you will be in trouble,
similar to RTE200 of TP. BTW, don't forget, that future CPUs will be

EA> I hope you can help me with some feedback on this, even if it is just
EA> "nobody needs broadcasts,

     I have neither experience, nor opinion about this, so I quiet.

SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
Freedos-user mailing list

Reply via email to