In message <[EMAIL PROTECTED]>, "M. Warner Losh" writes: >In message: <[EMAIL PROTECTED]> > Poul-Henning Kamp <[EMAIL PROTECTED]> writes: >: But the i8254 is a piece of shit in this context, and due to >: circumstances (apm being enabled0 most machines end up using the >: i8254 by default. >: >: My (and I belive Bruce's) diagnosis so far is that most problems >: come from the i8254 timecounter. > >We measured pps interrupts with the i8254 timecounter in a fast >interrupt handler via the parallel port (yes, we hacked it to give us >a fast interrupt). We found lots of outliers on the order of a few >milliseconds in the data that we had to discard because they were >obviously bogus. We don't know if this is because of interrupt >latency or because of bugs in the 8254 timecounter code/hardware. At >the time, it wasn't important enough to do a detailed numerology on to >see if more data couldn't be mined from it or not. And the data that >we saw the outliers in was somewhat processed from the original >data...
I have not tried to measure the i8254 against my hardware solution, but by now I belive that certain bogus chipsets may have bummed the 'latch' command or more than that maybe. Anyway, some, but not all of the i8254 issues could be eased up a bit if we lost the pcaudio crap and used the RTC's 128 Hz signal for Hz and let the i8254 run at a 65536 count all the time. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
