Re: granularity
Posted by
ccs@m...
on 2002-01-25 20:12:58 UTC
Art,
That is a very clever hack, and I want to thank your for inventing and
describitng it. But I don't personally like the idea of having the
computer thrash around in calibrated delay loops to generate
more-or-less even timing. Sure, one can buy a 600 mhz or 1.2 ghz
processor... I like my $100 hamfest Pentium-200's quite fine though.
And I'd rather hang a hardware timer on them than buy a faster
processor with which to imitate a hardware timer.
But to each his own! There are certainly situations where a more
expensive platform that can be bought at the local mall ends up
cheaper than an inexpensive one requireing a special part or
a some-assembly-required operating system.
Chris Stratton
Christopher C. Stratton
Engineer, Instrument Maker, and Horn Player
ccs@... 617 628 1062
http://web.mit.edu/~stratton/www/brassbuild.html
That is a very clever hack, and I want to thank your for inventing and
describitng it. But I don't personally like the idea of having the
computer thrash around in calibrated delay loops to generate
more-or-less even timing. Sure, one can buy a 600 mhz or 1.2 ghz
processor... I like my $100 hamfest Pentium-200's quite fine though.
And I'd rather hang a hardware timer on them than buy a faster
processor with which to imitate a hardware timer.
But to each his own! There are certainly situations where a more
expensive platform that can be bought at the local mall ends up
cheaper than an inexpensive one requireing a special part or
a some-assembly-required operating system.
Chris Stratton
> It is a common fallacy that a PC timer running at 8Khz has the next step-------------------------------------------------
> of resolution at 4Khz. This would seem logical on a hardware timer, but most
> people fail to consider the power of an algorithm controlled by a timer. In
> Master5 the next step after 8192HZ downward is 8191Hz. It was this very
> argument that started me on the path to insanity writing the software in the
> first place. Some of the tech's among us were convinced that a 8192Hz
> interrupt rate would have to go to 4096 as the next frequency. Think about a
> timer going 8192 Hz. It Pulses on every interrupt (every 122uS). When it
> wants to pulse at 8000Hz ( 125uS) it really only needs to pulse immediately
> on interrupt #1, then on Interrupt #2, wait 3uS (122 + 3), then pulse, on
> interrupt number 3 wait 6uS ( (122-3)+6) more then pulse again, etc.... The
> wait times will stretch until the next even boundry time and then repeat.
> This increases the burden of the processor, but who cares, there's lots of
> cycles and lots of free time in each of the 8192Hz cycles to release back to
> the processor.
Christopher C. Stratton
Engineer, Instrument Maker, and Horn Player
ccs@... 617 628 1062
http://web.mit.edu/~stratton/www/brassbuild.html