[RFD] Alternative PC speaker drivers?

[RFD] Alternative PC speaker drivers?

[Robert B. Hamilton]

Just wondering: is anyone interested in exploring alternative methods
for the pc speaker driver?

I've one idea, with some rudimentary code, thats been gathering dust for 
a while. It uses the real time clock for timing, along with udelay.  
It also has a (very) rudimentary soundcore interface. The only advantage 
it would have as an alternative to the existing driver is that the only 
kernel code it touches is in rtc.c, and then only to export a couple of 
functions.  If I understand correctly the objections to the current driver, 
this means it *might* have the potential to have an outside chance of getting 
slipped into the cannonical kernel distribution.

Disadvantage: it sucks up cpu cycles like a black hole. One would need
at least a 266 MHz processor, and even then not recommended on a heavily 
loaded machine.  Sound quality is not quite as good as existing driver, 
although I've gotten functional speech using rsynth, and can also tweak 
Maelstrom, eg, to get reasonable sound.

Any comments, discussion, helpful advice, would be much appreciated.
                                                 --Robert

Re: [RFD] Alternative PC speaker drivers?

[Jeff Tranter]

The PC speaker driver is a neat hack, but with the cost of low-end
sound cards ($25-$30 in these parts for a SB16), I wonder if anyone
would really use it.

On Wed, 21 Apr 1999, Robert B. Hamilton wrote:
> Just wondering: is anyone interested in exploring alternative methods
> for the pc speaker driver?

--
Jeff Tranter (mailto:tranter@pobox.com http://www.pobox.com/~tranter)
Currently running Linux 2.2.6      Learn more at http://www.linux.org

Re: [RFD] Alternative PC speaker drivers?

[Robert B. Hamilton]

> The PC speaker driver is a neat hack, but with the cost of low-end
> sound cards ($25-$30 in these parts for a SB16), I wonder if anyone
> would really use it.

Certainly that's true.  Still, people might find a use for it as
a "baseline" sound device (however crude) that is nearly guaranteed 
to be on every ix86 machine; and there is the occasional inquiry on l-k.
So perhaps there is no good reason not to support it, *provided* the 
support can be done in a technically acceptable manner.

Which brings us back to my original note.  Conventional wisdom has
had it that the maximum 8192Hz rate of the RTC is too low to support
the speaker driver.  I don't think that is true any more.  With faster
cpus one can use udelay's to get a finer time resolution (although
there are drawbacks).  If this is possible then it is a much cleaner 
solution than the current method which varies the system clock frequency.

But I've only tested this on my computer, a K6, so I don't know how well
it can work on others....I'd be willing to make the preliminary code 
available, and to work on midi, mixer and sequencer emulation, etc, if 
there were sufficient interest.....

----------------------------------------------------
Robert Hamilton
r-hamilton@physics.tamu.edu