Re: [ANNOUNCE] New PC-Speaker driver

Re: [ANNOUNCE] New PC-Speaker driver

[Stas Sergeev]

Hello.

Pavel Machek wrote:
> What problems does Jaroslav have with the driver?
This one:
http://www.alsa-project.org/archive/alsa-user/msg04284.html
But this is 2 years old so it may be obsolete:)

Re: [ANNOUNCE] New PC-Speaker driver

[Thunder from the hill]

Hi,

On Mon, 26 Aug 2002, Pavel Machek wrote:
> What problems does Jaroslav have with the driver?

I remember someone mentioned a PC-speaker driver would eat up too many 
interrupts, and would also not be maintained but rot in the corner...

			Thunder
-- 
--./../...-/. -.--/---/..-/.-./..././.-../..-. .---/..-/.../- .-
--/../-./..-/-/./--..-- ../.----./.-../.-.. --./../...-/. -.--/---/..-
.- -/---/--/---/.-./.-./---/.--/.-.-.-
--./.-/-.../.-./.././.-../.-.-.-

Re: [ANNOUNCE] New PC-Speaker driver

[Pavel Machek]

Hi!

> > Maybe ALSA team have some member
> > crazy enough to join you.
> The main problem with ALSA, which can also be a final
> problem of getting that in, is that Jaroslav doesn't
> like the whole idea of a pc-speaker driver. But I have
> a hope that if the driver works well and the code is
> also good, he may reconsider:)

What problems does Jaroslav have with the driver?
									Pavel
-- 
I'm pavel@ucw.cz. "In my country we have almost anarchy and I don't care."
Panos Katsaloulis describing me w.r.t. patents at discuss@linmodems.org

Re: [ANNOUNCE] New PC-Speaker driver

[Daniel Phillips]

On Thursday 15 August 2002 16:34, Helge Hafting wrote:
> The senseless cpu usage for something as simple as sound
> is worse.

Sound is simple?  Okaaaay....

I first coded this technique up on a 20 MHz 386, running with a 20 KHz 
interrupt.  My current machine is more than 1,000 times faster.

If you can spare the 3-5% cpu anyway, who cares?  And finally, nobody is
forcing anybody to configure this driver.

-- 
Daniel

Re: [ANNOUNCE] New PC-Speaker driver

[Stas Sergeev]

Hello.

Albert D. Cahalan wrote:
>> as it's opened it starts generating some horrible clicks and a
>> high-pitched whine.
>> Do I blame my motherboard (actually, a laptop)? Is there any way to 
>> fix this, or at least improve it?
> Adding a capacitor is supposed to help a PC speaker
> sound driver.
Ah, of course, that's what Denis Vlasenko was talking
about, now I see.
Clicks are caused by disabled ints, but whine is another
problem.
Recalling my old 386 board where I tried similar driver
under DOS and it also produced high pitched tone so I
ended up adding a capacitor, which solved the problem.
Now I think that the better solution is to add an option
to use higher modulation rates.
I have added an option to use 36KHz instead of 18.5KHz.
I beleive that this is far above the frequency range of
any motherboard and can solve the problem.

Andrew Rodland, could you please test the new patch from
my page and see if enabling the high freq modulation fixes
your problem, at least whine, not clicks?

Re: [ANNOUNCE] New PC-Speaker driver

[Albert D. Cahalan]

Andrew Rodland writes:

> I can get some pretty decent sound out of it, but I also get some
> horrible noise. Even if I send the driver a stream of zeroes, as soon
> as it's opened it starts generating some horrible clicks and a
> high-pitched whine.
> 
> Do I blame my motherboard (actually, a laptop)? Is there any way to fix
> this, or at least improve it?

Adding a capacitor is supposed to help a PC speaker
sound driver.

Re: [ANNOUNCE] New PC-Speaker driver

[Denis Vlasenko]

On 15 August 2002 08:42, Daniel Phillips wrote:
> > Of course I understand the desire to make simple hardware do nice and
> > unexpected things which it even wasn't designed to do.  :-)  Maybe ALSA
> > team have some member crazy enough to join you.
>
> You simply argued that because it might not work well for everybody, then
> nobody should have it.  I hope you see the fallacy.

It won't work well for everybody, then it won't live in mainline.
Because newcomers will enable it, be pissed off with crap sound etc...
"Political" reasons I'm afraid...

I don't say that driver is useless or something similar.
I'd object to such statement myself. It is fun to play with.

Keeping it as a separate patch is completely sane thing to do.

> I have two machines here that want it, and on which it works fine.  One of
> them is a modern server.
>
> It's a small patch and decently coded.  Sure, it could use a little more
> work, but that is exactly what it will get if it's in mainline.  I'm in
> favor of seeing this in mainline.

Me too, but my initials are not LT :(
--
vda

Re: [ANNOUNCE] New PC-Speaker driver

[Helge Hafting]

Denis Vlasenko wrote:

> It won't work well for everybody, then it won't live in mainline.
Bad reason.  You can select IDE without the fixes for your
particular buggy IDE adapter and have it eat the disks.  Still,
turning off RZ1000 and CMDxxx fixes is possible for those
that know they have a good adapter.

So, no need to reject the speaker driver for "crap sound".
It'll be usable with a good speaker, and the config help
text can simply state that it is a last-resort driver
which might work badly because it pushes the hardware.

> Because newcomers will enable it, be pissed off with crap sound etc...
> "Political" reasons I'm afraid...

The senseless cpu usage for something as simple as sound
is worse.  Consider putting a old voice modem on a serial
port and connect a speaker to the phone output...

Helge Hafting

Re: [ANNOUNCE] New PC-Speaker driver

[Stas Sergeev]

Hello.

Denis Vlasenko wrote:
 > I'm afraid I'll disappoint you guys but chances of
 > getting this into mainline
Well, I not counting on that too much myself too,
but only because the code is currently not ready.

 > 1.New motherboards have built-in sound,
Not all af them (my K7M doesn't) and there are a
lot of old ones. Note that linux still supports 386.

 > it may be crappy but definitely
 >   better than PC speaker.
Of course but they require an external speakers while
the speaker is always present:)

 > Since they may
 >   be wired differently, you can't be sure which way you
 > can force maximum amplitude
 >   on a particular mobo (there are 2 or 3 ways to reach 
 > max on different mobos.
Hmm, that sounds strange to me... Could you provide
some details?

 > 3.It loads CPU enormously.
Hmm, but it shouldn't. It used to work on my 486DX4-120
without loading CPU too much (not more than 3% - 5%) and
on my Athlon700 the load is not noticeable absolutely.
Yes, it speeds up the timer ints from 100Hz to 18.5KHz,
but the handler is very lightweight, what it does it
mainly just restarts the timer. What is this CPU you
have there that gets overloaded?

 > In short: making it work right on wide variety of 
 > hardware is next to impossible
And I think quite opposite: it works already on most
hardware and with trivial workarounds like disabling
APM idle calls, it can work on even more. What it really
doesn't like is when interrupts are disabled for too long,
but, as noted by Daniel Phillips, this is a bad condition
for many other things as well.

 > and even then results are mediocre (low volume, radio
 > quality).
Yep, that's the tradeoff for getting 6-bit sound from
a single-bit device:)

 > Maybe ALSA team have some member
 > crazy enough to join you.
The main problem with ALSA, which can also be a final
problem of getting that in, is that Jaroslav doesn't
like the whole idea of a pc-speaker driver. But I have
a hope that if the driver works well and the code is
also good, he may reconsider:)

 > It won't work well for everybody, then it won't live in
 > mainline.
Hey, how many drivers are marked Absolete, Experemental
or even Dangerous? This driver can be marked Experemental
at the end.

 > Because newcomers will enable it, be pissed off with crap 
 > sound etc...
Guess what the most newcomers does with the
CONFIG_EXPEREMENTAL option? :)

 > Keeping it as a separate patch is completely sane thing 
 > to do.
... at least for the near future, yes. But probably not forever
as there appears to be some demand, even more than I 
initially expected. After all I think that user's demand is 
rather strong argument for having any particular feature 
when the contrary arguments is only that it won't work in
some rare cases (but won't break anything either).

Re: [ANNOUNCE] New PC-Speaker driver

[Denis Vlasenko]

On 14 August 2002 20:44, Andrew Rodland wrote:
> > >> The latest patch for 2.4.18 kernel
> > >> is available here:
> > >> http://www.geocities.com/stssppnn/pcsp.html
> > >
> > > Tested. Works for playing MP3s.
> >
> > Thanks for your testing, indeed my primary
> > goal was to make the sound quality acceptable
> > even for playing MP3s.
> > With the motherboard's output attached to an
> > external speakers the quality is definitely
> > acceptable, but for the internal beeper I am not
> > shure if it is possible to really enjoy MP3s however:)

I'm afraid I'll disappoint you guys but chances of getting this into mainline
are slim for the following reasons:

1.New motherboards have built-in sound, it may be crappy but definitely
  better than PC speaker.
2.PC speaker hardware is not standardized enough. It is designed to beep reliably,
  but no manufacturer tests it for good frequency diagram and such. Since they may
  be wired differently, you can't be sure which way you can force maximum amplitude
  on a particular mobo (there are 2 or 3 ways to reach max on different mobos.
  Or so I read in a magazine a long ago).
3.It loads CPU enormously. Even more so considering that some recent chipsets _emulate_
  speaker via their integrated sound and SMM mode (ick).

These are typical symptoms:
> I can get some pretty decent sound out of it, but I also get some
> horrible noise. Even if I send the driver a stream of zeroes, as soon
> as it's opened it starts generating some horrible clicks and a
> high-pitched whine.
>
> Do I blame my motherboard (actually, a laptop)? Is there any way to fix
> this, or at least improve it?

In short: making it work right on wide variety of hardware is next to impossible
and even then results are mediocre (low volume, radio quality).

Of course I understand the desire to make simple hardware do nice and unexpected
things which it even wasn't designed to do.  :-)  Maybe ALSA team have some member
crazy enough to join you.
--
vda

Re: [ANNOUNCE] New PC-Speaker driver

[Daniel Phillips]

On Thursday 15 August 2002 18:34, Denis Vlasenko wrote:
> Keeping it as a separate patch is completely sane thing to do.

No it's not, it will rot and not be maintained as it should be.

-- 
Daniel

Re: [ANNOUNCE] New PC-Speaker driver

[Daniel Phillips]

On Thursday 15 August 2002 12:57, Alan Cox wrote:
> There are easier equivalent ways to accurately measure the IRQ
> behaviour. Flip the state of a parallel port pin when you mask
> interrupts. You can even slap an oscilloscope on it that way

Right, I'll just go get my oscilloscope now...

In fact I relied on the technique you mentioned a great deal when
I was doing realtime work with PCs.  With a multi-trace digital
scope the effect is something like a poor-man's logic analyser
with a resolution of 10 us or so.

Another technique I can recommend highly is panicking by halting
in a tight loop with the speaker set to emit some tone.  It may
be annoying for the guy at the next bench but in some code, such
as transitioning to/from virtual86 mode, there aren't a lot of
alternatives.

-- 
Daniel

Re: [ANNOUNCE] New PC-Speaker driver

[Alan Cox]

There are easier equivalent ways to accurately measure the IRQ
behaviour. Flip the state of a parallel port pin when you mask
interrupts. You can even slap an oscilloscope on it that way

Re: [ANNOUNCE] New PC-Speaker driver

[Daniel Phillips]

On Thursday 15 August 2002 01:28, Stas Sergeev wrote:
> Also try disabling APM manually. If this doesn't
> help then this is another problem but anyway someone is
> disabling interrups for the large periods, (hopefully)
> only this can cause such an effect.

And if that's the case, he's broken for a lot more reasons than the
speaker sound driver.  We've just discovered a new use for this
driver: detecting interrupt-disable borkness that needs to be hunted
down and killed.

-- 
Daniel

Re: [ANNOUNCE] New PC-Speaker driver

[Daniel Phillips]

On Thursday 15 August 2002 15:18, Denis Vlasenko wrote:
> On 14 August 2002 20:44, Andrew Rodland wrote:
> > > >> The latest patch for 2.4.18 kernel
> > > >> is available here:
> > > >> http://www.geocities.com/stssppnn/pcsp.html
> > > >
> > > > Tested. Works for playing MP3s.
> > >
> > > Thanks for your testing, indeed my primary
> > > goal was to make the sound quality acceptable
> > > even for playing MP3s.
> > > With the motherboard's output attached to an
> > > external speakers the quality is definitely
> > > acceptable, but for the internal beeper I am not
> > > shure if it is possible to really enjoy MP3s however:)
> 
> I'm afraid I'll disappoint you guys but chances of getting this into mainline
> are slim for the following reasons:
> 
> 1.New motherboards have built-in sound, it may be crappy but definitely
>   better than PC speaker.

Non argument.  What about old motherboards?  What about modern motherboards that
don't have sound, and yes, they do exist.

> 2.PC speaker hardware is not standardized enough. It is designed to beep reliably,
>   but no manufacturer tests it for good frequency diagram and such. Since they may
>   be wired differently, you can't be sure which way you can force maximum amplitude
>   on a particular mobo (there are 2 or 3 ways to reach max on different mobos.
>   Or so I read in a magazine a long ago).

Non argument.  What about that hardware that it works fine on?

> 3.It loads CPU enormously. Even more so considering that some recent chipsets _emulate_
>   speaker via their integrated sound and SMM mode (ick).

Non argument.  What about if you have cycles to burn, but no sound hardware?

> These are typical symptoms:
> > I can get some pretty decent sound out of it, but I also get some
> > horrible noise. Even if I send the driver a stream of zeroes, as soon
> > as it's opened it starts generating some horrible clicks and a
> > high-pitched whine.
> >
> > Do I blame my motherboard (actually, a laptop)? Is there any way to fix
> > this, or at least improve it?
> 
> In short: making it work right on wide variety of hardware is next to impossible
> and even then results are mediocre (low volume, radio quality).

So what?  If it works on *your* hardware then you want the option.

> Of course I understand the desire to make simple hardware do nice and unexpected
> things which it even wasn't designed to do.  :-)  Maybe ALSA team have some member
> crazy enough to join you.

You simply argued that because it might not work well for everybody, then
nobody should have it.  I hope you see the fallacy.

I have two machines here that want it, and on which it works fine.  One of
them is a modern server.

It's a small patch and decently coded.  Sure, it could use a little more work,
but that is exactly what it will get if it's in mainline.  I'm in favor of
seeing this in mainline.

-- 
Daniel

Re: [ANNOUNCE] New PC-Speaker driver

[Stas Sergeev]

Hello.

Andrew Rodland wrote:
> I can get some pretty decent sound out of it, but I also get some
> horrible noise. Even if I send the driver a stream of zeroes, as soon
> as it's opened it starts generating some horrible clicks and a
> high-pitched whine.
> Do I blame my motherboard (actually, a laptop)?
Yes because for most people it works pretty
fine:)

> Is there any way to fix this, or at least improve it?
Well, it was reported that CONFIG_APM_CPU_IDLE
causes such noises on some machines and even
CONFIG_APM_ALLOW_INTS doesn't fix the problem.
I have included a workaround that disables idle calls
during a playback. Are you using the latest patch from
my page (for 2.4.19 now)?
Also try disabling APM manually. If this doesn't
help then this is another problem but anyway someone is
disabling interrups for the large periods, (hopefully)
only this can cause such an effect. Make sure that
the hard drive is not active during a playback
because it also distorts sound by disabling interrupts
for too long.

Re: [ANNOUNCE] New PC-Speaker driver

[Daniel Phillips]

On Thursday 15 August 2002 01:14, Daniel Phillips wrote:
> On Wednesday 14 August 2002 14:24, David Woodhouse wrote:
> > Michael.Knigge@set-software.de said:
> > > > Well, there is (currently) no intention to get it into the mainstream
> > > > kernel so don't treat it too seriously.
> > 
> > > Oh, I would love to see that thing in the Standard-Kernel....
> > 
> > Wait for people to stop using the 8254 timer for timer ticks, and you can 
> > have it all to yourself -- the timer abuse is the main reason the driver
> > was never suitable for inclusion.
> 
> Ah, if I recall correctly this technique uses a different timer channel
> from the timer tick.  What's the abuse?
>
> > Actually, now that HZ is easier to vary, you can switch it to a power of 2 
> > and use the RTC for it, again leaving you the 8254 for your own nefarious 
> > purposes.

Never mind, I realize it's the speeded up timer that's at issue.  Still, I
don't see what the problem is, just step it down to a suitable timer tick
speed, and it doesn't have to be an even division either.

-- 
Daniel

Re: [ANNOUNCE] New PC-Speaker driver

[Daniel Phillips]

On Wednesday 14 August 2002 14:24, David Woodhouse wrote:
> Michael.Knigge@set-software.de said:
> > > Well, there is (currently) no intention to get it into the mainstream
> > > kernel so don't treat it too seriously.
> 
> > Oh, I would love to see that thing in the Standard-Kernel....
> 
> Wait for people to stop using the 8254 timer for timer ticks, and you can 
> have it all to yourself -- the timer abuse is the main reason the driver
> was never suitable for inclusion.

Ah, if I recall correctly this technique uses a different timer channel
from the timer tick.  What's the abuse?

> Actually, now that HZ is easier to vary, you can switch it to a power of 2 
> and use the RTC for it, again leaving you the 8254 for your own nefarious 
> purposes.

-- 
Daniel

Re: [ANNOUNCE] New PC-Speaker driver

[Andrew Rodland]

On Wed, 14 Aug 2002 20:58:20 +0400
Stas Sergeev <stssppnn@yahoo.com> wrote:

> Hello.
> 
> Denis Vlasenko wrote:
> >> The latest patch for 2.4.18 kernel
> >> is available here:
> >> http://www.geocities.com/stssppnn/pcsp.html
> > Tested. Works for playing MP3s.
> Thanks for your testing, indeed my primary
> goal was to make the sound quality acceptable
> even for playing MP3s.
> With the motherboard's output attached to an
> external speakers the quality is definitely
> acceptable, but for the internal beeper I am not
> shure if it is possible to really enjoy MP3s however:)

I can get some pretty decent sound out of it, but I also get some
horrible noise. Even if I send the driver a stream of zeroes, as soon
as it's opened it starts generating some horrible clicks and a
high-pitched whine.

Do I blame my motherboard (actually, a laptop)? Is there any way to fix
this, or at least improve it?

Re: [ANNOUNCE] New PC-Speaker driver

[Stas Sergeev]

Hello.

Denis Vlasenko wrote:
>> The latest patch for 2.4.18 kernel
>> is available here:
>> http://www.geocities.com/stssppnn/pcsp.html
> Tested. Works for playing MP3s.
Thanks for your testing, indeed my primary
goal was to make the sound quality acceptable
even for playing MP3s.
With the motherboard's output attached to an
external speakers the quality is definitely
acceptable, but for the internal beeper I am not
shure if it is possible to really enjoy MP3s however:)

Re: [ANNOUNCE] New PC-Speaker driver

[Stas Sergeev]

Hello.

Michael Knigge wrote:
 > Oh, I would love to see that thing in the Standard-Kernel....
Yes, I'd also like to see it there, but
for now I don't know if/when this can be
achieved.
As pointed by David Woodhouse (from whom
I've stolen the code just because I felt
it was not updated for tooo long:) the main
problem is that this driver, dispite no longer
touching irq.c (the latest patch doesn't touch
it) still speeds up the timer (this is
unavoidable). Well, as in 2.5 HZ is configurable,
this is probably a way to go (but I haven't
yet played with 2.5).
However, since 2.5 uses ALSA rather than OSS, I
think I have to make an ALSA port before thinking
about an integration. But even if this is done,
somehow this must be dealt with:
http://www.alsa-project.org/archive/alsa-user/msg04284.html

So there are anough of long-term issues and for
2.4 solving them is not possible.
My goal for 2.4 is to get rid of the native fops
and use OSS fops instead (USE_OSS_FOPS in pcsp.h
must be set to 1) but even this doesn't work very
well: when CPU load is high, my output routines
are called with the sound buffer entirely zeroed
out! I don't know who wipes out the buffer on a
high CPU load and as the problem is deeply buried
in the OSS internals (not in my driver) I don't
know how to deal with that.

 > Thanks for your work! This is something I was missing for
 > years!
Thanks:) I'll try to make this driver acceptable
for inclusion, but this will take *a lot* of time.

Re: [ANNOUNCE] New PC-Speaker driver

[Denis Vlasenko]

On 29 July 2002 07:29, Stas Sergeev wrote:
> For all those people who still
> don't have a sound card I want to
> introduce a pc-speaker driver.
> There were some other pc-speaker
> drivers floating over the net, but
> AFAIK no one is really finished and
> usable.
> My driver is originally based on
> Michael Beck and David Woodhouse
> driver, but it is havily reworked
> and pretends to be 100% OSS compatible
> producing nearly the best sound
> one can ever get from pc-speaker.
> Well, there is (currently) no
> intention to get it into the mainstream
> kernel so don't treat it too seriously.
> However any comments or bugreports are
> appreciated.
>
> The latest patch for 2.4.18 kernel
> is available here:
> http://www.geocities.com/stssppnn/pcsp.html

Tested. Works for playing MP3s.
--
vda

Re: [ANNOUNCE] New PC-Speaker driver

[David Woodhouse]

Michael.Knigge@set-software.de said:
> > Well, there is (currently) no intention to get it into the mainstream
> > kernel so don't treat it too seriously.

> Oh, I would love to see that thing in the Standard-Kernel....

Wait for people to stop using the 8254 timer for timer ticks, and you can 
have it all to yourself -- the timer abuse is the main reason the driver
was never suitable for inclusion.

Actually, now that HZ is easier to vary, you can switch it to a power of 2 
and use the RTC for it, again leaving you the 8254 for your own nefarious 
purposes.

--
dwmw2

Re: [ANNOUNCE] New PC-Speaker driver

[Michael Knigge]

Hi,

> Well, there is (currently) no
> intention to get it into the mainstream
> kernel so don't treat it too seriously.
> However any comments or bugreports are

Oh, I would love to see that thing in the Standard-Kernel.... My 
Server at home uses Dial-On-Demand for my Internet connection and the 
beautiful Gillian Anderson is talking to me if the connection gets up 
("wow!") and down ("uuuhhh, you scare me") ;-))))

**THIS** is the only reason why I have a soundcard in my PC - just to 
get some accoustic information for a few events (another sample could 
be a finished backup, incoming mail from fetchmail, incoming faxes or 
incoming ISDN-Calls from friends...). 

For this your driver is more than enough! A "real" soundcard isn't 
necessary anymore (and PCI-Slots are always rare)..... This evening I 
will remove my old soundcard from my server!


Thanks for your work! This is something I was missing for years!


Bye
  Michael

[ANNOUNCE] New PC-Speaker driver

[Stas Sergeev]

Hello.

For all those people who still
don't have a sound card I want to
introduce a pc-speaker driver.
There were some other pc-speaker
drivers floating over the net, but
AFAIK no one is really finished and
usable.
My driver is originally based on
Michael Beck and David Woodhouse
driver, but it is havily reworked
and pretends to be 100% OSS compatible
producing nearly the best sound
one can ever get from pc-speaker.
Well, there is (currently) no
intention to get it into the mainstream
kernel so don't treat it too seriously.
However any comments or bugreports are
appreciated.

The latest patch for 2.4.18 kernel
is available here:
http://www.geocities.com/stssppnn/pcsp.html

Re: [ANNOUNCE] New PC-Speaker driver

[Pavel Machek]

Hi!

> > In short: making it work right on wide variety of hardware is next to impossible
> > and even then results are mediocre (low volume, radio quality).
> 
> So what?  If it works on *your* hardware then you want the option.

It will work well enough to be used for speech synthesis on most hw. It only
eats CPU when in use. [Integrating festival into kernel for *speaking* panics?]


									Pavel
-- 
Philips Velo 1: 1"x4"x8", 300gram, 60, 12MB, 40bogomips, linux, mutt,
details at http://atrey.karlin.mff.cuni.cz/~pavel/velo/index.html.