Re: [linux-audio-dev] an open letter to Linus re: low latency

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: Re: [linux-audio-dev] an open letter to Linus re: low latency
From: Jay Ts (jay_AT_toltec.metran.cx)
Date: Fri Jun 23 2000 - 18:31:38 EEST


Paul wrote:
>
> Jay wrote:
> >The daemon would handle any number of inputs, outputs, synths, MIDI
> >devices, etc., on whatever sound hardware is available (no design limit).
> > [etc.]
>
> you clearly need to know more about ALSA.

Yes, I definitely should go study that immediately after I get this
Perl programming project out of the way.

> it doesn't have a sound
> daemon, but it can already do most of what you describe above, and/or
> has hooks for the remaining stuff.

Oh, goodie. Maybe it really is time to do some more audio programming! :)

> latency is not a function of bandwidth.

Yes, I know. At the bottom end, we can have extremely high bandwidth,
and still be limited in latency due to the finite speed of light along
optical fibers. Even if we could transmit terabits per second, it would
still take a little while for the first packet to get from one end to
the other. For a connection that goes half the way around the planet
(say, Sydney to NY) that's at least a 11,000 mile distance. With light
going "only" 186,000 miles per second, it's a significant delay of
about 60 msec, theoretical minimum. (I hope I got both my theory and
arithmetic are right on that. :-)

> network interface latency has
> been going down too, but not as fast as bandwidth has been going up.

One thing that bugs me a little is that the requirement for
VoIP is a maximum delay of about 120 msec from one end to the other.
That's acceptable for a voice conversation, but terrible for digital
audio. Considering how long we've been limited to a 3 KHz bandwidth
on the telephone (unchanged by digital telephony and fiber optics)
I wonder if the kinds of things I want to have possible over the
Internet will ever happen.

But even if not on the Internet, perhaps just on the LAN.

> >One of my goals is to be able to dedicate one computer as a multitrack
> >recorder, and use another one or two for the synthesizer(s), another to
> >do pitch to MIDI conversion to MIDIfy my electric guitar, maybe another to
> >convert pulses from drum pads to MIDI, etc., and maybe a bunch of rack mount
> >Linux systems to replace the rack mount digital effects boxes in the studio,
> >and tie it all together with an ethernet LAN for the digital data flow.
> >(S/PDIF? Digital ADAT interfaces? Ancient, archaic stuff. Who needs them? :)
>
> you've just spent way more than you would have done with dedicated
> boxes. using one computer for each of these tasks is absurd.

I was exaggerating a bit for purposes of illustration, but actually there
are good reasons to dedicate a computer to at least some of the above
tasks. Notice that I said "to be able to"; not that I will necessarily
need or desire to actually do it in every case.

> if all
> you are doing is a single task (eg. fx), you're much better off with
> no disk, no monitor, no keyboard, no mouse, no floppy, no IDE or SCSI
> bus, no video bus, no PCI bus, no ISA bus, etc. etc. etc. Thats why
> you can get a kick-ass TC multi-fx unit for US$400, and only a pretty poor
> computer for anything close to that price.

Last I checked, a t.c. electronic Fireworx costs $1800 (street), not
$400, and comes with a tiny monochrome LCD display. I'd much rather
program a device such as that with my 19" Viewsonic monitor, or at
least a 15" model. Also, I recently purchased a used Pentium for
$200 including monitor, and it is much more versatile than a dedicated
digital effects box or synthesizer. It can be a groove box or synth one
minute and a multitrack recorder (with digital effects) another minute. But I
found if I try to do more than one or two at a time, there are serious
limitations. The 200 MHz CPU is a limitation, especially during moments
when more than one process needs a lot of real-time CPU power at the
same time. Going to 600+ MHz would help a little, but a more conservative
solution is to dedicate a box to each function, just as a large ISP might
dedicate a single box to being an email server, and another as a web server.

What I've noticed recently while playing around with some Windows software
is that even this little 200 MHz Pentium has more processing power (and
RAM and disk space) than some synths on the market, and I'm much better
off running FruityLoops on my cheap Pentium than going out and buying a
Roland MC-505 (currently about $1100, street). And we're getting
close to where a computer can replace something like a t.c. electronic
Fireworx or Finalizer or Eventide Harmonizer (currently about $3500-4000,
street). Perhaps this is why we are now seeing so many effects and
synthesizers showing up in the form of VST plugins. You can now buy
a Waldorf keyboard (with physical keyboard, knobs and buttons), or get
another model as standalone software, or as a VST instrument. And the cool
thing about that is that there are different reasons to prefer any
incarnation over any of the others.

- Jay Ts
jay_AT_jayts.cx


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Fri Jun 23 2000 - 19:05:10 EEST