Re: [LAD] audio/midi app development

On Thursday 17 June 2010, Ralf Mardorf wrote:
>Gene Heskett wrote:
>> On Wednesday 16 June 2010, Patrick Shirkey wrote:
>>> On 06/17/2010 04:52 AM, Ralf Mardorf wrote:
>>>> Paul Davis wrote:
>>>>> On Wed, Jun 16, 2010 at 2:30 PM, Ralf Mardorf
>>>>>
>>>>> <ralf.mardorf@alice-dsl.net> wrote:
>>>>>> PS: Why not programming for savant syndrome musical gifted and
>>>>>> 'fast' watching people too?
>>>>>
>>>>> the limits under discussion relate to monitor technology, not human
>>>>> capabilities.
>>>>
>>>> I'm not a 'fast watching savant' ;) and even if the GUI is too slow, I
>>>> won't care. I'm listening to music with my very good ears, but my bad
>>>> eyes. No doubt, Linux is a good choice, but MIDI real-time could be
>>>> better. For me the GUI is unimportant. BUT I prefer to do audio
>>>> recordings using Linux, but MIDI recordings. It's a real pity, because
>>>> MIDI would add some very cool features.
>>>
>>> This is only on your system right? I know a lot of people are working
>>> with midi recording using linux tools.
>>>
>>> You see jitter at low latency but have you tried changing your hardware
>>> or working with the driver developers to isolate and fix the bugs you
>>> are seeing?
>>
>> One of the test tools that might be enlightening for the MIDI folks
>> here, is the machine control program called emc. Because jitter is very
>> important when feeding a stepper motor controller a steady heartbeat at
>> high audio and somewhat above frequencies, the coders have developed a
>> 'latency-test' script, which you run on one screen, then abuse the heck
>> out of the machine doing other things, (browsing the web, moving windows
>> around, compiling a kernel, whatever warms up the cpu) then come back
>> half an hour or more later and read the average and worst case latencies
>> as displayed in nanoseconds.
>>
>> Those are generally big figures so do the math and make milliseconds out
>> of them.
>>
>> Emc when running stepper motors is fussier that all get out, and that
>> tool just might point the finger at truly bad motherboard, or video
>> hardware. FWIW, an nvida video card, can only be used in a machine
>> running emc if the vesa driver is selected, all the others including nv,
>> tie up the interrupts, sometimes for many milliseconds. For emc, that
>> would equal a stalled motor and a wrecked part you were cutting at the
>> time it stalled. Similar things can be said about the APCI of some
>> motherboards. If that can't be fixed via a bios setting, toss the
>> board. Via chipsets seem to be the most popular in this latter
>> category.
>>
>> If its a complex part that you've already got several hours worth of
>> carving & cutting tool wear into, that will only happen once, because
>> whatever the culprit is, gets both found and a free airmail trip into
>> the bin.
>>
>> What? Oh, I'll go back to lurking now. ;-)
>
>I guess you are regarding to http://linuxwiki.de/EMC, but I just started
>searching the web. Btw. when 'we' some old dino computer freaks
>controlled stepper motors by DOS machines, we just controlled remoted
>pics (oldish micro controllers - but not very old -, I guess you would
>use DSPs or other micro controllers today,

There are several boards about that will let you treat a stepper equipt
machine like they were servo's. But in either case, if the stepper misses a
step, game over. So one stays within the limits of what the stepper can do.
I would explain why steppers and high speeds are mutually exclusive, but
then I'd really be off topic.

>but would you use your MacOS,
>Windows, Linux instead of external chips for fast and exact real-time
>control?). While oldish machines without multitasking, e.g. the Atari ST
>could control machines on real-time, at least for applications like
>MIDI, I don't know if the Atari would be able to control a CNC machine,

I have one friend down in TX that is running his stepper mill with a TRS-80
Color Computer whose clock is .79mhz. He has been making steam engines with
it. Slowly, very slowly, using a Dremel for a spindle. For that, Dremels
area POS.

Midi's timing is fairly critical, but steppers are even fussier at higher
speeds where the torque falls off.

>I guess there are reasons for using external micro controllers when
>controlling such machines, not only when using a MacOS, Windows PC,
>Linux PC I never heard of MacOS, Windows PC, Linux PC that are used to
>do it directly, without help of external micro controllers. But I hear -
>never tested it myself - that Nuendo (for Windows) should be the first
>and only MIDI capable PC software.
>
>If jitter for MIDI is depending to chip sets - and I do belief that the
>chip sets are important - why aren't there trustful black- and
>whitelists for that hardware?

A good question. I believe the wiki may have a list of hardware broken down
that way, but its nowhere near complete or definitive. My machine is driven
by an early athlon single core, running at 1.6Ghz. That is why the latency-
test was written, because we needed to be able to define if it was usable
before we carved up an expensive hunk of metal & wrecked the job.

One should also bear in mind that MIDI, was standardized when a 600 baud
circuit was _fast_, 30 years ago. And IMO, MIDI as a standard needs to be
replaced by something with a 16 bit word, and 10 megabit or more data rates.
We can do it, optically or electrically today and I see no real reason not
to be developing a new standard that reflects what the hardware today can
do. Some folks I hear are using ethernet in studio settings, but the cables
aren't durable enough for the way touring musicians tend to treat them. So
that isn't something the guy who has a gig once a week in a bar wants to put
up with.

Anyway, I don't see any real progress being made toward a new std way of
doing midi, one that reflects what todays hardware can do. Till that day,
we are stuck with playing a midi piano that sounds a wee bit like Floyd
Cramer with an overdose of speed. With midi, you simply cannot bring all 10
fingers of both hands down on the keyboard truly simultaneously like a good
human player can. With midi working perfectly, those 10 notes are going to
have at least a .003 second lag from first to last.

>I should shut my mouth and lurk too.

Me too.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If 7-11 is open 24 hours a day, 365 days a year, why are there locks on the 
doors? 
		-- Why Why Why n�6
_______________________________________________
Linux-audio-dev mailing list
Linux-audio-dev@lists.linuxaudio.org
http://lists.linuxaudio.org/listinfo/linux-audio-dev