The Logic Off Topic list

--- In logic-ot@y..., Dennis Gunn <dennisg@a...> wrote:
> 
> But they might be willing to pay for two machines that are designed 
> to work together harmoniously which is something that it is not 
> always safe to assume is going to be the case between even the 
> components of a single PC.

Oh, sure, agreed.

But if a Mac person intends to buy a G4, and fill it up with various 
pieces of legacy PCI kit from their PC (eg high-performance audio and 
video cards), then they might start getting similar problems. 

Flashy graphics cards seem to be particularly bad offenders, I had an 
Ati Radeon PCI card in my desktop PC and Ati's installation software 
was quite awful, and the software update procedures were a mess. I no 
longer use that card unless I have to. But that's Ati's fault, and 
for all I know, a mac version of the software might be just as bad.
My laptop PC has only crashed maybe four times in six months of heavy 
use (brief, minor, auto-recovered crashes) and each time the OS has 
blamed the built-in graphics hardware ... which turns out to be 
another Ati Radeon (mobility version).

At this point, I start thinking that perhaps this Ati product line 
suceeds on the basis of performance rather than on the amount of time 
that Ati have spent making it and its support software stable, and 
you might argue that this is symptomatic of the way that the PC 
market operates. 

But ... guess what graphics hardware Apple decided to build into the 
G4 range? 
<grin>  
Yep, it's Ati again!

 

>   Given the troubles I've seen with my PC I would be happy to pay 
the 
> premium so as not to have waste my precious time and creativity 
> trying to figure out what was causing some nasty glitch between two 
> CPUs of different types in some Rub Goldberg rig.  

Yep, but I woudn't set your hopes /too/ high on being able to achieve 
stability nirvana by upgrading to a mac, if you are then going to 
load up the poor machine with all your nasty third-party PCI 
hardware, and drivers equivalent to those that were previously 
screwing up your PC.

Some PC music people do pay premium prices for custom-designed music 
PC systems, partly so that they know that the people at the company 
have debugged the configuration that they are buying and so they 
won't have to tear their hair out worrying about driver updates and 
soundcard conflicts. I've got a couple of the nicer soundcards, but 
I've ended up taking them out of my main sequencing machine and 
relying on softsynths instead, because those are less hassle and more 
portable (if my main machine suffers a major coffee spill, I can run 
the same configuration on my backup machine, or even at a pinch on my 
laptop, without touching a screwdriver)

Talking to people who use BigMacs packed with all the proper 
soundtools stuff in a professional studio situation, the bad news is: 
apparently they crash just like PC's do. 

     

> I would not mind 
> paying for a couple of rack mount mac DP servers.  It would still 
> probably come out to less Bang for buck wise than buying Digidesign 
> cards.

Yeah, I saw those servers on the apple site and had a small "want 
one!" moment before I saw the price.

Then again, if you strip down a computer this far, it probably works 
pretty well irrespective of whether its a PC or a mac. 
If its running Win2000 or XP, and its just a glorified processing box 
with fast memory and a network connection and no funny CDwriter 
drives, or soundcards or graphics cards, or PCI bus ... then you've 
removed most of the scope for hardware conflicts, because you've 
removed most of the hardware. 




==If== Yamaha's mLAN driver is rock-solid (currently an unknown 
factor), and DAW software companies then embrace the idea of moving 
processing load over the network on demand, then I think the DAW 
market may move away from big expensive centralised single machines, 
towards networks, and successful network standards tend to be cross-
platform.



The usual problem with attempting to spread tasks across a network of 
processors is is the difficulty of splitting the job into self 
contained modules that can function independently on different 
processors, without the processing needed to split the job up and 
recombine the various elements afterwards (and handle the network) 
being so processor-intensive that you lose the advantage of the extra 
CPUs. 

But a music studio application like Logic is perfect for it, you 
already have multiple parallel chains of software modules that work 
with explicitly-defined signal streams, and some of the highest-
demand parts of the job, the effects plugins and software 
instruments, are not only already designed as independent modular 
code, they also already have published interfaces that let them run 
inside other apps on the PC or Mac, that can control them and load 
and save all their settings on demand!

Trying to turn a wordprocessor into a multi-processor application is 
probably a bit depressing (do you run a spell-checker module on a 
separate processor? big deal!), but for a music DAW, you already have 
the parallel data streams and processing blocks all shown right there 
on the screen and being configured by the user, and with mLAN we'll 
have an agreed standard for making those same (MIDI, audio) 
connections between codeblocks work across the network.
That just leaves the job of getting the central machine to ask 
satellite machines to load and configure links on demand, and I think 
the simplest way to do that is to leave it to the DAW companies to 
produce satellite apps that link with their main DAW's over the 
network, so you click on an audio instrument slot or mixer channel in 
Logic's environment, select the "run remotely" option, and the 
object's display and controls then become a controller for a copy of 
that object running on "Logic Satellite" on the second machine, which 
actually does all the audio processing and sends the results back 
over mLAN.


> Apple actually is working on some kind of cpu ganging scheme.  It 
> will be interesting to see if they come up with something that will 
> do what we need.

It might take Apple and Intel and Motorola years to come up with a 
proper distributed-processing architecture, legacy languages and 
datastructures weren't designed for proper fault-tolerant plug-and-
play modular processing. 

But luckily for those of us in this particular market, we don't need 
that, we have a much simpler solution coming into view: 

  1: Apple release OSX incorporating mLAN support (due this year?)   
  
  2: Yamaha release the mLAN drivers for Windows (due any time now?) 

  3: Companies who write DAW software issue network licenses and 
write a quick set of protocols into their software so that if machine 
A is running a load of softinstruments and is getting a bit choked, 
it asks the user if they want to "autofarm"  and if they click yes, 
machine B (running either the same software or a cut-down "shell" 
version of the same DAW) loads the same softinstrument with the same 
parameters, the instrument is bypassed on A, and the slot's MIDI 
stream is piped to machine B over mLAN, and the resulting audio comes 
back into the slot over mLAN. The user doesn't need to intervene, all 
their editing and saving is done on machine A, and the only 
difference is that the effects stack or instrument or mixer channel 
gets displayed on A with some subtle signifier that it's a "ghost 
instrument" or "ghost channel" that's actually being processed 
somewhere else on the network (the user doesn't care where).

I'm guessing that the most intimidating "nuts and bolts" part of 
getting logic to be able to work like this would be the job of 
rewriting the audio mixer channel code (or writing a dummy 
instrument) so that a slot can be converted into audio and MIDI send 
and return signals that are then linked to a pair of free audio 
ports ... 
... and guess what, that new feature is supposed to be appearing 
under Logic 5.3 next month, on the final cross-platform release.



 
> >Cross-platform processing is liable to end up being the domain of
> >companies like Steinberg and Native Instruments, who are 
maintaining
> >a presence (and investment) on both platforms.
> 
> Liable?  E-mapples position is pretty clear at this point don't you 
think.

Well, before July 1st, emagic's public position on cross-platform 
support was pretty clear, too! Apple might still have the option of 
changing their mind about dropping Windows support (although each day 
that passes makes this more difficult). 

mLAN is already on some hardware, and the emagic guys have probably 
already been playing with it. It should show up under Windows as 
conventional MIDI and Audio ports, which logic can already access, so 
I guess its conceivable that emagic might have put in a bunch of mLAN-
friendly features into 5.3 as a final finishing touch to logic-as-we-
know-it, and to wind up the features that they had planned for v5.x 
before they move on and start on the next mac-only incarnations.

I don't know whether its /likely/ or not, but its conceivable!  
Yamaha aren't saying when their mLAN windows drivers are expected, so 
I suppose we are waiting for an announcement of some sort. 
/Hopefully/ an imminent announcement.

And hopfully Apple haven't struck a deal with Yamaha to get them to 
delay or drop their planned mLAN Windows support, that would REALLY 
piss me off.

 
> >Unless apple/emagic change their minds about dropping all Windows
> >audio product development, I don't see how they can hope to be a
> >credible player in this new sort of market.
> 
> By making something that works well.  Only time will tell.

Well, Betamax worked well, but Sony refused to allow their superior 
heads to be used in other manufacturers machines, and the publc 
perception of betamax as an "unpopular" format became self-
fulfilling. It lost critical mass, and VHS (which lots of 
manufacturers could "support" by buying in the core mechanism and 
putting their own box and badge around it) won out. 

Music is a "niche" computing application, and macs are a "niche" 
computer product, by becoming mac-only, emagic is now selling to a 
niche within a niche. That might not be a large enough target 
audience to sustain decent sales, especially since Steinberg are 
going to be able to leverage their new Windows dominance to move 
further into the mac market. Its not just the customers they have to 
convince, its also the music shops who now have to be persuaded to 
continue stocking a product that is now mac-only, and not to try to 
switch sales to Steinberg. 

If "network-awareness" becomes a key DAW feature, and logic's 
implementation only works with macs, and Steinberg's works cross-
platform, then the Steinberg solution will tend to be the default 
winner.

I guess a lot of this depends on when Yamaha's Windows mLAN driver 
appears, on what features 5.3 has, and on how well 5.3 utilises the 
mLAN remote configuration features (if it does).
If 5.3 implements some sort of remote control and autofarm features, 
but doesn't do them particularly well, I suppose they could get away 
with a further minor release to tidy things up a few months later,  
call it 5.3.1, and by calling it a maintenance release for features 
added on 5.3.0, they wouldn't be going against the position stated in 
the July 1st press release.


Maybe all of this will take shape in the next ~four weeks, or maybe 
none of it will.

=Erk=