[sdiy] (errata) The goose-stepping morons are at it again

[Richard Wentk]

At 08:54 02/03/2003 -0800, Scott Gravenhorst wrote:

>I remember a thread about this USB-MIDI thing before and I believe someone
>stated that USB isn't really the best way to go in the first place, something
>about overhead and latency?

USB has no guaranteed performance. If you start choking up your USB line 
with all kinds of other stuff, there's no guarantee your bytes will get 
through when you send them. However... given that the various commercial 
interfaces work well, I'd say it's more of a problem in theory than practice.

In fact the *only* way to guarantee performance, esp. with a multiport 
interface, is to send out time-stamped events ahead of time to a 
subprocessor, and let it sort out what to send and when.

>I would think that EPP would be better.  Personally, I prefer a bus card for
>MIDI.  I have a USB dual port MIDI interface, but it's mainly for my laptop
>and for a secondary set of ports should I need more than the two that my bus
>card has.
>
>Anyone care to comment on what the differences between the 3 methods (bus
>card, EPP and USB) might be?  What is the best?

The big problem with both bus cards and EPP is that specs are liable to 
change without notice. EPP is *already on its way out*. There are moves to 
eliminate parallel and serial ports from PCs over the next year or three.

I'd expect PCI to be around for at least that long, but not as long as USB 
will. Although now that USB 2 is here, it's possible even that's up for grabs.

FireWire looks more interesting. Does anyone have any experience?

Wireless looks more interesting still, but I suppose that's a whole other 
bag of bytes. By my calculations you should be able to easily get 8 mono 
24/96 channels of audio through an 802.11g interface. MIDI would be - 
rather less demanding.

Richard