On Tue, 2009-10-06 at 17:30 +0200, Nick Copeland wrote: > Out of interest, what was the problem with MIDI sequencing and SYSEX? > The only problem I knew about was that they are atomic and cannot be > interupted hence on 'legacy' 5-pin DIN at a meagre few kilobaud there > were timing issues, largely overcome now with USB unless you want to > download gigasamples.
For a given interface, you cannot send more midi-data down an USB line than would fit in legacy midi, so the bottleneck still persists. Otherwise you'd have created a standard that would overflow any midi-endpoint connecting to the "real" world. What you can do is to have up to 16 midi-endpoints (real or virtual 5-pin DIN) for each device in both directions. If you know that the device is virtual and that it won't pass on any messages to the next device, you can sometimes get away with sending usb-midi at a higher rate. This has to be implemented at the driver level though. FWIW, I had something like that going for a while on a BCR2000 for subtle (fast switching) glow-in-the-dark effects as well as flashing blinkenlights, using parameter feedback. _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev
