I read:
Do MIDI In jacks still use opto-isolator components? I think they did in the original spec... The rise- and fall-times could account for the added delay perhaps?
I get about 40-60 with the alsa virmidi driver (-nosound, or with audio, makes no difference), if I use -audiobuf 5 I get ~20-30 but that is about as low as it gets.
so basically this is too long, and the jitter is too high
But where does that come from ? 1/44.1*64 = 1.45ms for each block (i.e. between ctrl rate evaluations) means ~35 ctrl rate 'cycles' between sending and receiving midi ? since win2k suffers aswell it might not be specific to oss or alsa's oss emulation ...
regards,
x